Consumer Behaviour

Applications in Marketing
Consumer Behaviour
Applications in Marketing

3rd edition

Robert East
Jaywant Singh
Malcolm Wright
& Marc Vanhuele
SAGE Publications Ltd

1 Oliver’s Yard

55 City Road

London EC1Y 1SP

SAGE Publications Inc.

2455 Teller Road

Thousand Oaks, California 91320

SAGE Publications India Pvt Ltd

B 1/I 1 Mohan Cooperative Industrial Area

Mathura Road

New Delhi 110 044

SAGE Publications Asia-Pacific Pte Ltd

3 Church Street

#10-04 Samsung Hub

Singapore 049483
 © Robert East, Jaywant Singh, Malcolm Wright and Marc Vanhuele 2017
First published 2008.
Reprinted 2008, 2009, 2011 (twice), 2012.
Second edition published 2013.
Reprinted 2014 (twice).
This edition published 2017.

Apart from any fair dealing for the purposes of research or private study, or
criticism or review, as permitted under the Copyright, Designs and Patents
Act, 1988, this publication may be reproduced, stored or transmitted in any
form, or by any means, only with the prior permission in writing of the
publishers, or in the case of reprographic reproduction, in accordance with
the terms of licences issued by the Copyright Licensing Agency.Enquiries
concerning reproduction outside those terms should be sent to the publishers.
Library of Congress Control Number: 2016938426

British Library Cataloguing in Publication data

A catalogue record for this book is available from the British Library

ISBN 978-1-47391-949-5

ISBN 978-1-47391-950-1 (pbk)

Editor: Matthew Waters

Editorial assistant: Lyndsay Aitken

Production editor: Sarah Cooke

Marketing manager: Alison Borg

Cover design: Francis Kenney

Typeset by: C&M Digitals (P) Ltd, Chennai, India

Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY
Brief Contents
About the Authors
Preface
Acknowledgements
Praise for Previous Editions
PART 1 INTRODUCTION
1 Ideas and Explanations in Consumer Research
PART 2 CONSUMption PATTERNS
2 Customer Loyalty
3 Brand Knowledge, Brand Equity and Brand Extension
4 Stationary Markets
5 Market Dynamics
6 Consumer Group Differences
PART 3 EXPLAINING DECISION-MAKING
7 Predicting and Explaining Behaviour
8 Information Processing and Decision-Making
9 Consumer Satisfaction and Quality
PART 4 MARKET RESPONSE
10 Consumer Response to Price and Sales Promotions
11 The Retail Context
12 Word-of-Mouth Influence
13 The Response to Advertising
References
Index
Extended Contents
About the Authors
Preface
Acknowledgements
Praise for Previous Editions
PART 1 INTRODUCTION
1 Ideas and Explanations in Consumer Research
Section 1: The scope of consumer behaviour
Section 2: Consumer decision models
Section 3: Classifications and explanations
PART 2 CONSUMption PATTERNS
2 Customer Loyalty
Section 1: Brand loyalty in repertoire categories
Section 2: The rise of relationship marketing – customer
loyalty as retention
Section 3: Combination definitions of loyalty
Section 4: Reasons for defection
3 Brand Knowledge, Brand Equity and Brand Extension
Section 1: The mental representation of brands
Section 2: Brand equity, brand extension and brand alliances
Section 3: Sales losses by the parent of a line extension
4 Stationary Markets
Section 1: Modelling mature markets
Section 2: Single brand purchase patterns
Section 3: Patterns of purchase in the whole category
5 Market Dynamics
Section 1: Changes in aggregate sales
Section 2: Dynamic effects and brand loyalty
Section 3: The dynamics of new product adoption
Section 4: The sales dynamics of frequently bought
categories
6 Consumer Group Differences
Section 1: Relevant differences for consumer research
Section 2: National cultural differences
Section 3: Cultural differences in consumer research
 Section 4: Age and gender differences in consumption
PART 3 EXPLAINING DECISION-MAKING
7 Predicting and Explaining Behaviour
Section 1: Definitions and measurements
Section 2: The theory of planned behaviour (TPB)
Section 3: Problems with the theory of planned behaviour
8 Information Processing and Decision-Making
Section 1: Schemas and attention
Section 2: Heuristics
Section 3: Processing value and probability
Section 4: Financial applications of prospect theory
9 Consumer Satisfaction and Quality
Section 1: Introduction
Section 2: Theories of consumer satisfaction
Section 3: Measuring satisfaction and service quality
Section 4: Outcomes of satisfaction and dissatisfaction
PART 4 MARKET RESPONSE
10 Consumer Response to Price and Sales Promotions
Section 1: Consumer response to price
Section 2: Estimating price sensitivity
Section 3: Psychological reactions to prices and price
changes
Section 4: Consumer response to sales promotions
11 The Retail Context
Section 1: Shopper choice
Section 2: Customer typologies
Section 3: The store environment
12 Word-of-Mouth Influence
Section 1: The nature of word of mouth
Section 2: The occurrence of word of mouth
Section 3: The impact of word of mouth
Section 4: WOM in the social network
Section 5: Applications of word-of-mouth research
13 The Response to Advertising
Section 1: Effective advertising
Section 2: Advertising frequency and concentration
Section 3: A model of advertising effect
 Section 4: Specific effects
Section 5: The evolving media landscape
Section 6: Biometrics and consumer neuroscience
References
Index
About the authors

Robert East
is Emeritus Professor at Kingston Business School, Kingston
University, London, and Adjunct Professor at the Ehrenberg-Bass
Institute of the University of South Australia. He trained as a social
psychologist and is a postgraduate of London Business School. His
research has mainly focused on word-of-mouth patterns, where his
new evidence has shown that some widely-held beliefs are mistaken.
As a teacher of consumer behaviour, he has been keen to deliver
knowledge that is useful to students while not oversimplifying the
subject. This book reflects his iconoclastic research and his
commitment to a curriculum that is both intellectual and useful.
Jaywant Singh
is Professor of Marketing at Kingston Business School, Kingston
University, London, where he teaches consumer behaviour and
research methods. He applies quantitative research methods to unravel
how and why people buy goods and services, through the lenses of
brand management, ethical consumption, status consumption and
service management. He has published in leading academic journals
and regularly presents his research at international conferences. In his
teaching, he endeavours to deliver theoretically and empirically
grounded research knowledge that has direct applications in marketing
and business.
Malcolm Wright
is Professor of Marketing at Massey University, New Zealand, and
Adjunct Professor at the Ehrenberg-Bass Institute of the University of
South Australia. He applies empirical principles to marketing problems
and has made interrelated discoveries about brand loyalty, the use of
probability scales, new product forecasting and optimizing the
advertising budget. He has also published many articles critically
examining the foundations of popular marketing knowledge.
Marc Vanhuele
holds a PhD from UCLA and is Professor of Marketing at HEC Paris.
As an expert in consumer information processing, his research focuses
on how consumers treat price information. A second field of research
is how marketing managers can improve their decision-making
through better use of quantitative information on the consumer’s
mindset. His research has been published in leading academic journals.
He also works as a consultant to consumer goods and market research
companies.
Preface
Readership and Scope
We have designed this book to support courses in consumer behaviour at
Master’s level. It is also suited to more advanced teaching at first-degree
level. Our intended audience is those who see consumer behaviour as a
research-based discipline that addresses the problems raised by marketing
and consumer policy. The problems we explore are found in all advanced
and emerging economies, and for this reason we believe that the book will
be useful throughout the world.

This new edition updates the subject matter of the 2013 edition, reflecting
changes in the field in recent years, but its structure remains unchanged.
The book is selective in the research it covers, dealing in some detail with
the areas chosen. As before, the chapters are quite short and are intended to
support students who will also be reading original research papers. In
updating and revising the book, we found that we could often simplify and
clarify the text. The result is a book that is easier to read and no longer than
the previous edition.

Consumer Behaviour: Applications in Marketing stresses well-researched

aspects of consumer behaviour that are of widespread importance.
Following the Introduction, we describe the patterns of customer
purchasing that are usually observed in market economies and the way
those patterns can be explained and used in practical marketing
applications. We then look at research that has illuminated our
understanding of consumer decision-making and show how this
understanding can be used by marketers and public policy-makers. The last
section of the book deals with the observed consumer response to market
intervention and covers research findings on price, promotion, word of
mouth and advertising.
Approach
Most textbooks on consumer behaviour are extensive and well illustrated,
but may present the subject in a rather uncritical manner. Often, the
treatment illustrates fashionable topics rather than providing evidence that
helps us understand long-standing marketing problems. Such books do not
make sufficient call on the expanding research in our field, and when they
do cite research may give limited attention to the uncertainties or opposing
views that persist in our discipline. In practice, there are competing findings
and explanations in all areas of consumer behaviour and marketing, and we
have tried to recognize these and discuss their relative merits.

This touches on a problem familiar to those who teach business students.

Some of these students find arguments from evidence quite unfamiliar and
may instead provide accounts of current business practice as though these
were conclusive. Our approach opposes such uncritical thinking. We
believe that those who learn to use evidence as students acquire a technique
that will serve them well as practitioners.

One hazard of research-based texts is the sheer weight of evidence. We

have tried to emphasize the most recent work and key papers on topics
while also acknowledging those early researchers who first identified
problems in consumer behaviour – problems that are usually still current.
We therefore make no apology for some of the more ancient citations in this
book, as these help to describe the origin of current thinking.

As subjects become more fragmented, textbooks acquire importance as

integrators of different perspectives. In scientific consumer behaviour, we
can discern two rather different approaches to research and application. On
the one hand, there is the tradition that dominates in the large conferences
of the Association for Consumer Research. Put baldly, this endorses
theorizing and hypothesis-testing, often within experimental designs, and
tends to emphasize explanations in terms of the beliefs, the preferences and
the culture of consumers – a cognitive orientation. In contrast to this is the
approach of those who belong to the Marketing Science grouping, who
place an emphasis on behaviour, measures rather than concepts,
generalization from an accumulation of findings, and on the use of
mathematical models rather than psychological theories for explanation.
Textbooks have generally emphasized the cognitive tradition. We give more
space than usual to the marketing science orientation; in particular, we
emphasize behavioural explanations, the role of habit and the modelling of
market patterns and market change. However, we also provide an extensive
treatment of the techniques and theory that underlie the cognitive approach
to consumer behaviour.

Consumer behaviour is a changing field. New research is providing answers

to questions of major importance, and in due course will give rise to a new
breed of professional marketer. All four authors are active researchers and
use their own research in this book; we hope that, in doing so, we manage
to convey the excitement that new discoveries arouse.
Exercises
Good education gives students the confidence to use and criticize ideas. We
try to enlarge this confidence through practical exercises that help the
reader to apply and reflect on ideas about consumer behaviour. The
exercises require self-appraisal, calculation, observation, measurement of
attitudes and the use of computer programs. In many cases these can be
swiftly completed, and the reader will benefit by doing these as they occur.
Plan of the Book
The book is divided into four Parts. Part 1 (Chapter 1) introduces the reader
to explanations for the different forms of consumer purchase. Part 2
(Chapters 2, 3, 4, 5 and 6) focuses on patterns of purchase; we cover
customer loyalty and brand equity, the recurrent features of mature and
changing markets and the relevant differences between cultures and
consumer segments. Part 3 (Chapters 7, 8 and 9) looks at decision-making;
we deal with methods for predicting and explaining decisions, the way that
decisions can be biased and the post-decision effects relating to satisfaction
and quality. Part 4 (Chapters 10, 11, 12 and 13) considers the responses of
consumers to conditions that affect consumption; these are price, the retail
environment, social influence and advertising.
Acknowledgements

A number of people assisted us in the production of the previous edition:

Dag Bennett, Brian Birkhead, Walter Carl, Cullen Habel, Kathy Hammond,
Bruce Hardie, Paul Marsh, Jenni Romaniuk, Deborah Russell, John
Scriven, Byron Sharp, Mark Uncles and Jim Wiley. We are particularly
indebted to Kathy Hammond, Magda Nenycz-Thiel, Cathy Nguyen,
Francesca Dall’Olmo Riley, Jenni Romaniuk, Deborah Russell and Mark
Uncles. Finally, we appreciate very much the influence of our students.
With them in mind, we have tried to be relevant and clear.
Praise for previous editions

‘A thought-provoking text that challenges readers to consider consumer

behaviour in new and refreshing ways. The approach is distinctive. Readers
are encouraged to reflect on their own experiences, as well as appreciate the
insights provided by scholarship in psychology, sociology and marketing
science. The authors not only discuss complex, one-off decisions, but also
help us to understand routine behaviours that occupy so much of daily life –
buying brands, patronising stores, watching adverts, making
recommendations.’

Professor Mark Uncles, Deputy Dean, Australian School of Business,

University of New South Wales, Sydney

‘This book provides a wonderful (and very unusual) balance between areas
of marketing that are often at odds with each other (or, worse yet, unaware
of each other) … I recommend it to any student, researcher or manager in
marketing.’

Peter Fader, Frances and Pei-Yuan Chia Professor, and Professor of

Marketing,Wharton School, University of Pennsylvania, USA

‘This textbook is exceptional for the amount of relevant research that is

presented and explained. Students who have read and understood this text
are likely to be much more of use to industry.’

Fergus Hampton, Managing Director, Millward Brown Precis, London,

UK

‘A serious, thoughtful consumer behaviour text, that focuses on substance

rather than what’s fashionable in academic circles.’

Professor Byron Sharp, Ehrenberg-Bass Institute, University of South

Australia
Part 1 Introduction
1 Ideas and Explanations in Consumer
Research
Learning Objectives
When you have completed this chapter, you should be able to:

1. Explain why it is important to study consumer behaviour.

2. Discuss the limitations of a common-sense approach to consumer behaviour.
3. Compare and contrast different approaches to decision-making by consumers.
4. Discuss the effects of the consumer environment on choice.
5. Explain how markets are usually classified.
Overview
In this chapter, we show that findings about the way in which consumers behave when
they buy and use products and services can be quite unexpected and that research is
needed if we are to answer the questions posed by marketers and regulators. We then
describe three ways in which consumer choice can occur. Following this, we introduce
some classifications that are commonly used to describe different aspects of marketing
and consumer research.
Section 1: The Scope of Consumer Behaviour
How do people buy and use goods and services? How do they react to
prices, advertising and store interiors? What underlying mechanisms
operate to produce these responses? If marketers have answers to such
questions, they can make better managerial decisions. If regulators have
answers, they can design better policy. It is the role of consumer behaviour
research to provide these answers.

In this book, we provide an up-to-date account of the main issues studied by

consumer behaviour researchers, our current understanding based on these
research findings, and we show how our understanding can be applied to
marketing problems. Knowledge has grown rapidly in some areas, and we
have reflected these advances by describing some work in more depth. In
such cases, we explain why an issue is important, how it is investigated and
what the findings are. This approach culminates in empirical
generalizations. These are general findings, based on evidence, that have
stood the test of repeated investigation. Such general findings summarize
the state of our knowledge and are useful to practitioners and researchers
alike. All too often, popular pronouncements on marketing issues contain
little evidence of this sort and it is our purpose to reverse this approach.

Where our knowledge is still sketchy, we have tried to indicate doubts about
the evidence or its interpretation. Such uncertainty propels research and as a
result creates new knowledge. Though not always welcome to students,
doubt is part of good education. Students who see the uncertainties in
consumer research should be more sceptical of unsupported opinions and
may be better placed to interpret and adapt to new findings when these
emerge. Each of the authors is an active researcher and has struggled to
understand the complexities of consumer behaviour over many decades. We
hope that this sharpens the account that we give. Inevitably, we have
omitted some fields of knowledge; in particular, we have left out some
topics that are well covered in more introductory consumer behaviour texts.

In this chapter, we introduce some general ideas about consumer behaviour

and marketing that are explored further in following chapters. In Section 1
of this chapter, we look at the sorts of question raised by marketing and the
answers that are provided by consumer behaviour research. Section 2 will
discuss models that provide descriptions of consumer decision processes
and Section 3 will focus on some of the classifications and explanations
that we use to describe different types of consumer research.
Questions and Answers
There is a close affinity between marketing and consumer behaviour. In a
sense, marketing is a customer of consumer research. Marketers want
answers to a number of problems raised by their practices, and consumer
researchers can provide these answers. Examples of marketing practices
are:

the use of price incentives

customer loyalty programmes
the use of particular colours, music and aromas in the retail
environment
launching new products using existing brand names (brand extension).

Often, the direction of an effect fits common sense; for example, consumers
buy more when the price is dropped. However, the benefit of a discount
depends on the amount of extra purchasing generated by, say, a 10 per cent
price cut, and here common sense does not supply an answer. For informed
action, we need to conduct systematic research, which allows us to measure
the size of any effect. Evidence is gathered using the methods of market
research, psychology and the social sciences. Using such methods, we seek
answers to such questions as:

How much do sales change when the price of a product is cut by 10

per cent? What happens to sales after a discount has ended? Why do
these effects occur?
How much do colours, music and aromas affect consumers’ behaviour
in a store? Which underlying mechanisms explain any effects we see?
When a new product is launched under an old brand name, how much
does the old name affect purchase of the new product?

Another set of questions comes from legislators and regulators, who have to
set the rules that affect marketing. Examples of their questions are:

How do consumers react to product benefits such as increased energy

efficiency or high nutritional value? What explains their behaviour?
 Do childproof packs save lives? How are such packs used?

Sometimes, marketers give little attention to the explanation for an effect.

An example is the identification of specific groups who are very heavy
buyers of a product. If such people can be identified, they can be selectively
targeted. This type of empirical approach can work well but an explanation
still helps. If marketers know why some groups buy a product much more
than other groups, they may be able to design communications that
capitalize on this and also predict other products that these groups will
want.

In any applied subject, practitioners need to use their judgement when

evidence is lacking. Those who have to take decisions cannot delay action
until problems have been fully researched. However, it is important that
marketing practitioners do accept new evidence when this becomes
available. Some apparently sensible practices may need to be adjusted
because of new findings.

For example, it is generally assumed that consumers across Asia are

homogeneous, and accordingly their value perceptions towards the
consumption of luxury items should be similar. Marketers usually adopt a
standardized strategy for the region. In a multi-country study, Shukla, Singh
and Banerjee (2015) show that this assumption is erroneous. How
consumers value luxury differs across countries within Asia. Managers of
luxury brands, therefore, need to adapt according to regional differences.
Importantly, this type of work reminds us about the need for empirical tests.
Section 2: Consumer Decision Models
The traditional approach to problems in consumer behaviour used to employ
a comprehensive model of the purchase decision process. Such models were
often the centrepiece of undergraduate consumer behaviour texts and were
expressed with boxes and arrows representing all the components and
connections of an elaborate rational decision. In these models, a consumer is
supposed to attend to product information and process it into their memory.
The consumer retrieves that memory when a need emerges, and after a
further search and evaluation of all relevant alternatives a purchase is made.
After this, post-purchase evaluation may create satisfaction or dissatisfaction
with the chosen product, and this can result in a review of needs for later
decisions. Figure 1.1 shows the basic form of such a model.

Figure 1.1 Is this how you choose?

These days, there is less enthusiasm for such models. One problem has
always been that they are hard to test because it is difficult to find
satisfactory measures for all the components (Ehrenberg, 1988). Another
problem with comprehensive models is that they overstate the rationality of
how consumers choose. If there is plenty of time and the decision is
important, then sometimes people will discover all the alternatives, evaluate
these, and then select the one that seems to be the best, but we know from
our own experience that we often simplify the process. Sometimes, we will
choose first and justify our behaviour afterwards – if we justify it at all.
Thus, although rational decision models might suggest what people ought to
do (normative), they are a poor guide to what people actually do
(descriptive). In practice, managers want to know what people actually do
since it is this behaviour that they are seeking to influence.
Textbooks now give more attention to ‘partial decision models’ where the
rationality of the process is incomplete; also, it is accepted that much repeat
purchase occurs automatically as a habit. Often, this range of decision-
making from rational to automatic is related to the degree of involvement
with the product. People are likely to be more involved and give more
thought to the choice when they are buying something for the first time and
this has important outcomes. To explain decision-making in more detail, we
focus on three models of consumer decision which have different
implications for managers (see Box 1.1). The models are:

1. Cognitive – treating purchase as the outcome of rational decision-

making processes.
2. Reinforcement – treating purchase as behaviour which is learned and
modified in response to the opportunities, rewards and costs present in
the consumer’s environment.
3. Habit – treating purchase as already learned behaviour, which is
elicited by particular stimuli in the consumer’s environment.
Box 1.1 Models of consumer choice
The cognitive model – this assumes rationality. The decision rests on beliefs about
alternatives, which are investigated and compared. Marketers can influence cognitive
decision-making by providing information that leads the consumer to prefer or reject
alternatives.

The reinforcement model – choice is controlled by factors in the environment that

reward and facilitate some alternatives more than others. Marketing influence is achieved
by changing the consumer’s situation. However, what is rewarding to some persons may
not be so to others and this limits influence.

The habit model – choice is controlled by managing stimuli (brand name, logo, pack
features, etc.) that have become associated with a product as a result of past purchases.
Sometimes this is called stimulus control.
The Cognitive Model
When consumers make an important purchase for the first time, they may
reflect on alternatives and discuss the pros and cons with others with the
intention of securing benefits and avoiding costs. This model, sometimes
called extended problem-solving, has always had its critics. Olshavsky and
Granbois (1979: 98–9) noted:

for many purchases a decision never occurs, not even on the first
purchase … even when purchase behaviour is preceded by a choice
process, it is likely to be very limited. It typically involves the
evaluation of few alternatives, little external search, few evaluative
criteria, and simple evaluation process models.

It is quite hard to find behaviour that fits the elaborate sequence of extended
problem-solving. Beatty and Smith (1987) found that people did not search
much before the purchase of durables and Beales et al. (1981) found that
few people in the USA consulted Consumer Reports. Carefully thought-out
decision-making is only likely for first purchases but these are quite rare,
even in consumer durable markets, since most purchasers are either buying
a replacement for an existing product or making an additional purchase. In a
study of white goods purchases in the USA, Wilkie and Dickson (1985)
found that two-thirds of the purchasers had bought the category before and
Bayus (1991), quoting US industry sources, found that 88 per cent of
refrigerators and 78 per cent of washing machines were replacements. In
these circumstances, a carefully thought-out comparison of brands is likely
to be the exception rather than the rule.

But, when it does occur, is a carefully thought-out decision likely to result

in the best choice? When people attempt to be rational about a first-time
choice, they may make mistakes because they lack experience. However,
they are likely to make a better choice than those who abandon any rational
processing and plump for an alternative (see Box 1.2).
 Box 1.2 When a pension is converted into
an annuity
People build up pension funds over their working lives and may then convert the
accumulated investment into an annuity when they retire. They may use their pension
company for the annuity or search for better value from another company. According to
Hargreaves Lansdown, a large financial services firm in the UK, the majority of people
will buy their annuity from their pension company. Since annuity rates across pension
companies can vary by as much as 15 per cent, this careless choice can mean that many
retirees lose income that they could have enjoyed for the rest of their lives. The most
likely explanation for this behaviour is that the retirees have a very poor understanding of
the issues and they plump for the company that is familiar.

The tendency to simplify decision-making is also observed in industry. One

study of investment decisions in British industry revealed that these were
often made first and then justified later. Marsh et al. (1988) found that
faulty financial analysis and a lack of coherence with stated strategic
objectives were common in major acquisitions and that the company
rulebooks were often ignored. More generally, industrial decisions often fit
a ‘satisficing’ model (Simon, 1957). Simon describes how executives tend
to accept the first option that is good enough to solve a problem; this means
that there is little comparison between alternatives. Klein (1989) found that
many decisions in operational settings follow a pattern that is consistent
with Simon’s ideas. Typically, people assess the situation and generate a
prospective action based on this assessment. Then, they evaluate this action
to see whether it will provide a solution. If it fails, they generate another
prospective action and evaluate this, but they do not usually compare
prospective actions.

When the satisficing model applies, the order in which products are
evaluated is important since the first satisfactory solution will be the one
that is adopted. This means that more prominent alternatives have a better
chance of selection (see Box 1.3). Managers and marketers may be able to
use this fact to their advantage by keeping awareness of their brands high in
consumers’ minds.
 Box 1.3 Diagnosis
Even in medicine, decisions may be simplified. Often, the symptoms are assessed and a
preliminary diagnosis is made, taking account of common illnesses; then other symptoms
are checked to see whether they confirm this diagnosis. Only if these other symptoms fail
to support the first diagnosis is a second one considered. This procedure may lead to the
over-diagnosis of common illnesses.

These examples of decision-making in industry and medicine suggest that

the simplification of choice is the norm rather than the exception and we
might expect consumers to follow much the same pattern. For example, if
the freezer needs replacing and a preliminary inquiry establishes that there
is an appropriate model in a convenient shop, consumers may complete the
purchase there and then. If the shop does not offer a suitable freezer, they
may then turn to other stores and look at other models. On the Internet, the
comparison between alternatives is made easier but consumers may still cut
the process short by making a satisficing choice.

Although satisficing may not result in the optimal solution, it may use time
efficiently when this is scarce. However, when the outcome of the decision
is important, consumers and managers would make better decisions if they
forced themselves to consider a second alternative before deciding.
Influences on Decision-Making
It is easy to fall into the trap of assuming that decisions are made by people
acting on their own. Many choices are made in groups, and even when
people decide on their own they are often influenced by word of mouth
received previously from other people. At other times, people may base
their decisions on information received through the mass media (e.g.
advertising, newspaper, television and Internet reviews). People are
particularly likely to seek advice on matters that are obscure or difficult to
test in other ways; this is common when the recipient of the advice is
choosing for the first time or acting under changed circumstances, such as
when they move home and need to find service providers such as a dentist.
In later chapters on word of mouth and advertising, we consider in more
detail how these influences may affect choice.

Since advice affects consumer decisions, marketers need to take account of

this process. For example, advertising can include information that is easily
passed on in conversation, and the design of the ad can reflect the process
of giving advice. However, word of mouth is under consumer control, not
marketing control, so normally marketers can only affect it indirectly.
Exercise 1.1 Decision-making
Identify an important purchase that you have made, for example a holiday, electronic device,
financial investment or education course:

Were you clear about what you wanted?

How much investigation did you do before purchase?
Did you consider one option and move on to others if it was unsuitable, or did you keep
several alternatives in mind before choosing?
Did you use the Internet to search for others’ opinions?
Did you consult friends or relatives?

In retrospect, you may be able to see defects in your decision-making process. Often, we will
lack enough prior experience, time or motivation to fully compare the options.
Purchase as Learned Behaviour
A person’s environment controls behaviour in two ways. First, the
environment makes some actions possible and other actions impossible to
perform; for example, some physical items can only be bought if they are
stocked by retailers. Second, when actions lead to positive outcomes they
are more likely to be repeated, and conversely negative outcomes make it
less likely that the action will be repeated. These reinforcement effects on
behaviour have been examined in learning theory; this is a systematic
description of the relationship between initial behaviour, its outcomes and
subsequent behaviour. Learning theory is relevant to both the reinforcement
and habit models.
Reinforcement
Early research in learning theory was done by Thorndike (1911), who
confined a hungry cat to a cage and placed food outside. The erratic
movements of the cat eventually released a simple catch and the cat
escaped. The cat took less time on subsequent trials and eventually it
released the catch immediately when it was placed in the cage. Thorndike
called this trial and error learning and it has some relevance to
consumption. People entering new markets are faced with a range of brands
and may make near random trials of alternatives until they come upon a
brand that they like.

In Thorndike’s work, the cat’s actions were driven by the outcomes: gaining
food and freedom. Skinner (1938, 1953) called such outcomes reinforcers.
He defined a reinforcer as an experience that raises the frequency of
responses associated with it, while a punisher reduces the frequency of such
responses. Reinforcers may be rewards or reductions in cost, while
punishers may be costs or reductions in reward. Reinforcement has most
effect when it occurs at the same time as, or just after, the response. Skinner
placed an emphasis on the way in which reinforcement changes the
frequency of the response, but reinforcement also strengthens the
association between stimulus and response and this is important for the
habit model. Figure 1.2 illustrates the effect of reinforcement.

Figure 1.2 Reinforcement learning

The principles of reinforcement are applied in many sales promotions, such
as discounts that offset the cost of a product. Skinner also introduced the
idea of shaping – the process whereby behaviour is gradually shifted from
one form to another by selectively reinforcing those performances that
show change in the desired direction. Shaping is sometimes apparent in
sales techniques where the salesperson moves the prospect towards the
sales goal by reinforcing shifts in the preferred direction with nods,
agreement and approval. Products also shape us. We become more expert at
using computers and cars, partly because of the reinforcers that such
products deliver; as a result of this, we may seek more sophisticated
models.

Learning can be reinforced each time a response is produced, i.e.

continuously, or it can be reinforced intermittently. Learning is faster if the
reinforcement schedule is continuous but the final effect of a given amount
of reinforcement is greater when it is used intermittently. This helps to
explain why people are prepared to lose money by gambling on fruit
machines. The cost of playing a slot machine is a fairly continuous
punishment but the machine rewards intermittently. Over time, the gains are
fewer than the losses, but the effect on behaviour of the irregular reward is
greater than the effect of the regular cost.
Both stimuli and reinforcers can lose their effect if they are used too
frequently. Stimulus satiation, called desensitization, helps people to put up
with recurring unpleasant experiences. An important effect of
desensitization in consumer behaviour is the way in which people get used
to conditions that are inadequate or unpleasant, and as a result may not
complain or demand compensation. Examples of this are the way people
tolerate litter in streets, overcrowding on public transport, and being kept
waiting ‘on hold’ on the phone. Similarly, consumers may put up with
defective goods because they have grown used to the defects. Examples are
the continuing use of lumpy mattresses, broken refrigerator shelves, and
inadequate carving knives. The job of the marketer is to overcome the
inertia in these situations so that the consumer sees the problem afresh and
seeks a solution.
Stimulus Control: Classical Conditioning
One type of learning, called classical conditioning, was studied by Pavlov
(1927). Pavlov noticed that dogs started to salivate at the sight of the person
who fed them. The older dogs showed this most, and Pavlov thought that
over time the salivation reflex that normally occurred at the presentation of
food had become associated with a new stimulus –the dogs’ handler. Pavlov
set up a series of experiments to demonstrate this process of classical
conditioning using the sound of a buzzer as the conditioned stimulus instead
of the dogs’ handler. Figure 1.3 illustrates this process.

Classical conditioning has considerable relevance to consumer behaviour.

Packaging, brand names, colours, smells, music and the contexts of purchase
and consumption may become associated with the buying of particular
products. Some advertising is clearly intended to forge associations between
brands and particular stimuli that can be used in further advertising and at
the point of sale – for example, McDonald’s and the big ‘M’ sign, ‘i’ and
phone, pod, pad … and, more generally, all logos and their respective brands
and companies. The idea here is that the conditioned stimulus may help in
identification and add to purchasing tendency. It is also noticeable that, to
compete in some markets, manufacturers have to adopt the colours and pack
shape that are conventional for that type of product. The power of such
associations is revealed by a trip to an unfamiliar country. The absence of
familiar features makes the high street confusing. A simple task, like posting
a letter, requires investigation and effort in order to identify the colour, shape
and location of the postbox.

Figure 1.3 Classical conditioning: Pavlov’s experiment

A stimulus that is associated with a rewarding product may induce a more
generalized tendency to buy other products that appear similar. A direct
application of such generalization in marketing is the use of an existing
brand name for a new product. By this process of brand extension, some of
the buying tendency (often termed propensity by marketers) that consumers
have for the old brand may attach to the new brand. For example, Mars used
the positive consumer propensity towards the brand when introducing Mars
ice-cream, and this was helped by the similarity in the appearance of the ice-
cream and the confectionery bar.
Habits of Purchase
The cognitive and reinforcement models emphasize the modification of
consumer behaviour and thus may explain the changes that occur in our
purchasing. However, much consumption has a settled form: people buy the
same brands and use the same stores over long periods. This repetitive
purchase is of great value to firms.

We say that people have a habit when they regularly produce much the
same behaviour on encountering a particular stimulus. In the case of
supermarket goods, important stimuli are the colour, size and shape of the
pack. Williams (1966) found that colour positively affected buying
behaviour most, followed by size and then shape. Response to such stimuli
is automatic, so that no conscious thought is required when we pick a
laundry detergent brand in the supermarket. Habits sidestep cognitive
decision-making and leave us free to concentrate on other problems where
experience does not provide us with a ready response. However, even in
novel situations people may trade on already acquired habits. Consider the
person who is about to buy a car for the first time. Most first-time car
purchasers are familiar with cars, have been to car showrooms before, may
have bargained for goods before, may be knowledgeable about the ways of
salespersons, and may understand credit arrangements. Thus, even first-
time car purchasing may draw on previous learning, some of which may
have become habitual. Viewed in this way, even complex and novel
behaviour may call upon behaviours in a habit repertoire.

The habit model of consumption excludes planning before action but does
not imply that consumers never think about their habitual behaviour. People
may reflect on their actions after purchase either because of discussions
with others or because their purchase outcomes were exceptionally good or
bad. But this is unusual; generally, habit restricts experimentation, and as a
result consumers may be unaware of improvements in products from which
they could benefit. This suggests that, although habitual purchase is
frequently satisfactory, it is not always the best solution. Exercise 1.2 draws
your attention to habits you may have which may not best meet your needs.
 Exercise 1.2 Habits
It is hard to detect habits that work against your own interests but consider these two areas:

1. Taking sugar in tea and coffee is a habit that adds to body weight and contributes to
tooth decay. When people give up sugar, they get used to it fairly soon, and after a few
weeks may prefer unsweetened tea or coffee. Is this not a habit worth changing?
2. If you make a regular journey to work, is the route optimal? People can discover journey
improvements after years of using a less suitable route that has become habitual.

How should marketers present new brands in markets where purchase is strongly habitual?

When purchase is habitual, a new brand must be marketed in a way that

disrupts habit and provokes a review of past purchase. This is not easily
achieved. Advertising may be ignored, while discounts and free samples
may be used by consumers without much effect on later purchase. Most of
the time, consumers will carry on buying what they have bought before. But
that’s why marketers need to understand which types of marketing
interventions work best for different customer groups in different
environments.
How Free Are Consumers?
It is often claimed that the consumer is king but this may exaggerate the
flexibility of action that consumers have. To be free you should be able to
choose from more than one option without pressure, and be able to reject all
options if they are unattractive. Many choices are controlled by the
consumer’s environment rather than by reflective thought by the consumer,
and this casts doubt on how much freedom of action consumers can
exercise.

The constraints on consumers are considerable and are not just

environmental. Consumers may lack knowledge of alternatives when these
are not displayed. Sometimes, people have to use products; they must put
petrol in a car and laundry detergent in a washing machine, and the fact that
they have a choice between near identical brands is often, to them, a matter
of indifference. Freedom of action is also affected by limited access to
goods and services, by physical dependence on products like cigarettes and
alcohol, and by psychological dependence when the consumer is a
compulsive purchaser or gambler.

People also do many things that they would prefer to avoid, such as going
to work on congested public transport and waiting for flights in airports. In
many areas, such as education, medicine and legal advice, the opportunity
to influence a service by withdrawing custom or complaining is effectively
limited by the continuing need to use the service. There are other areas
where a lack of money prevents people from doing the things they might
wish to do; large houses and luxurious cars are possible for only a few. For
these reasons, we are sceptical of claims about the almost unlimited choice
available to consumers and how much autonomy they have. However, the
growth of the Internet has raised access to knowledge about goods and
services and assisted purchase; this may lead to a genuine increase in
consumer choice.
Decision-Making on the Internet
Increased use of the Internet and the facilities that websites offer may change
the rationality of choice. The ability to compare prices online generally
drives down the average price of goods and services bought online. The
proportion of UK shoppers who say they often consult the Internet before
making a purchase was 78 per cent in 2011, unchanged from the previous
year (Nielsen, 2011a). However, shoppers seem to be making more
purchases on the Internet. According to the Office for National Statistics
(UK), daily Internet use more than doubled in eight years to 2014 (see
Figure 1.4). In the UK, 74% of all adults reported buying goods or services
online. This had risen from 53% in 2008, with clothes the most popular
online purchase, bought by 49% of all adults (ONS data, 2014).

Figure 1.4 Changes in the use of the Internet by UK shoppers (Office for
National Statistics, UK, Statistical Bulletin 2014)

The Internet makes it easier to compare prices and specifications, and can
take some of the effort out of shopping. Search engines such as Google.com
assist in the identification of sources and products, while chatrooms and
blogs often provide user comment on different brands. Comparison sites
such as Shopping.com show the prices charged by different suppliers. Other
sites, such as Uswitch.co.uk, can compute the best value among service
providers and may facilitate transfer to a new provider. Websites for those
buying houses, shares, books and many other items aid choice by providing
easy comparison between alternatives. For example, an Australasian buyer
can use a site such as realestate.com.au to specify properties by location,
price and type, and can then inspect pictures of interiors. This helps to focus
attention only on those properties that meet the needs of the buyer. A
subscriber buying shares through a Web-based stockbroker such as
Hargreaves Lansdown (www.hl.co.uk/) can see the past return on specific
shares over different periods, and can compare this performance with other
shares and with standard indexes. On Amazon.co.uk, customers can read
reviews of a book before buying and be provided with information on new
books that are related to their previous purchases. On airline sites such as
ba.com, a traveller can pick travel dates that are cheaper. Quite clearly, the
Internet can be used by consumers to better assess alternatives, but how
much do consumers do this to improve their choices and lower their costs?

A study by Zettelmeyer, Morton and Silva-Risso (2006) suggests that

Internet customers may bring down the price that they pay for cars by an
average of 1.5 per cent. However, consumers who use the Internet may be
more price-sensitive and these people might also drive a hard bargain in an
offline context. In addition, it appears that even Internet customers rarely
secure the lowest price. According to Shopping.com, 80 per cent of Internet
customers pay more than they have to. It seems that use of the Internet to
obtain better value is restrained by loyalty to particular websites. Once they
are familiar with a site, consumers may return to it later because it is easy to
use and saves time. A consumer may agree that a book might be cheaper
elsewhere but still use Amazon because of convenience – and this
convenience can be considerable; customers may allow trusted sites to store
their credit card and delivery details so that purchasing really is simplified.
Similarly, buyers normally use one online grocer because of the trouble of
getting to know another site. In short, habits take over.

This evidence presents a somewhat confused picture. The Internet can assist
people to make better decisions and buy more cheaply, but the technology
may discourage experimentation when goods are regularly bought. In
addition, there are some sectors, such as grocery, where choice limitations
and delivery cost may raise the price that is paid online.
Section 3: Classifications And Explanations
Disciplines must organize and classify information in order to explain it.
Marketing is no exception and uses a number of classifications, some of
which are shared with other subjects. We start with one distinction that is so
ubiquitous that we scarcely notice it. This is the use of comparison in the
assessment of evidence.
The Principle of Comparison
Any judgement rests on implicit or explicit comparison. When we say ‘that’s
cheap’, we are comparing the price that is presented with some standard. That
standard might be given by another brand that is physically present, or it might
be an internal standard that we have built up from experience. Such
comparisons are fundamental to human judgements. We make sense of any raw
data by comparing these with objective standards, or with personal or social
norms. Comparison also occurs in the scientific assessment of findings. To
illustrate this, consider Table 1.1.

This table shows the ratings that owners gave to their car compared with the
best alternative that they could have purchased instead. The data come from an
Internet survey of 495 owners conducted in the UK. The numbers show that 64
per cent of respondents thought that their car was better than the best
alternative and 9 per cent thought that it was worse. This seems to show great
confidence among respondents in their choice of car. Our finding reflects a
general phenomenon called the endowment effect: objects are rated more
highly once they are owned (Kahneman, Knetsch and Thaler, 1991a).
However, the assessments shown in Table 1.1 are difficult to justify. When
there are many alternatives, which are difficult to fully compare, it is quite
likely that another brand would have been better than the one chosen. Thus,
there seems to be an optimism bias in the assessment of possessions which is
revealed by making the question comparative. In Chapter 8, we study these
judgemental effects in more detail.

Sometimes, the standard of comparison that people use for judgements has an
objective basis: for example, the average price of a basket of goods in the
different supermarket chains or the fuel economy of different cars. But notice
that consumers have to discover and accept such standards if these are to affect
their judgements. Standards may be affected by marketing communications,
but mostly people appear to acquire price or quality norms from experience.
Such internal norms will be based on observations, discussions with other
consumers and information from the media, and are likely to be quite stable. In
these circumstances, what changes when marketers are successful in modifying
consumer behaviour? Usually, marketing activity alters the immediate
perception rather than the internal norm. When the price is cut, and more
people buy, it is because the new price is seen as cheap, compared with the
norm.
Categories, Brands, Variants and SKUs
Classifications are also made on the basis of the context in which decisions
are taken. We call anything a customer buys, whether a good or service, a
product. Then all products are divided into categories such as soup, wine,
mobile phone airtime supply, cars and hotels. Within a category, there will
be a number of brands available for consumers to choose from. Brands are
easily recognizable entities – such as Apple, Toyota and Disney – and
customers can become attached to one brand rather than another when
making repeated choices. Sometimes there are sub-brands, for example
Volkswagen has Polo, Golf and Passat. The branding is signalled primarily
by name, but also by logo, and the shape, colour and design of the pack or
product when this has a physical form. Advertising may attach other
associations to the brand, such as cartoon animals and musical themes. In
many cases, a company name is synonymous with the brand, such as BP,
but in other cases, the company owns a variety of brand names – for
instance, General Motors, Procter & Gamble and Unilever each manage
many brand names (e.g. Unilever owns Ben & Jerry’s, Bertolli, PG Tips,
Dove, Lynx and Timotei among other brands). Variants are subdivisions of
the product type so the Volkswagen Passat is available as a saloon or an
estate, and Heinz soups are available in different flavours or pack sizes.

In business, the term SKU (stock keeping unit) is used widely. This is a
unique combination of brand, variety, pack size, etc. that is required for
manufacturing and filling the shelves. The SKU is barcoded so that
automated systems can specify it in production scheduling and stock
control. Manufacturers and retailers often analyse consumer choice at the
level of the SKU. Research by Singh, Ehrenberg and Goodhardt (2004)
shows that buyers switch between SKUs in much the same way that they
switch between brands. In both cases, switching is related to the market
share of the SKUs.

The consumer’s purchasing of specific brands from specific outlets at

specific prices controls the profit that is made by the brand owner, retailer
and other suppliers; marketing activities are therefore coordinated to
promote brand and outlet preferences. This means that the branding must be
distinctive enough for consumers to distinguish one brand from another, but
at the same time the brands in a category often have features in common,
such as pack size, colour and shape, which help the consumer to recognize
what they are buying when they search for the product on a shop shelf or on
a website. In fact, one brand does not have to be physically different from
other brands in a category. At one time, Volkswagen, Seat and Ford offered
SUVs that were almost the same except for the name badge and price.
Similarly, there may be no detectable difference between the granulated
sugar offered by two different manufacturers; consumers know this, but this
does not stop them from regularly buying one brand over another. Often,
each brand will cover much the same range of variants. Sugar brands will
offer granulated, castor, icing and Demerara variants; soup brands will have
much the same range of flavours; and car brands will be available in SUV,
sports, saloon and estate forms. In fact, the differences between the variants
of a single brand are often much greater than the differences between the
corresponding variants of different brands.

Singh, Ehrenberg and Goodhardt (2008) show that product variants can
attract markedly different levels of loyalty. These different loyalty levels are
found to be closely related to the variants’ market shares – higher loyalty
predictably goes with higher sales. Some variants were found to be very
popular, and some were bought by only a small fraction of the market.
However, neither large nor small variants seem generally to attract a special
or unusually loyal customer base. Although product variants have their own
specific functional differentiation, this is seldom the focus of advertising
and promotion (except perhaps at launch). Instead, they are expected to ‘sell
themselves’, by their labelling, shelf space and familiarity. A problem here
is that, when the variants of different manufacturers are very similar,
advertising the variant may assist the sales of other manufacturers as well.
Choosing between product attributes and the variants offered remains
complex (e.g. Sharp and Dawes, 2001) and is subject to substantial market
research using focus groups, trade-off analysis and modelling.

In many categories, brands compete only with each other for the customer’s
attention, such as Colgate versus Aquafresh toothpaste. However, in the
food and entertainment fields this is less true. A frozen meal brand
competes with home cooking and restaurant meals, as well as with other
brands of frozen meals. Similarly, beer competes with wine and ten-pin
bowling competes with the cinema.

Other differentiators beside brands are used to distinguish one market

offering from another. An interesting example is provided by wine. French
wine has traditionally been branded by the producer, sub-region and region
– for instance, Château Cheval Blanc is a St Emilion production in the
Bordeaux region. There are many other producers, and this produces a
complex choice for the consumer. By contrast, Australasian wine is sold
more on the basis of grape variety. Although there are many varieties of
grape, a small number dominate the field (e.g. Cabernet, Chardonnay,
Malbec, Merlot, Pinot Noir, Reisling, Sauvignon, Shiraz and Tempranillo)
and several of these varieties are grown in each region. Grape variety
creates major differences between wines and provides an easy ‘handle’ for
the consumer. When the grape variety has been chosen, regions like the
Barossa, producers like Penfolds, and the year of the vintage may be used in
the choice process of the more discriminating buyer.
Goods and Services
A familiar grouping of categories is into goods and services. A good has a
physical form, such as a can of soup or a bed, or in business-to-business
(B2B) markets, aluminium sheets or bus wheels, whereas a service is
intangible and is used by the recipient as it is created, such as a haircut, a
visit to the dentist, professional advice or a phone call. Thus, the essence of
a service is that it exists in time and must be consumed at that time if a loss
of sale is to be avoided. By contrast, goods such as frozen peas can be
stockpiled by the retailer and supplied when there is demand. Most service
products incorporate a goods component, for instance the meal is consumed
in a restaurant and your phone call is made from an electronic device such
as a phone handset or a computer.

Goods can be subdivided into classes such as groceries, electronics and

fashion, or in B2B markets, electronic components, food commodities, etc.
Similarly, services divide into classes such as telephony, transport, surgery,
entertainment and financial services. The fact that there are textbooks
devoted to the marketing of services suggests that this is substantially
different from the marketing of goods. One difference is that, because they
are delivered over time, services can suffer problems of uneven demand,
leading to an inefficient use of resources and delay and frustration among
customers. We cover research on the consumer response to delay in Chapter
9. There are also differences that arise from the interaction between service
supplier and customer; this cannot easily be standardized because customers
differ and attempts at uniform treatment may cause dissatisfaction. Most
goods can be examined before purchase and this helps consumers to find
what will be satisfactory. Services cannot usually be examined in this way,
and as a result those who are thinking of adopting a new service provider
may seek advice from existing customers, whose word of mouth provides a
proxy for personal experience. In other respects, goods and services are
similar. Our three models of consumer decision-making apply to both, and
so does the distinction between repertoire and subscription categories that
we discuss below.
Vargo and Lusch (2004) have suggested that services rather than goods are
the fundamental product form since goods are made by the service of
workers. This new ‘dominant logic’ in marketing has echoes in the work of
early economists, particularly in Marx’s theory of value, first expounded in
Capital (volume 1 was published in 1867), which relates value to the labour
input. At the time, economists argued that the value of goods was defined
by an exchange process; it is what others are prepared to give for the goods
and no amount of labour input will raise the price of something that people
do not want. The character of transactions may have changed and become
more cooperative, but in our view such exchanges remain the basis of
value. Marketers must be concerned with profitable trading, and for this
reason we are sceptical about making service fundamental in marketing.

Service failures are a common concern, and service providers struggle to

understand the psychology behind customer perceptions of a fair response
by the supplier after failure (see Singh and Crisafulli, 2015 for a review of
service recovery after failure). This topic is considered in more detail in
Chapter 10.
Repertoire and Subscription Markets
Categories can be divided into those that are repertoire, where consumers
commonly purchase more than one brand over a fairly short time such as a
year (e.g. groceries, restaurants and airline tickets), and subscription, where
consumers mostly use only one brand at a time (e.g. bank accounts, dentists
and refrigerators). Research by Sharp, Wright and Goodhardt (2002) shows
that most categories fall clearly into either the repertoire or the subscription
division. In repertoire markets, we can measure a type of behavioural brand
loyalty called share-of-category requirement (SCR). This is the percentage
of category purchases that a customer gives to a specific brand over a
period. For example, if a person buys instant coffee on ten occasions in a
year and five purchases are Nescafe, the customer’s SCR for Nescafe is 50
per cent. By contrast, loyalty in subscription markets is shown at the time of
repurchase when the customer either retains the brand or switches to
another.
Market Concentration
In many categories, there are relatively few brands. Laundry detergents,
toothpaste and mobile phone airtime supply are examples. In other fields,
such as wine, cheese and biscuits, there are a great many producers, none of
which commands a large market share. In some other fields, such as
supermarkets, fashion stores, chemists and investment advisers, a few large
chains compete with many smaller suppliers in western markets. When a
few producers command a large part of the category, we describe the
market as high concentration. Usually, large suppliers are more profitable
because of economies of scale in manufacture, distribution and advertising.
Retailers feel compelled to stock more familiar brands because of demand
and this helps the manufacturer (the brand owner) to maintain the price paid
by the retailer.

Consumers are not necessarily disadvantaged by high market concentration.

The large scale and efficiencies of big producers mean that product
development can occur and the wide distribution of big brands ensures that
consumers can easily find the larger brand. One concern is that high
concentration may reduce competition but it is not difficult to find high
levels of competition in concentrated markets. For instance, the worldwide
cola market is highly concentrated, yet both Pepsi and Coca-Cola remain
fiercely competitive suppliers.
Market Share
A brand’s market share is a robust indicator of its customer loyalty. Ehrenberg
(1988) has explained that many aspects of aggregate buyer behaviour can be
seen as an outcome of market share. For example, the average SCR loyalty for
a big brand tends to be higher than that for a small brand (i.e. the evidence is
that, in general, the bigger the brand the more likely it is that customers will
buy it again compared with the customers for smaller brands). In Table 1.2, we
illustrate how another variable, the share of recommendation, relates to market
share in the mobile phone category, using data gathered before the advent of
smartphones.

Table 1.2 shows that the share of brand recommendations closely follows the
market share of the brand. There is no mystery about this. As we saw earlier
with regard to cars, people are usually happy with the products that they own,
and East, Romaniuk and Lomax (2011) found an average of 71 per cent of
recommendations related to the informant’s main brand. So, the bigger the
brand, and therefore the greater the number of users, the larger will be the share
of recommendations. For this reason, managers need to take account of market
share before they assess the word of mouth about their brand. In Table 1.2,
Motorola is doing well because the rate of recommendation is ahead of market
share. If the rate of recommendation was assessed without taking account of
market share, Nokia would come top, but we can see that its performance is
just average for its size.
Exercise 1.3 Do big brands get more, or less,
negative word of mouth?
Recommendation is positive word of mouth. What about negative word of mouth? Develop
ideas about how negative word of mouth is produced. What will be the resulting relationship
between market share and the share of negative word of mouth?

When you get to Chapter 12, you will see our evidence on this topic.
Consumer Segmentation and Causal
Relationships
We often compare population segments: those who retain a brand versus
those who switch, heavy TV viewers versus light viewers, high
recommenders versus low recommenders, men versus women, etc. If we
have evidence about the consumption habits of different segments, we can
target those that appear to be most likely to purchase the category or most
open to switch brands. Consumer segmentation is an approach that is very
popular in marketing; it can work well even when we do not know why the
behaviour of one segment differs from that of another. For example, a
method used by those trying to harness word of mouth is to try to identify
those consumers who give more advice than others (the influentials). Once
they have been identified, the job of the marketer is to recruit them on
behalf of a promoted brand.

However, in consumer behaviour, we want to explain behaviour, preferably

by finding causes for it. Why is it that one segment is more active in giving
advice to others than another segment? We can investigate how segments
differ with respect to possible causes. As the picture of the different factors
underlying recommendation builds up, a new strategy becomes available to
marketers. Instead of identifying a segment that gives more word of mouth,
marketers can try to influence the factors that cause word of mouth, and this
can be done without identifying the influentials.
Behaviourism and Cognitivism
Does a change in thinking cause a change in behaviour, or does a change in
behaviour cause a change in thinking? The answer is that we can find
support for both processes. In psychology, the primacy of behaviour is
called behaviourism. This approach was developed by Skinner (1953). The
traditional behaviourist rejects the idea that thought and feelings are the
initiators of action. Instead, action is explained by reference to the
environmental circumstances that act on a person. This fits the
reinforcement and habit models of consumer decision-making.

In traditional behaviourist research, it used to be believed that thought and

feeling are effects but not causes; like ripples on the surface of a pond, they
indicate the fish’s movements but do not move the fish. If this account is
correct, we can use people’s thoughts and feelings as indicators of their
potential behaviour but not as explanations for it. Such narrow
behaviourism is usually rejected today. One reason is that it is difficult to
describe action without taking account of the thoughts and feelings that lie
behind it; words become insults or praise only through an understanding of
the motives of the person uttering them. The traditional behaviourist
position is not subtle enough to deal with this complexity in the nature of
human behaviour.

Opposed to behaviourism is the view that thought and feeling can produce
change in action directly. This is cognitivism and it lies behind rational
accounts of consumer decision-making. In its strongest form, experience is
interpreted and used to change attitudes and knowledge, which then control
behaviour. Thus, from a cognitivist perspective, behaviour may be modified
by communications that change attitudes and knowledge. Some support for
the cognitivist position can be found in the way public information
campaigns change behaviour (for example, anti-smoking advertising,
featuring the hazards of smoking, has been shown to be effective; see
Chapter 13).

There are also examples where behaviour precedes attitudes that support
behaviourism. Clare and Kiser (1951) asked parents of completed families
about the number and sex of the children that they thought were desirable.
There was a strong tendency for parents to prefer both the size and the sex
mix of the family that they already had; for example, if they had two girls
they stated that they felt two girls were what they would like if they were to
have their family again. At the time of the study, there were no ways of
controlling the sex of offspring, so a preference for the same sex balance
can only be explained as a product of experience.

In many other cases, the causal direction between attitude and behaviour
may be in doubt. The preferred number of children is a case in point.
Parents might have had two children because they wanted two; or, having
had two children, they might have come to prefer this number. Such
alternative explanations can often be seen in the social sciences. For
example, Marx argued that it was not ideology that determined social
relations but social relations that determined ideology. This is the
sociological equivalent of the primacy of behaviour over attitude and it is
contrasted with Hegelian philosophy favouring the primacy of ideas.
Sometimes Hegel’s account fits; paradoxically, Marxism itself was a
revolutionary ideology that created change.

In consumer research, there is substantial evidence for the effect of prior

behaviour. Bird and Ehrenberg (1966) found that two-thirds of those who
have used a brand at some time express an intention to buy it. A declining
brand has a long tail of past users, and as a result a larger number of
consumers state that they are going to buy it again, compared with a
growing brand with the same share. There is also evidence that brand
attitudes follow the purchase of groceries. Dall’Olmo Riley et al. (1997)
found that brand attributions (e.g. that ‘Persil washes whiter’) may depend
on recent purchase.
Summary
Key questions for consumer behaviour come from marketing strategy and
consumer policy. In order to answer these questions, we need to understand
how consumers make decisions. When people face difficult and involving
choices, the cognitive model of choice may describe the process of
decision-making, but the process is often simplified, even when the
decision is difficult. When action is steered by the environment, the
reinforcement model provides an explanation of how purchasing is learned:
consumer action is constrained by the opportunities available and directed
by the rewards and costs that are present. Once actions such as brand
purchase are learned, they may be induced by specific stimuli, such as
brand name, and the habit model can apply. To change the behaviour of
consumers, the influencing agent (e.g. advertising, promotions, word of
mouth) must either alter the beliefs and values involved in a complex
decision, or where the context controls behaviour, modify the consumer’s
environment. Learning principles help us to explain some marketing
practices, such as brand extension.

The growth of the Internet suggests that people are now able to make better
choices (more suitable brands, lower prices), but it is not yet clear how
much this occurs.

In this chapter, we also introduced some of the ways in which data are
organized to create meaning: the use of comparison, types of category,
brands and variants, goods and services and market share.
Additional Resources
For an early challenge to comprehensive models of consumer behaviour,
read Olshavsky and Granbois (1979).
Part 2 Consumption Patterns
2 Customer Loyalty
Learning Objectives
When you have completed this chapter, you should be able to:

1. Report the different terms and measures that have been used to describe customer
loyalty.
2. Explain how different ideas about loyalty developed.
3. Explain how customer loyalty is divided between brands in repertoire categories.
4. Describe other habitual features of consumer purchase.
5. Discuss and criticize the main ideas in favour of encouraging retention in consumer
markets.
6. Show how design features of loyalty programmes trigger differences in consumer
behaviour.
7. Report research on the associations between different forms of loyalty.
8. Report on the reasons for defection in services.

(Loyalty schemes, which are really forms of retail promotion, are also covered in Chapter 11.)
Overview
There are three types of loyalty behaviour that consumers can show. First, when they buy
several brands in a category, consumers can give a high share to one of them. Second,
they can continue to buy a brand for a long time; this is retention. Third, they can give
positive advice about a brand and, by this action, recruit new customers. These three
forms of customer loyalty – share, retention and recommendation – ensure a continuing
revenue stream to the brand owner and reduce the need for the parent company to
promote the brand. Marketers therefore want to find and keep customers who exhibit
these forms of loyalty and, where possible, they want to encourage this behaviour.
Marketers are also keen to understand why customers switch away from a brand.

A second aspect to loyalty is the feeling that customers have about brands. We talk of
being satisfied by or liking a brand, being committed to the brand and, in the case of
business and service suppliers, trusting and being dependent upon them.

This subject is quite complicated. We have a common term, loyalty, but it has many
different forms and one form of loyalty may have a strong or weak relationship with
another. Also, the measure of loyalty that we use depends on the category. We use repeat
purchase to show retention in consumer durables and duration as a customer to show
retention for utilities and other services. In some fields, where consumers have a portfolio
of brands they regularly buy, we can use both share and retention to show the loyalty of
customers (e.g. to grocery brands, stores and airlines). To explore these issues, we
approach the subject historically, show how different measures of loyalty originated, and
examine some of the evidence associated with each form of loyalty.
Section 1: Brand Loyalty in Repertoire Categories
The Development of Panel Research
Research on brand loyalty, as a share of purchase, began with a paper by
Copeland (1923) in the first issue of the Harvard Business Review.
Copeland discussed a phenomenon he called brand insistence, which occurs
when a consumer refuses to substitute one brand for another. Copeland was
concerned with repertoire markets like groceries, where consumers often
purchase more than one brand in a category. In these markets, brand
insistence is an extreme form of share loyalty and is now called sole-brand
loyalty.

Initially, research into this field was held back because there were no sound
methods for measuring brand purchases. Retrospective surveys of purchase
may be used but consumers can easily forget some of the purchases that
they have made. To reduce this recall error, Churchill (1942) advocated the
use of panels of consumers, who agreed to make regular reports about their
household purchases. Initially, members were asked to provide weekly
reports of their household purchases, usually by keeping a diary of daily
buys.
 Box 2.1 Methods for measuring purchases
The methods for measuring purchases by panel members have now evolved. One early
form of measurement was the ‘dustbin’ method, where the consumer retained all product
wrappers and agency staff counted purchases from those wrappers. But all wrappers may
not be kept, so this method is also fallible. When bar codes became universal, panel
members were given a bar-code reader and they used this to record their purchases when
they brought their groceries home. But even this is fallible because members may forget
to process purchases. An alternative method was developed by Information Resources
Inc. (IRI) in the USA. They provided the checkout scanners used in the stores of a
number of towns. When panel members used a store in these towns, they showed an
identification card and the store scanner sent data on their purchases directly to IRI for
processing. IRI used this method to link sales to advertising, but as we report in Chapter
13 even this approach may have weaknesses.

The first regular panel was run by a newspaper, the Chicago Tribune.
Brown (1953) used data from this panel and found that brand loyalty in a
household fitted one of four patterns:

Sole-brand loyalty.
Divided brand loyalty (polygamous).
Unstable loyalty (switching between brands).
No brand loyalty (promiscuous).

Brown classified people on the basis of runs of purchase of brands in each

category. Thus, AAAAAA shows sole-brand loyalty, a mix such as
AABABA indicates loyalty divided between brand A and brand B, and
AAABBB might indicate unstable loyalty with a switch from A to B,
though it is not possible to distinguish true switching from divided loyalty
without an extended period of measurement. It is now clear that divided (or
multi-brand) loyalty is the usual pattern of grocery purchase (see Box 2.2).
Box 2.2 Reasons for divided loyalty
Why do people buy more than one brand in a category? There seem to be two sorts of
explanation for having a portfolio of brands – which we call genuine and apparent.
Genuine portfolio
This may occur because:

1. There is little brand awareness and the consumer does not remember previously
bought brands.
2. The category is one where consumers appreciate variety (biscuits, cereal, wine).
3. Customers buy discounted brands, which spreads their range of purchase.
4. The brand that the customer wanted was not available.
Apparent portfolio
1. The panel collects data on household expenditure. Members of a household may
prefer different brands. Individually, they could be 100 per cent loyal, but as a
household, they may show divided loyalty.
2. A household may buy different brands in sub-categories such as biological and
non-biological detergent. The household could be 100 per cent loyal in each sub-
category, but if the two forms of detergent are aggregated in market research data,
the household will not be 100 per cent loyal over all.
Share-of-Category Requirement
Cunningham’s (1956) share-of-purchase approach is now standard and is
illustrated with invented data in Table 2.1. In the table, the last three
numbers of row 1 show that, over one year, Household 1 devotes 50 per
cent of purchases to Brand A, 30 per cent to Brand B and 20 per cent to
Brand C. These percentages are the share-of-category requirement (SCR)
measures that were introduced in Chapter 1. Another measure that is often
used is first-brand loyalty. This is the share given to the most heavily
bought brand (e.g. Household 1 has a first-brand loyalty of 50 per cent). We
see in Table 2.1 how purchase patterns can vary and that some households
buy very little.
 Exercise 2.1 Market share and average SCR
How do average SCRs relate to market share? Are they the same or different? Think about this
before looking below.

The average SCRs per brand are quite close to market share. Purchase frequencies differ across
households and light buyers tend to focus on market leaders (like consumers 6 and 10 in Table
2.1). This means that the average SCR of brand leaders tends to be a little above their market
share: 51 of the 89 purchases in Table 2.1 are for Brand A, which gives it a market share of 57
per cent, slightly below the mean SCR of 64 per cent for Brand A.

Another measure is the average first-brand loyalty in a category. What is the average first-
brand loyalty in Table 2.1? This is 50 + 60 + 67 + 50 + 64 + 100 + 56 + 50 + 80 + 100 divided
by 10, which is nearly 68. Figures of 50–70 per cent for first-brand loyalty are common for
grocery brands.

Customers who buy a brand only once in a period must have an SCR of 100
per cent; when the brand is bought twice, the SCR cannot be less than 50
per cent; and when it is bought three times, the minimum is 33 per cent.
These small-number effects mean that customers who rarely purchase in a
period tend to have higher SCRs than average and, conversely, those who
are sole-brand loyal are often light buyers. When more cases are obtained
by gathering data over a long period, the small-number effect disappears
and then light buyers are found to be somewhat less loyal (Stern and
Hammond, 2004).
Loyalty Proneness and its Correlates
Cunningham (1956) also wanted to know whether the loyalty that a consumer
showed in one category was related to their loyalty in another; he called this
loyalty proneness. In his research, Cunningham found little evidence of loyalty
proneness. Among 21 correlations between share loyalties for individuals
across different categories, the highest was 0.3. East et al. (1995) found
correlations averaging 0.46 between share-loyalty measures across four
grocery categories in a survey. This evidence indicates that it is realistic to
average a consumer’s loyalty scores across a range of categories to obtain a
score for individual loyalty proneness. Using this method, East et al. found that
a customer’s share loyalty to grocery brands was correlated with their store
loyalty (measured as share), total supermarket spending, lack of interest in
discounts and household income.

Note: The last three columns give the SCR loyalty to brands of each household and the means at
the base of these columns give the average SCR per brand; brand A, with a score of 64 per cent, is
more popular than brands B and C.

The association between brand loyalty and store loyalty that East et al. found
has been noted in other studies and a number of explanations have been
offered. One possibility is that loyalty to retailer brands (own label, private
label) explains the effect because the customer who buys more of a particular
retailer brand has to do this by shopping with that retailer. However, Rao
(1969) and East et al. (1995) both found that the correlation persisted after
removing store-brand loyalty, and Flavián, Martínez and Polo (2001) supported
this view with their finding that brand-loyal customers buy fewer private-label
goods. Another explanation is that those who use a wider range of stores (low
store loyalty) have a wider range of brands to choose from and this would tend
to reduce their brand loyalty. A third possibility is that the correlation between
brand loyalty and store loyalty may be explained if these forms of loyalty are
habits, and that some people are more habit-prone. This explanation is
supported by the finding that those with high brand and high store loyalty are
more likely to show another habit by having a routine day for supermarket
shopping (East et al., 2000). Habit proneness could relate to personality or
lifestyle. Habits, by their nature, tend to exclude new experience but they may
save time and effort (see Box 2.3).
Box 2.3 The habits of Gilbert & George
(from the Observer Magazine, 28 January
2007)
The artists Gilbert & George wear the same tailored suits day in, day out, and follow the
same routines 365 days a year. They get up at 6.30am and go round the corner to a café
for breakfast (they do not have a kitchen at home). They then work till 11, when it’s back
to the café for lunch, after which they put in a full afternoon until Paul O’Grady’s show
comes on ITV at 5pm … Dinner is taken in the same Turkish restaurant in Hackney every
night …They are often asked about these routines and complain that no one ever seems to
grasp that they stick to them, not for show, but to save time.
Other Habits
Purchase habits also apply to brands that we routinely do not buy. Most of
us will admit to avoiding certain brands and service providers. Research by
Hunt, Hunt and Hunt (1988) has thrown light on the way consumers hold
grudges against such brands or providers. Hunt et al. find that grudges
persist for a long time and usually begin with an emotionally upsetting
experience as a customer. Grudge-holders may give negative word of mouth
about the offending product when talking to others. Such brand avoidance
could have dire consequences for a manufacturer, but despite this it has
received little systematic study.

Beside brands, there are other product differentiators, and consumers can be
loyal to pack size, price level, country of origin, flavour and formulation
characteristics. For example, Romaniuk and Dawes (2005) found that,
although people bought a variety of different wines, they tended to have a
consistent pattern of preference for price tiers. Singh, Ehrenberg and
Goodhardt (2004) have illustrated regular patterns of purchase with regard
to other category divisions. The point that we emphasize here is that no
emotional commitment is needed for such effects. We argue that most
patterns of purchase, including loyalty, reflect habit rather than deeply felt
commitment.
Section 2: The Rise of Relationship Marketing –
Customer Loyalty As Retention
Relationship marketing (RM) has been described as ‘attracting, maintaining
and enhancing customer relationships’ (Berry, 1983: 25). In a business-to-
business (B2B) context, RM is an industrial philosophy that replaces the
competitive transaction between buyer and seller with a more cooperative
relationship (Grönroos, 1994). In a cooperative relationship, partners learn to
trust each other and reveal more detail about their needs to the other, which
improves mutual support. Relationship marketing has also been applied in
the business-to-consumer (B2C) field, particularly by those concerned with
services. As in B2B, some service relationships (e.g. dentist and patient) can
be characterized by trust and cooperation but this does not apply so well
when the business is large. There is still interdependence between a large
firm and its customers but any initiatives are likely to come from the firm
and to be automated using a customer database. Firms call this customer
relationship management (CRM). Much of the CRM conducted by large
firms is designed to increase sales by exploiting customer purchase habits
and this has little to do with cooperation. Most firms follow good-practice
rules so that their customers can trust them to deliver consistent quality
goods and services, but it does not go much further than that. For their part,
customers can be quite calculating. For example, they may participate in
loyalty schemes because they get a discount on purchases, or gain other
benefits, rather than because they like the firm.

In relationship marketing, there is more emphasis placed on retaining

existing customers than attracting new ones. CRM can help here; for
example, when a sudden lack of spending indicates that a customer has
switched supermarkets, customized vouchers can be issued that may bring
that customer back. Most firms are keen to see increases in satisfaction
among customers because this is thought to retain customers. This emphasis
on retaining customers is based on the idea that it is more expensive to
acquire customers than to retain them. So, instead of losing a customer and
gaining another, it is cheaper not to lose the customer in the first place. A
review by Rosenberg and Czepiel (1984) suggested that the average
company spends six times as much acquiring a customer as keeping a
customer. The ‘six times as much’ rule has now become an item of
marketing folklore; the reality is that the relative cost varies with the
category. For example, supermarket customers are acquired at little cost
whereas credit card customers are expensive to acquire because they must be
checked with credit agencies and offered financial inducements to switch.

The idea that customer retention increases long-term profit was given added
impetus by Reichheld and his associates in a series of papers (Reichheld and
Kenny, 1990; Reichheld and Sasser, 1990; Reichheld, 1993; Jones and
Sasser, 1995; Reichheld, 1996a). These ideas were brought together in a
book by Reichheld (1996b). Reichheld suggests that the value of a customer
grows with the length of time that they remain a customer (called customer
tenure). The reasons for this are illustrated in Figure 2.1. Reichheld argues
that, for each added year of tenure, the profit from a customer rises as the
acquisition cost is amortized, and as the customer spends more (revenue
growth), becomes easier to deal with (cost savings), introduces more new
customers (referrals) and is more tolerant of higher prices (price premium).
Also, and not shown in Figure 2.1, the longer customers stay, the more likely
they are to remain in the following year. This means that those who are
currently long-term customers are likely to give more profit than current
short-term customers in the future. An admirable feature of Reichheld’s
work is that he is very precise about the potential effects of customer
retention so that others can test these claims. Reichheld’s own evidence
tends to be based on case studies. Case studies serve well for teaching about
management practice, but as evidence they are not as valuable as systematic
studies that are set up to test a hypothesis. Case study evidence is often
already available when marketers begin to hypothesize and they may
unintentionally focus on the evidence that fits their theory. Below, we review
Reichheld’s claims.

Figure 2.1 Factors in customer lifetime value (adapted from Reichheld,

1996b)
Customer Tenure and Profitability
We now review five assumptions for how customer loyalty might be turned
into higher profits.

First, do long-term customers spend more? East, Hammond and Gendall

(2006) reported on 17 services where customers were asked how much they
spent and how long they had used the supplier. Examples of services were
supermarkets, credit cards, dry cleaners, fashion stores, mobile phone airtime
and car servicing. Of the 17 studies, only three showed a statistically
significant positive association between tenure and spending: credit cards
(UK), outdoor clothing (USA) and mobile phone airtime (UK). The average
correlation between tenure and spend for the 17 studies was 0.09. This shows
that, usually, there is no substantial association between tenure and spending
that would justify management attention. In a few categories, long-term
customers may spend significantly more than new customers, but such cases
need to be established by research and not assumed by managers.

Second, are long-term customers cheaper to serve? Long-term customers

become familiar with company procedures and need less ‘hand holding’, but
they may also exploit company services more. Dowling and Uncles (1997) first
expressed doubt that long-term customers were cheaper to serve. Later,
Reinartz and Kumar (2000, 2002) found that long-term customers in one firm
made more use of the free services available, thus raising their cost to serve.
They also found that loyalty programme costs increased with tenure. It seems
that total costs do not routinely decline with tenure.

Third, do long-term customers refer more new customers than recently

acquired customers? Long-term customers may value their providers more for
two reasons. First they may learn more about the merits of the supplier’s
offering over time, and second, as those who dislike the supplier switch, the
more appreciative customers remain. Despite these effects, Smith and Higgins
(2000) and Fournier, Dobscha and Mick (1998) have illustrated how
relationships can sometimes sour over time. Also, a brand may be salient when
first acquired but may then become so familiar that consumers give it no
thought and therefore do not talk about it. This loss of salience is more likely
when the category does not change much (e.g. house insurance) and/or is
frequently used (e.g. credit cards). When there is change, for example in the
merchandise of a fashion store, the brand might be recommended repeatedly,
but unchanging products such as motor insurance may lose salience and, as a
result, recommendation rates will fall.

In their review of previous evidence, East et al. (2005a) found either no

association between recommendation rates and tenure (e.g. Kumar, Scheer and
Steenkamp, 1995; Verhoef, Franses and Hoekstra, 2002) or a negative
association (e.g. East, Lomax and Narain, 2001; Wangenheim and Bayón,
2004). In their own research, East et al. (2005a) reported evidence from 23
studies on tenure and recommendation rates (shown in Table 2.2). They found
that the overall association between tenure and recommendation was neutral (–
0.01) but that individual associations ranged from significantly negative to
significantly positive. The significant negative associations were for cheque
accounts, credit cards and car insurance. The significant positive associations
were for car servicing and main fashion stores. These effects may be because
car servicing is infrequent so it takes time for a new customer to be reassured
about the quality of work and change in the merchandise of fashion stores may
support comment. Car servicing was one of the service categories mentioned
by Reichheld, and in this specific case the evidence supports the assertion that
long-tenure customers recommend more. Notice that East et al. studied credit
cards and car servicing twice in the UK (with different samples); the pairs of
studies gave similar results and this makes the work more convincing. Overall,
East et al. (2005a) do not support Reichheld’s claim that long-tenure customers
recommend more than short-tenure customers. The association depends on the
category and may be positive, negative or neutral.
 Note: *Significant at p< 0.05

Figure 2.2 Normal customer survival pattern (adapted from Reichheld, 1996b)
Fourth, are long-term customers more price-tolerant? Price tolerance is
particularly exploited by providers of financial services. For example, firms
may reduce interest rates on investments after an introductory period and rely
on the inertia of customers to minimize switching. In addition, mortgage,
insurance and credit card offers to new customers are often better than those to
existing customers. These tactics may produce short-term profit for the firm but
can irritate customers and lead to them switching. Reichheld makes it clear that
this sort of exploitation of customers is likely to be detrimental to the firm’s
profit in the longer run. In some fields, there may be no long-term price
premium. In three B2C companies that they studied, Reinartz and Kumar
(2002) found that long-tenure customers did not pay more than short-term
customers for the same goods. They also found that long-tenure customers
were more price-sensitive and that these customers expected better value when
compared with recent customers.

Fifth, do defection rates decline with tenure? In general, this is true. Reichheld
finds that a company typically loses about 15 per cent of its current customers
in the first year and 50 per cent over five years. If we follow a cohort of
customers and examine them over a period, we find that fewer and fewer
customers defect each year and that the decay curve levels out, as shown in
Figure 2.2. A study by East and Hammond (1996) estimated defection rates for
a range of groceries and found an average of 15 per cent defection in the first
year. In the second year, defection halved. This means that the customer’s
likelihood of defection declines with tenure. However, there must come a time
when changes in life stage (and even death) mean that customers no longer
need the category and defection may then rise. In addition, there are some
products and services which are only used for a limited period, for example
disposable nappies (diapers) and crèche facilities, and here we would see a
different pattern from that shown in Figure 2.2.
The Strategy of Customer Retention
The evidence summarized above indicates that the benefits of customer
retention in consumer markets have been exaggerated, and that those
benefits differ substantially between categories. One implication of this
evidence is that customer acquisition may bring more advantage, relative to
retention, than is conventionally recognized. We need more evidence of the
relative cost of sales gains through customer acquisition and retention, and
we also need more evidence on how increases in market share come about –
are such increases due primarily to the acquisition or retention of
customers? Two relevant studies are by East and Hogg (1997) and Riebe,
Wright, Stern and Sharp (2014). East and Hogg found that, when Tesco
overtook Sainsburys in 1995 as the leading UK supermarket, the Tesco
gains came equally from increased customer acquisitions and reduced
defection and Sainsburys lost customers because of reduced recruitment,
not because of increased defection. Riebe et al. found that acquisition was
much more important than defection in explaining market share changes for
(i) drug prescriptions among doctors and (ii) the choice of main bank for
consumer finance customers. Sharp (2010), and more recently Romaniuk
and Sharp (2016), also argue that brand gains come mainly from customer
acquisition and present evidence from several studies to support this view.

There are a number of other points that are relevant to the retention versus
acquisition argument. First, it is quite difficult to reduce defection.
Reichheld’s calculations suggest large gains in profit if customer defection
is reduced from 15 per cent to 10 per cent, but a one-third drop in defection
is a substantial amount. Reichheld did give some examples where defection
averaged only a few per cent a year, but in services where there is a specific
location for service delivery, a large part of the defection occurs because of
the relative inaccessibility of the service. For example, East, Lomax and
Narain (2001) found that 43 per cent of the defections from a main
supermarket were because the customer had moved home or because a
more convenient store had been built nearby. This sort of customer loss is
very difficult to counter.
A second point is mentioned by Reichheld but is sometimes forgotten by
those who espouse his arguments. This is that those customers who are
retained by a successful marketing intervention or service improvement are
not necessarily typical of the other customers of the service provider.
Customers who defect are obviously more mobile; when these customers
are retained through a marketing intervention, they may be more likely to
defect later.

A third point is that it is in the nature of loyal customers that they stay put.
They may not need incentives to stay; if this is so, investments in rewards
and product improvements may give little return with this group. Similarly,
it may be very difficult to prise away the loyal customers of competitors.
This leads to a paradox of loyalty. The most loyal customers may have the
highest value but they may not be the best segment for marketing
intervention because of their inertia. So which customers should be targeted
when we have evidence of their loyalty? Should you target your high-share
customers, who cannot increase their share much and also may not be
willing to change their habits? Or medium-share customers, who can
increase their share and may be more changeable? Or low-share customers,
who can increase their share substantially but again may be difficult to
keep? This is a complicated problem which requires category-specific
research.

Although customer retention is emphasized in relationship marketing, this

evidence shows that customer acquisition may be more important.
Reichheld (1996b) does not ignore customer acquisition. He gives the
example of the MBNA credit card organization. This company managed to
acquire high-spending and high-retention customers by the careful design of
the service and by well-chosen targeting. It is also well accepted that not all
customers are profitable. Company costs can exceed returns on small-
spending customers and sometimes retail facilities are so overstretched that
more profit is made when some customers defect.
Exercise 2.2 Brand switching
Consumers switch for many reasons which may vary across categories. Your own experience
may be a guide. If you have switched banks, mobile phone companies, doctors, hairdressers,
supermarkets or alcoholic beverages, why did you do this? Choose two categories in which
you have switched your main brand:

1. List the key reasons for your switch.

2. Identify three things the supplier could have done to try to retain you.
3. Evaluate how effective each of these supplier initiatives would have been in your
particular case.
Loyalty Programmes
Many B2C companies now have loyalty programmes. These programmes
have been introduced for different reasons. Some companies saw a genuine
interest in rewarding their most loyal customers, and business books like
The Loyalty Effect (Reichheld, 1996b) encouraged them in this direction.
Other companies felt forced to imitate their competitors for fear of losing
customers. In some industries, loyalty programmes increased the overall
costs for all players without any real effect on loyalty (a zero-sum game).
Overall, recent academic research suggests that loyalty programmes
generate small effects (Verhoef, 2003; Lewis, 2006; Leenheer, Heerde,
Bijmolt and Smidts, 2007) or no effect (DeWulf, Odekerken-Schroder and
Iacobucci, 2001; Mägi, 2003) on purchase behaviour. It is likely that loyalty
programmes have different effects on different consumer segments; Liu
(2007) found more effect for initially light buyers, while heavy buyers made
little change in their purchasing although they claimed rewards. However,
given that these programmes are in place, how should they be optimized?
Drèze and Nunes (2011) developed a programme of research to examine
how insights from consumer behaviour can help marketers design and
improve their loyalty programmes.

In most loyalty programmes, rewards can be earned repeatedly (e.g.

discount certificates at a retailer, free nights at a hotel chain, cheap tickets
on an airline). Consumers work towards the goal of obtaining the reward
and, once the reward is obtained, have to build up their reward credit again.
Drèze and Nunes (2011) examined what happens to consumption behaviour
after a reward has been obtained. One might expect a post-reward reset and
deceleration in purchases after the consumer attains the reward. Using a
large-scale dataset from a frequent-flier programme, the authors show that,
instead, success is followed by an increase in effort to reach the same goal
again. Interestingly, this effect is only obtained when success requires
perseverance. Successes that come too easily, when small rewards are
frequently obtained, do not have this effect. Creating larger rewards with
greater purchase requirements leads to more overall effort. In a subsequent
experimental study, Drèze and Nunes showed that the consumer’s self-
perceptions of efficacy play an important role and that some form of
learning takes place when a goal is attained which affects action designed to
secure the second and subsequent rewards. Overall, this work shows that
cleverly designed loyalty programmes really can stimulate purchases.

A loyalty programme can also be designed to give consumers additional

incentives to purchase, without additional cost. In a field experiment at a
professional car wash, Nunes and Drèze (2006) randomly distributed two
types of loyalty cards. For the first type, eight car wash purchases were
required for a free car wash. With the second type of card, ten purchases
were necessary but, as part of a special promotion, two free stamps were
given so that the number of required new purchases was also eight. The
respective redemption rates were 19 per cent and 34 per cent. Framing the
task as already begun apparently enhanced the effort to reach the reward
goal, although the distance to the goal was no different between the two
groups. Consumers in that framing condition also accelerated their
purchases: they left 2.9 fewer days between washes.

Another design feature of loyalty programmes is that they can have a

hierarchical structure with different tiers to give the highest spenders a
special status. Companies have to decide on the number of tiers they want
to introduce (a single tier is an option), and in the case of multiple tiers on
the number of customers they want to admit to each. Drèze and Nunes
(2009) show, in a series of experiments, that the desire for tier status can
drive behaviour. Companies face the trade-off of making as many
customers as possible feel special, without disenfranchising the very best
customers by diluting their special status. The authors show that a three-
tiered programme is more satisfying than a programme with two tiers, and
this applies to all customers, not just those qualifying for elite status. The
size of the top tier can be increased without decreasing the status
perceptions of its existing members, as long as a second tier is added.
Adding a third tier enhances the self-perceptions of status of those in the
second tier. (Other aspects of loyalty programmes are considered in Chapter
11.)
Section 3: Combination Definitions of Loyalty
So far we have described loyalty in terms of share, retention,
recommendation and satisfaction. We have not combined these different
forms of loyalty into a more complex definition. Most marketing scientists
use a single behavioural definition, usually share or retention (see East et al.,
2005a). By contrast, most of those who have theorized about loyalty suggest
that loyalty is not behaviour alone, and many feel strongly that attitude
should appear in the definition. For example, Jacoby and Olson (1970)
defined loyalty as the biased (i.e. non-random) behavioural response (i.e.
purchase), expressed over time, by some decision-making unit (e.g.
household, person), with respect to one or more alternative brands, which is
a function of psychological processes (decision-making, evaluation). Oliver
(1999: 34) emphasized the role of feeling as well as behaviour when he
described loyalty as ‘a deeply held commitment to re-buy or re-patronize a
preferred product/service consistently in the future, thereby causing
repetitive same-brand or same brand-set purchasing, despite situational
influences and marketing efforts having the potential to cause switching
behavior’. Also, Day (1969) suggested that ‘true’ or intentional loyalty
occurred when there was a positive attitude to the brand, and he
distinguished this from ‘spurious’ loyalty where purchase of the brand was
not supported by any commitment. Another widely quoted paper by Dick
and Basu (1994) used Day’s distinction and divided customers into four
segments of the typology shown in Figure 2.3. True loyalty occurs in the top
left-hand quadrant of Figure 2.3. Latent loyalty covers those who would like
to buy the brand but have not been able to do so in the past because it was
not available, too expensive, or they had no need for it. Spurious loyalty
occurs when consumers buy the brand but regard it as little better than
others. In this typology, ‘relative attitude’ means that the attitude measure
includes a term such as compared with available alternatives. Then, if a
brand gets a high score, it is because it is rated much higher than the nearest
alternative.

Figure 2.3 Forms of loyalty (adapted from Dick and Basu, 1994)
These definitions of loyalty that combine feeling and action exclude
customers who stay with suppliers because of unthinking habit or simple
convenience. However, marketing is concerned with profits and it is through
customers’ actions that these are made. So should we see loyalty simply as
behaviour or as behaviour with attitude? Ryanair can make profits from
customers who use the airline regularly but may not like it. Indeed, for
brands in some utilitarian categories like bleach and sugar it is difficult to
generate much feeling. But when loyal behaviour is supported by a liking for
the brand, share loyalty and retention may be greater and more profit may be
made, so evidence is required on this. One study that did this was conducted
by East et al. (2005a); they used the Dick and Basu typology and
investigated retention and recommendation in car purchase and
supermarkets. For both supermarkets and cars, they found that greater
relative attitude was associated with more recommendation but had little
relationship with retention. They also found that customers who bought
mostly from one supermarket or had bought the same make of car on the last
two successive occasions (high past patronage) were somewhat more likely
to retain the supermarket or buy the last make of car again at the next
purchase, but that the level of past patronage had no effect on word of
mouth. Figure 2.4 summarizes these relationships with the dotted line
indicating a weak relationship. We conclude that retention and word of
mouth have largely different causes. One simple way of explaining the
relationships in Figure 2.4 is that ‘like correlates with like’. Past patronage
and later retention are alike because they are the same measure at different
points in time. Relative attitude and recommendation are alike because the
reasons that people have for liking a brand are likely to be much the same as
the reasons they give when they recommend it. This evidence indicates that
customer sentiments are important indicators of potential recommendation
and that past practice is a guide to future practice, but the study does not
show that there is an appreciable combination effect, for instance past
practice being a better guide to future practice when the brand is liked.

Figure 2.4 Consequences of different forms of loyalty

 Box 2.4 Checking findings
Research should be confirmed by replication studies to see whether the same results are
obtained when small adjustments are made to the original format. Often, replication
studies fail to confirm the original findings, which is a matter of considerable concern
(Hubbard and Armstrong, 1994). The rather surprising claims of the East et al. (2005a)
research were tested by Praewa Chitpanorak in unpublished research for an MA at
Kingston University. The categories were changed to mobile phones and computers.
Again, attitude was the main predictor of recommendation and there was no significant
evidence that attitude worked in combination with past loyalty to predict retention.

Research studies have generally shown a weak positive association between

satisfaction and retention. Crosby and Stephens (1987) found that life
assurance renewal was slightly greater when customers were satisfied with
the provider. Kordupleski, Rust and Zahoric (1993) found limited evidence
that satisfaction increased retention in company research by AT&T.
Reichheld (1993) reported that between 65 and 85 per cent of customers
who defected were satisfied with their former supplier. Ennew and Binks
(1996) did not find clear evidence of a positive association between
retention and service quality (the latter is usually closely related to
satisfaction), and Hennig-Thurau and Klee (1997) generally found moderate
associations between satisfaction and retention in a review of studies in this
field. Against this weak evidence, we can find a few stronger findings but
these are where dissatisfaction has caused customers to switch. Andreasen
(1985) studied ten patients who reported serious problems with their
medical care and found that six of them switched physicians; Bolton (1998)
found that dissatisfaction with a mobile phone airtime provider led to
switching if the users had recently adopted the service and lacked
knowledge of the supplier’s longer-term performance. Winchester,
Romaniuk and Bogomolova (2008) showed that defection is indicated by
prior negative beliefs about the product. Overall, the research evidence
indicates that satisfaction with a brand or supplier provides a limited
prospect of increased retention but dissatisfaction may be a spur to
switching.

Some readers may find it puzzling that feelings like satisfaction have such a
weak relationship with retention but there are some good reasons why this
should be so. One reason is that usually the measure of satisfaction
employed is not relative. People retain a supplier because of the superiority
of that supplier over others and using a relative satisfaction measure is more
sensitive to this relative difference between alternatives. A second reason
for the weak association between satisfaction and retention can be found in
the reasons for defection. In services, defection may occur as a consequence
of specific failures, price changes or the emergence of superior competition,
as found by Keaveney (1995); such reasons are unlikely to be anticipated
by an earlier measure of satisfaction. A third reason why satisfaction fails to
predict defection well is that the defection may be involuntary, as was found
by East, Lomax and Narain (2001). People who move house will change
their main supermarket if their old one is now inaccessible; this does not
mean that it was unsatisfactory.

What does all this mean? Those who propose combination measures of
loyalty need to show that this approach is useful and that the segments in
Dick and Basu’s typology behave in different ways. In particular, they need
to show that the top-left loyalty segment shows more retention, share
loyalty and recommendation than other segments, at least in some
categories. If they cannot do this, the case for such combination measures
fails. Meanwhile, those who are interested in the prediction of retention and
recommendation need to treat these behaviours separately because they do
not appear to have much common causation.
Section 4: Reasons for Defection
As we have seen, retention is often based on inertia and consumers remain
loyal by continuing to do what they have done before. In some cases, that
inertia can be quite thoughtless as people continue with savings accounts and
utility suppliers, even when alternative suppliers offer much better value. In the
case of consumer durables, where an act of repurchase is required for retention,
there may be more thought, but even here the process may be fairly automatic.
A study by Lapersonne, Laurent and Le Goff (1995) showed that 17 per cent of
the respondents considered only the brand of their current car when
considering a replacement, while Dawes, Mundt and Sharp (2009) found that a
majority of customers considered only one brand when using financial
services.

We have mentioned how existing customers can attract new ones through
positive word of mouth. It turns out that defection is also contagious. Using
data on one million customers of a cellular phone company, Nitzan and Libai
(2011) showed that exposure to a defecting neighbour in their close social
network increased a person’s chances of defection by 80 per cent. Consumers
with more social connections were more affected. But heavy users and
customers with a longer tenure were less affected by their neighbours’
defections.

When people defect, they often have reasons. The reasons for defecting from
service providers were studied by Keaveney (1995) using her postgraduate
students as investigators. They gathered evidence from people outside the
university, asking them to focus on their most recent service defection, report
the service and describe what occurred. The narratives of what happened were
reviewed by judges, who produced a typology of eight reasons for defection
plus an ‘other’ category. Then the frequency of the different reasons was
assessed from the narratives. Some people had more than one reason for
defection, and in Table 2.3 we show the percentages for all the reasons cited
and the percentages when one reason was given. There is not much difference
between these two sets of percentages and we treat the average as typical of
what Keaveney found (column 4).

The method used by Keaveney, called Critical Incident Technique (CIT), is

suitable for establishing the typology of reasons for defection but rather less
appropriate for measuring the frequency of the different reasons since some
memories are more easily recalled because of retrieval bias. This was
described by Taylor (1982: 192), who states that ‘colorful, dynamic, or other
distinctive stimuli disproportionately engage attention and, accordingly,
disproportionately affect judgments’. Events are more changing and distinctive
than conditions and are thus more easily retrieved. In Keaveney’s list of
reasons, the first three (core service failures, failed service encounters and
responses to failed service) are clearly events (50 per cent), whereas
inconvenience and involuntary switching (16 per cent) are likely to relate to
persisting conditions. Keaveney asserts that managers can make changes that
prevent events and can thus stop defection. But if Keaveney’s method raises
the proportion of events reported at the expense of conditions, a strategy of
customer acquisition may be more profitable.

East, Grandcolas, Dall’Olmo Riley and Lomax (2012) used a different method
of measurement. They put Keaveney’s reasons into one item of a questionnaire
and asked survey respondents to state which of these reasons was the most
important in their decision to defect. This method should have reduced retrieval
bias because all the possible reasons were prompted. Keaveney aggregated the
data on all services mentioned by respondents but East et al. reported findings
for specific services. They chose some services that were delivered in a
particular location (e.g. a favourite restaurant) and some services where
delivery was independent of location (e.g. mobile phone airtime). They
reasoned that conditions would be much more important as a reason for
defection when service delivery was located because the inaccessibility of
some locations would cause involuntary switching. East et al. combined data
on the three types of service failure event and the two types of condition and
their aggregate results are shown in Table 2.4, together with Keaveney’s
frequencies for comparison.
Table 2.4 shows that East et al. found a much lower proportion of service
failure events compared with Keaveney (18 per cent instead of 50 per cent) and
a higher proportion of conditions (31 per cent versus 16 per cent). The
conditions came mainly from the located services, as expected. This evidence
suggests that it is difficult to retain customers in located services, and here it
seems better to go for a strategy of customer acquisition. Even when the firm is
not physically located, the scope for retaining customers may have been over-
estimated; Bogomolova and Romaniuk (2009) found that about 60 per cent of
brand defection in business-to-business financial services occurred for reasons
outside the control of managers, and only 4 per cent defected because of
service-related issues. This evidence indicates that attempts to retain customers
may have small effects.
Summary
Several behaviours indicate customer loyalty. These are share-of-category
requirement (in repertoire markets), retention and recommendation.
Feelings may also indicate loyalty. The main feelings are attitude,
satisfaction, commitment and trust.

Early work investigated SCR in grocery purchase and showed that

consumers frequently divided their purchasing across several brands in a
category. Divided loyalty patterns persist over long periods and reflect
consumer habits.

With the rise in relationship marketing, attention turned to the retention of

customers, particularly with regard to services. Reichheld (1996b) argued
that there was substantial profit to be gained if customers could be retained
for longer periods; however, many of his assertions have been undermined
by subsequent research. Loyalty programmes are nevertheless very
common. They can be optimized by thinking carefully about the reward
frequency and status perceptions when there are different tiers of loyalty.

Marketing scientists have generally used behaviour measures such as share

and retention, while others have argued that loyalty should be seen as a
composite of attitude and behaviour and should be tested by segment
comparisons. The best-known composite model is that of Dick and Basu
(1994), however there is evidence that attitude and past behaviour do not
have much combination effect: attitude tends to predict recommendation
and past behaviour predicts future behaviour.

Defection from services has been studied by Keaveney (1995), who

developed a typology of eight reasons for defection. Keaveney found that
many reasons for defection relate to events that managers could control and
thus she thought that much defection could be prevented. However, recent
work suggests that much defection may not be controllable, which turns
attention toward customer acquisition rather than customer retention.
Additional Resources
The articles by Reichheld and his colleagues are clearly written and are
more concise than his book. Reinartz and Kumar (2002) provide a critical
review in this area. Keaveney’s (1995) article is easy to follow.
3 Brand Knowledge, Brand Equity and
Brand Extension
Learning Objectives
When you have completed this chapter, you should be able to:

1. Understand ideas about the mental representation of concepts.

2. Understand what is meant by brand image, brand awareness, brand strength and brand
equity.
3. Explain the potential gains and losses from brand extensions.
4. Understand how brand shares alter when new brands enter a market.
Overview
Brands are represented in the minds of customers as ideas, associations, feelings and
purchase dispositions. Managers search for ways of exploiting these mental
representations of brands by introducing line extensions (new variants in a brand’s
category) and category extensions (products with the same brand name in different
categories). These extensions are not without risk. If the extension fails to capture new
sales, marketing effort will have been wasted. In addition, if the new introduction reduces
allegiance to the brand, all the sales associated with the brand could suffer. We start by
considering how consumers store and process brand knowledge in a network of related
concepts. Then, we consider the way in which this processing gives brands strength and
value, and the way in which brand names are recalled and recognized. Building on this,
we consider the idea of brand equity, that brands are capital that can be exploited by their
owner. Finally, we examine research on brand extension and the way in which new
entrants to a market take sales from existing players.
Section 1: The Mental Representation of Brands
In this chapter, we are first concerned with the way in which concepts such
as companies, brands, services and categories are represented in memory
along with all the other concepts that we have. Psychologists have described
this cognitive world using mental representations theory, which is reviewed
by Smith and Queller (2001). The fundamental idea is that memory is an
associative network of interlinked nodes in which each node is a concept.
The meaning of a concept is therefore given by the interrelationships it has
with other concepts. Applied to brands, the interrelationships define how the
brand is perceived and what developments of the brand are most likely to be
commercially successful. These interrelationships are established in people
through direct experience, from information received from other persons and
from the mass media. In mental representations theory:

Perception activates nodes that correspond to the perceived object.

Thinking occurs when these activated nodes spread activation to other
linked nodes.
Links between nodes are strengthened slowly over time by such
activation.
Those concepts relating to more frequently activated nodes are more
easily retrieved from memory.
Nodes are valenced (positive or negative), thus forming the basis of
attitude.
Long-term memory is the single, large associative structure covering all
concepts, whereas short-term memory is a currently activated subset.
Simultaneous activations may occur (parallel processing), but much of
this processing is unconscious.

To illustrate this thinking, we have set out one possible mental representation
map for iPhone in Figure 3.1. The mental node for iPhone could be linked to
a wide range of concepts and only some are shown. These include product
features such as reliability, ease of use and cost, as well as a wider range of
ideas. iPhones may be associated with Apple stores, with their novel store
layout and merchandising. Another connection is to the Apple Corporation
and its founder, Steve Jobs, who has been credited with a string of original
product developments. Other links might be to the vast range of apps that
can be downloaded, such as the one that allows users to identify birds from
birdsong. Brands are often associated with their competitors; in this case we
have shown Samsung and Nokia.

Figure 3.1 Concepts that might be associated with iPhone

We should be wary of drawing too close a parallel between mental

representations and the evidence from brain research, but there is clearly
some correspondence between the neurons in the brain which are connected
to other neurons by synapses, and the nodes connected by links in mental
representations theory. Brain research shows that consciousness is associated
with the activation of hundreds of thousands of neurons and that each neuron
has as many as 2000 synapses connecting it to other neurons (Greenfield,
1997). This suggests that mental-representations maps indicate only a very
small part of the content and linkages involved when people think.
Brand Knowledge
The nodes and linkages in Figure 3.1 help us to describe two aspects of
brand knowledge – brand image and brand awareness – which are discussed
in the next two sections. Brand image is the set of ideas about the brand, as
illustrated in Figure 3.1. Brand awareness is how these ideas are brought to
mind, which depends on the nature of the linkages.

Brand Image
Gardner and Levy (1955) believed that brands had a social and
psychological meaning as well as a physical nature and that these feelings
and ideas about brands directed consumer choice; this thinking is also
conveyed by terms such as ‘the symbols by which we buy’, ‘brand
personality’ and ‘brand meaning’. These ideas are created in consumers’
minds by their experience with the brand, by word of mouth, and through
mass communications. To the extent that consumers have similar customer
experiences and similar exposure to word of mouth and mass
communication, we expect them to have similar brand images for products
such as iPhone.

Figure 3.2 Main forms of brand knowledge

Applying the theory of mental representations, brand image is given by the
concepts associated with the brand. These associations differ in respect of
valence (positive or negative), number, uniqueness and linkage strength, as
shown in Figure 3.2. Thus, we can break down brand image into a valence
concept, brand attitude, and a cognitive concept, brand strength.

Brand attitude is given by the positive and negative feelings about brand
features. Attributes like unreliability and cost are typically negative, while
ideas such as ease of use and wide functionality are usually positive. In
addition to this attribute basis for brand attitude, there is likely to be a ‘mere
exposure’ effect whereby links become stronger on repeated exposure
(usually usage but also exposure to advertising and word of mouth). This is
explained further in Chapter 8.

Brand strength is important because it relates to the ease of retrieving the

brand name from memory, together with the range of associations and the
resistance to change in thought and feeling. If the number of links affects
strength, we might expect brands in the choice set (the brands considered for
purchase) to have an above-average number of links. Supporting this,
Romaniuk (2003) has shown that brands with more image attributes are
more likely to be considered for choice, and Romaniuk and Gaillard (2007)
have shown that the biggest brands have more attributes. Romaniuk and
Sharp (2003) also found that respondents stated that they were less likely to
defect from brands with a greater number of positive attributes. Thus, more
associations and more positive linkage help to anchor brand use so that it is
difficult to change consumers’ choice sets. Romaniuk (2013) shows that the
bigger market share brands have a larger mental share of associative
networks.This thinking has now been developed into an industry service
measuring distinctive brand assets by the Ehrenberg-Bass Institute of the
University of South Australia.

If brand strength is affected by the uniqueness of nodes, retrieval of the

brand name from memory should be greater when nodes are uncommon and
differentiated from others in the brand image. However, the evidence does
not support this effect of uniqueness. Romaniuk and Gaillard (2007) did not
find that consumers thought that the brand they bought had more unique
attributes than brands they did not buy, and they also found that large and
smaller brands had similar numbers of unique characteristics. Thus, although
we have included uniqueness in Figure 3.2, it is likely that this feature is
overemphasized in brand theory. Customers need to be able to differentiate
brands but this can be accomplished through quite trivial differences (see
Box 3.1).
Box 3.1 Consumer confusion and look-alike
brands
The term ‘consumer confusion’ is used in a variety of ways. In courts of law, it relates to passing
off, when one manufacturer makes a product that is very similar to that of another manufacturer.
In these circumstances, consumers may make mistakes and buy the wrong brand. This has been
a concern in marketing but there is often little confusion on the part of the consumer. For
example, a person buying Perigan’s gin will normally know that they are not buying Gordon’s
gin, despite some similarities of bottle shape, colour and label design. But Gordon’s gin has built
up a buying propensity that is associated with the uniqueness of its pack, and some element of
this buying propensity may attach to look-alike competitors, even when consumers are well
aware of the differences between brands. This means that Perigan’s gin is using brand strength
that it did not create. However, it is not clear that Perigan’s product damages Gordon’s gin – it
could even strengthen this brand because it draws attention to the Gordon’s design. Also, if a
brand is sold at a distinctly different price from another in the same category, there will be
limited direct competition.
The variation in brand strength will relate to the extent to which categories are
promoted, their importance in everyday life and the awareness consumers have
of them which will depend on their salience in the environment. Box 3.2
illustrates two brands with rather different strengths.
Box 3.2 Brand strength
A weak brand
Some years ago, one of the DIY store groups in the UK found that the extra sales raised
by its advertising were much the same as the extra sales they received when its
competitor advertised. It seemed that the public made little differentiation between DIY
stores; advertising probably reminded them of work that they had intended to do and they
got any necessary supplies from the most convenient store. Thus, the advertising of any
one store group had the effect of promoting the DIY retail category as a whole, rather
than the named store group. In situations like this, where brand awareness is weak, store
groups may use sales promotions and advertise these. In this way, consumers only gain
advantage when they patronize the store group offering the promotion.
 A strong brand
In Britain, the case of Hellman’s mayonnaise illustrates the way in which strong brand
awareness can pay off. In 1981, Hellman’s was priced well above other brands of
mayonnaise and was open to fierce competition from those other brands, particularly
retailer brands. The advertising used the term ‘Hellman’s’ rather than ‘mayonnaise’ to
reinforce the brand name. The campaign was successful in making Hellman’s the
effective name of the category and the brand still has a large share of the market.
Becoming synonymous with the category has always been an attractive possibility for a
leading brand. In the UK, ‘Hoovering’ (from the one-time market leader, Hoover) means
vacuum cleaning and is one of the best-known examples of this effect.

Although the notion of strong and weak brands is popular with

practitioners, empirical evidence suggests that brand strength is more about
its size or market share. A highly influential stream of empirical research
challenging the prevalent notion of brand strength was given by Andrew
Ehrenberg and his colleagues. Ehrenberg demonstrated that brand-image
associations were related to past behaviour in a predictable manner (Bird,
Channon and Ehrenberg, 1970). Users of the brand reported a stronger
image association as compared to the non-users. Brands with bigger market
share are more easily recalled because they are bought and used more often.
The pattern was confirmed in several new datasets by Romaniuk,
Bogomolova and Dall’Olmo Riley (2012).

A number of techniques are used for measuring the components of brand

image. Driesener and Romaniuk (2006) reviewed three commonly used
methods: brands can be rated, ranked, or all those with a characteristic can
be named – the ‘pick-any’ method. These techniques can be used for
evaluative criteria such as ‘good value’, or for more descriptive criteria
such as ‘French-made’ or ‘organic’.
Brand Awareness
Brand awareness concerns the way in which ideas are brought to mind.
There are two mechanisms – recall and recognition (Bettman, 1979). When
people recall something, they use the links between concepts in their mental
representation to get to the idea that is recalled. For example, they might
recall an app by first remembering the context in which they heard about it
or used it. Recognition involves a direct match between an external
stimulus and a mental representation; this occurs via the senses, for
example when a person sees an iPhone and knows what it is. When the
stimulus is partial or the mental representation poorly defined, recognition
will be harder to achieve. Human beings have very great powers of
recognition, such as the ability to recognize people from their face, voice
and other cues. Recognition may be a specific-to-general process – for
example, the iPhone is recognized as a smart phone. Recall is usually a
general-to-specific process – for example, the thought of a smart phone
might bring the idea of an iPhone to mind. Often, the cue is the category or
a need for the category. Clearly, a number of alternative brands can be
recalled and it is the business of marketers to make their brand come to
mind more easily than other brands so that it is more often chosen or at least
considered.

The strength of brand recall may be measured as a top-of-mind effect. Top

of mind means the first brand retrieved in response to a category stimulus –
for example, FedEx might be recalled when a courier service is needed.
Ease of recognition may be measured as speed of response when a brand
stimulus is presented. Accuracy of recognition might be estimated by
asking people to report whether a particular brand is present when a picture
of several brands is exposed briefly.

Brands are thought about and chosen in a variety of contexts and these
contexts affect whether recall or recognition is used to retrieve a brand from
memory. For example, a person might want to repair a broken jug and recall
that Loctite will do the job. Alternatively, the person might be in a DIY
store and see Loctite on the shelf, recognize it and then recall that this
product is needed to repair the jug. Brands with a physical form are
generally suitable for recognition because the brand ‘makes itself known’ to
the consumer in the store. Services are often harder to represent in the
environment and here the need for the category often occurs first, and then
the brand is recalled in response. Rossiter and Bellman (2005) use the
distinction between recall and recognition as a cornerstone of their
approach to designing effective advertising and choosing suitable media.
Visual media such as television are good for the recognition of physical
brands because they can display the product, but radio is good for
strengthening the link from category to brand and this medium therefore
aids recall.

The distinction between recall and recognition would have less relevance if
advertising and brand experience facilitated recall and recognition equally.
However, the linkages in the mental representation have direction and a
person who has a brand → category recognition may not have the same
degree of category → brand recall, even though the same nodes are linked.
The measurement of brand awareness should test these links separately.
Section 2: Brand Equity, Brand Extension and
Brand Alliances
Brand Equity
Biel (1991) describes brand equity as the value of a brand beyond the
physical assets associated with its manufacture or provision, and says that it
can be thought of as the additional cash flow obtained by associating the
brand with the underlying product. Because brand knowledge usually
changes slowly, a brand that is currently profitable is likely to continue to be
so (barring unforeseen disasters such as serious PR gaffs by company
employees that devalue the brand or superseding designs).

We should therefore conserve and exploit brands in the same way that we do
with other assets. Some marketers have taken a financial approach and tried
to value the additional profit potential offered by a brand. Indeed, some
companies now estimate the value of their brands on their balance sheets –
and certainly brand equity is of crucial concern in company mergers and
take-overs. A second customer-based research approach has focused on
consumer responses to the brand as measured by image, awareness, quality,
loyalty and specific market advantages (e.g. Aaker, 1991).1 These are the
precursors to financial benefit; without these consumer responses to the
brand, there would be no extra revenue stream or scope for brand extension.
In consumer behaviour, we are primarily interested in this second approach
and Figure 3.3 shows brand equity as the outcome of brand attitude, brand
strength and context. Context covers such matters as market size and
category differences such as the prominence of the category in everyday life.

Figure 3.3 Determinants of brand equity

Because of the difficulty of measuring these different components of brand
equity, and uncertainties about how they should be aggregated, customer
judgements of quality are often used as a proxy for brand equity. For
example, using brands of high and low quality, Krishnan (1996) compared
the number of associations, valence and uniqueness and how the associations
were formed (whether by experience or indirectly through communication).
Krishnan’s results were generally in the direction expected but there is a
problem about using quality as a measure of brand equity; few people doubt
the quality of Porsche but most people lack the means to buy such cars,
which reduces Porsche’s brand equity (hence the introduction of context in
Figure 3.3). Kamakura and Russell (1991) and Keller (1993) have proposed
an alternative measurement of brand equity. Instead of trying to measure
aspects of brand knowledge, these researchers suggest that brand equity can
be measured by comparing consumer responses to changes in the product
specification, price, promotion and distribution of a named brand, with the
corresponding responses for an unnamed or unfamiliar product. The
differences indicate the benefit conferred by the branding.
 Exercise 3.1 The scope for brand extension
Put in order of brand quality the following brand names: Samsung, Toshiba, Sony, Bush.
What does this suggest about the profit potential of the four brands?
What, in particular, do you associate with Sony?

In western countries, Sony usually tops the list and Samsung does less well
(but not in South Korea where it is a national premium brand). Sony’s
ascendancy probably reflects product quality, innovation and advertising
over many years. However, Sony is particularly focused on electronic goods
and the assurance offered by the brand name might be much weaker when
applied to a field outside electronics such as kitchen utensils. This
introduces the idea that the extension should fit the parent brand strengths.
New Brand or Brand Extension?
Brand equity will affect the success of line, category and geographical
extensions. A line extension is a variant within the category of a parent
brand (e.g. a new pack size). One form of line extension, known as a
vertical extension, introduces new lines at different price points (e.g. a
premium version and a basic version). A category extension occurs when a
brand name used in one category is applied in another category (e.g. when
Stella Artois, known for lager, introduced a cider). Some extensions are
more of a leap (e.g. when Amazon moved from being solely a retailer to
being a manufacturer, when it introduced the Kindle for reading electronic
books).

A geographical extension occurs when a brand that is marketed in some

countries is introduced to other countries where it may already be well
known from films or other popular media; an example is the introduction of
US car brands such as Chrysler to Europe. In addition to extensions,
companies with a strong brand may gain by taking over other companies
and extending their name to the acquired company’s products. For example,
a well-established hotel brand, like Marriott, may be able to make more
profit out of another hotel company’s physical assets because of the strength
of the Marriott name. Similarly, co-branding combinations can work well,
such as Intel and Dell. Alternatively, a firm with a weak brand presence will
come to an arrangement with a firm with a strong brand – for example,
Caterpillar boots are made by Wolverine World Wide under licence.
How to use the brand and the sub-brands is a difficult decision for car
manufacturers (see Box 3.3).
 Box 3.3 The model name dilemma
The launch of a new car raises branding problems. Should the old model (or sub-brand)
name be abandoned or kept? Some equity attaches to a model name which is lost when
the name is dropped. On the other hand, it is important to show the novelty of the new
model, and a break with the past helps this. Some companies, such as Volkswagen, have a
policy of retaining model names such as Golf and Passat, whereas other manufacturers,
such as Peugeot, usually drop the model name (though when faced with the continuing
popularity of the 205, they retained this name for a while).

Nesting a new model name with the old one is not normally done in the car industry, but
it had to happen in Australia when Daihatsu found that its rather bizarre model name,
Charade, was better known than the parent brand, Daihatsu. When the time came to end
the life of the Daihatsu Charade, the company introduced the new model as the Daihatsu
Charade Centro in order to benefit from the positive brand image that the Charade had
gained amongst consumers (Sydney Morning Herald, 7 July 1995).

Line extensions are very common. In the USA, Aaker (1991) estimated that
about 90 per cent of new products in the packaged goods industry were line
extensions, though there was some cutback in line extension when the
Efficient Consumer Response movement got underway and reduced
wasteful marketing activity (Buzzell, Quelch and Salmon, 1991; Kahn and
McAlister, 1997). In 2004, Les Échos reported that new launches divided
into 18 per cent new brands, 17 per cent category extensions and 65 per
cent line extensions. Generally, the description of a product as a line
extension is appropriate when the variant can compete with its parent. For
example, those who buy fun-size Mars bars will usually do so instead of
buying the normal size Mars bar. Sometimes a line extension will raise
additional brand sales, but often it is defensive and designed to counter
competition and prevent sales erosion when new variants are introduced by
competitors. Normally, category extensions do not compete with sales of
the parent brand. For example, Porsche sunglasses do not compete with the
sale of Porsche cars and Caterpillar boots will not compete with the sale of
the company’s earth-moving equipment. Exceptions to this rule occur in
food and drink categories – for example, it is possible that instead of buying
a Mars bar, a customer may buy a Mars ice-cream or a Mars mini roll.

Although there may be substantial value attaching to a brand name, it is not

easy to decide whether to introduce a product as a new brand or to extend
an existing brand name. Sometimes, there will be an incompatibility
between categories that makes it unrealistic to extend a name. Procter &
Gamble might see problems in extending the Pampers brand (disposable
nappies/diapers and baby wipes) to baby food though they might be able to
use the name on baby clothing. The Toyota brand was deemed unsuitable
for the launch of a new luxury car because of its mass-market positioning.
As a result, Toyota developed Lexus. By contrast, high quality brands such
as BMW and Mercedes have chosen to offer smaller cars and utility
vehicles under the parent brand name. This underscores a general
observation that high quality brands can be extended to lesser quality
products more easily than lower quality brands can be extended to higher
quality products.

If a new name is used, managers should check that it is (1) different from
other brand names, (2) easy to remember, (3) translates well (the Vauxhall
Nova was unsuitable in Spain because it implied that the car would not go),
(4) available (many brand names are registered but unused by other
manufacturers), and most importantly, (5) extendable itself, since
extensions of a new brand can make further profits and help to justify
creating the new brand name. One reason for the loss of interest in
descriptive names such as ‘I Can’t Believe It’s Not Butter’ is that such
names offer limited scope for extension.

New brand names are expensive. McWilliam (1993) found that cost saving
was the most frequent reason cited by marketing practitioners for using an
extension. Smith and Park (1992) studied the effect of extensions on market
share and advertising efficiency, and concluded that extensions capture
greater market share and can be advertised more efficiently than new
brands. Doyle (1989) also found that extensions need less advertising and
noted that they are more readily accepted by distributors as well as by
customers. Tauber (1988) found that an existing name helps a brand to gain
shelf space in stores.

Smith and Park (1992) did not find that the efficiency of a new extension
was reduced by the number of extensions already made. Similarly, Dacin
and Smith (1993) found that consumer confidence in a new extension was
unaffected by the number of existing extensions, provided that the new
entrant was compatible with its predecessors. Dawar and Anderson (1993)
found that new lines were more acceptable if they were introduced in an
order that made them coherent with the products that had already been
introduced.

However, failure with a brand extension may damage the parent brand. This
may occur because the extension so enlarges the associations of the brand
name that it loses impact and all products under that name suffer in
consequence (Tauber, 1981). There may also be a negative effect on the
brand in downward vertical extension. Heath, DelVecchio and McCarthy
(2011) examined the effect on the parent brand of vertical line extension up
and down the quality range. They found that lower quality extensions
tended to reduce the parent brand’s rating a little and higher quality brands
tended to improve the rating quite substantially so that the effect was
asymmetric. Dall’Olmo Riley, Pina and Bravo (2015) found that the scale
of the negative effect depended on the product and that prestige brands were
more sensitive than luxury brands. However, they also found that the effect
was reduced when the downward extension had a much lower price.

Aaker and Keller (1990) found that potentially negative associations could
be neutralized by focusing on the attributes of the new brand rather than the
strengths of the parent brand. In a subsequent study, Keller and Aaker
(1992) examined how consumers saw the extensions to a brand when there
were, and were not, prior extensions. They found that, if the prior
extensions were regarded as successes, they improved the evaluation of a
new extension; if the prior extensions were unsuccessful, they diminished
the evaluation of a new extension. This suggests that brand owners should
be wary of extensions after a failure.
Consumer Acceptance of a Category Extension
Aaker and Keller (1990) took six well-known brand names and examined
how consumers reacted to 20 hypothetical category extensions. For example,
they suggested the idea of Crest toothpaste extending into chewing gum and
also Vidal Sassoon offering perfume. They found that three factors were
related to the attitude of consumers to the potential extension. One of these
was the fit between the categories of parent and offspring. A second was the
quality of the parent brand. The third factor was whether the extension was
seen as difficult to make by the owner of the parent brand. These
relationships are shown in Figure 3.4.

Fit was positively associated with acceptability. It was measured as

complementarity, substitutability and transferability. Complementarity
concerns the matching of the new product with the parent expertise, thus
skiing goggles would be a complementary product for a maker of skis.
Substitutability applies when the new product could be used instead of the
parent product – for example, snow boards would substitute for skis.
Transferability relates to manufacture rather than usage – whether or not the
producer of the parent product is believed to have the capacity to produce
the offspring product. For example, a firm that made skis would not
necessarily be seen as the best organization to manufacture snow-making
equipment but would be acceptable for ice skates. This account of fit is neat
but it may be inadequate when faced with brands like Chanel which embrace
a wide variety of product types (e.g. fragrances, jewellery, clothing, skincare
products). Although Chanel products share a quality image, they differ in
other respects.

Figure 3.4 Factors contributing to extension success (adapted from Aaker

and Keller, 1990)
The second factor, the perceived quality of the parent brand, was positively
associated with the acceptability of the extension according to Aaker and
Keller, but only when there was a good fit. The third factor, the degree of
difficulty perceived about making the extension, was positively associated
when the difficulty was moderate, while extensions that were very easy to
make and very difficult to make were less acceptable. This may be explained
in a number of ways. Aaker and Keller suggest that consumers might think
that an easily made extension is overpriced. An alternative explanation
relates to the cognitive effort involved. An easily made extension may take
less effort to comprehend and may therefore not disturb a person’s mental
representation. When people have to do cognitive work on an idea, they may
connect it with existing concepts, thus raising awareness of the extension.
However, if the cognitive work required is too complicated, the extension
may be rejected. Thus, an extension requiring moderate cognitive effort may
be most acceptable. Related to this explanation, Hartman, Price and Duncan
(1990) suggest that people will try to make sense of an extension, but if it
differs too much from the parent product they may dismiss the idea of the
new product. Supporting Hartman et al., Meyers-Levy and Tybout (1989)
showed that one unusual characteristic increased cognitive processing, while
too many unusual characteristics reduced it. In sum, the very obvious may be
ignored and the inexplicable may be rejected.

Sunde and Brodie (1993) failed to replicate Aaker and Keller’s (1990)
findings. Following this, Bottomley and Holden (2001) reviewed the
evidence on the acceptability of brand extension. They used data from the
original Aaker and Keller (1990) study and from seven replications. They
found that the original contentions of Aaker and Keller were broadly
supported: fit makes an extension more acceptable and the quality of the
parent brands increases acceptability provided that there is some fit.
However, we should note that Aaker and Keller’s method is quite weak; it
rests on the judgement of respondents about how they will behave in
hypothetical circumstances and this judgement can be mistaken.

Batra, Lenk and Wedel (2010) suggest that the success of an extension
depends on the atypicality of the parent brand as well as the fit of the
extension. By atypicality, they mean having abstract (rather than concrete)
associations since these can link to a wider range of categories. They suggest
that a beer brand such as Corona, which has ‘lifestyle’ associations, is
atypical compared to Heineken which is sold more on its performance as a
beer. Batra et al. proposed a personality-measuring procedure covering the
brand and the category that permits assessment of atypicality. Their initial
results are promising but this work still rests on perceptions of consumers
rather than their purchasing, which is the ultimate test. The circumstances
governing the success of an extension are also covered by Keller and
Lehmann (2006) in a wide-ranging review of the field.
Effective Marketing
Völckner and Sattler (2006) investigated ten factors that might predict
extension success and found that fit was the most important; however, they
also found that marketing support and retailer acceptance were needed as
well. This suggests that marketing effectiveness may play a large part in
brand extension success. Some poorly fitting extensions have worked
because the appropriate marketing structures were available. Marks and
Spencer, originally known for clothing, successfully diversified into food
because it had an effective system of sourcing and distribution; Bic, known
for disposable ballpoints and lighters, succeeded with a sailboard extension.
Neither of these extensions was an obvious fit. McWilliam (1993) points
out that marketers are very reluctant to see a failure as the result of poor
marketing, but this is often the reason. The potential of a category extension
is affected by market size, product quality, market growth, economies of
scale, distribution structures and profit margin, all of which should
influence the marketing strategy. From this standpoint, some incongruous
extensions may succeed because they are well marketed.
So, can we predict extension success?
It is clearly important that we understand how brands are accepted or
rejected and therefore what scope there is for extracting more profit from a
brand. However, the potential of a brand is difficult to measure. First, it
seems likely that the unconscious cognitive activity involved in brand
choice is large, and because it is unconscious it is difficult to represent and
measure this activity. Another problem is that brand choice is contextual. It
occurs under a variety of circumstances and these different circumstances
will relate to different parts of a consumer’s mental representation map.
This problem is aggravated when we take account of the way mental
representations differ between people. The theory of mental representations
may serve as a description of why certain effects are found, but it has
limited value as a predictive model.

We are also rather sceptical about the method used to determine the
acceptability of potential extensions such as that by Aaker and Keller
(1990). We do not dispute the findings of such work, now checked by
Bottomley and Holden (2001). What we dispute is the generalizability of
these findings to everyday life. Under research conditions people will report
on perceived quality and fit, but in the field marketing activity and
consumer adaptability may overcome a lack of fit. In addition, until the
work of Sattler et al. (2010), research in this area focused on the
acceptability of the extension without taking account of price and therefore
profitability.

Despite these problems, there are some areas of promise. It is clear that
brands do have value in the sense that people associate more benefits with
some well-known brands and may pay more for a branded product than a
functionally equivalent anonymous product. Tests based on such
comparisons are likely to indicate brand strength and extension potential.

Brand Alliance or Co-Branding

A more recent and popular branding strategy is the partnership between two
brands, known as a brand alliance or co-branding. Traditional brand
extension strategies involve one brand, while co-branding includes two or
more brands with both benefiting from each other’s strengths. A well-
known example is the partnership between Dell and Intel. In this field, we
distinguish between the fit or complementarity of the partnering brands
(brand fit) and that of their product categories (product fit). Positive
attitudes towards both types of fit are thought to enhance consumer
evaluation of the co-branded product, as well as leading to a strengthening
of favourable attitudes towards the partners, known as spillover effect. A
close fit between the positioning strategies of the partner brands also leads
to positive consumer evaluations of the co-brand, as shown by Singh,
Kalafatis and Ledden (2014). In an experimental study, Singh (2016) shows
the role of CSR-activity-based fit on consumer evaluation of the co-brand.
In industrial markets, Kalafatis, Remizova, Riley and Singh (2012) show
that a brand that partners with another brand with higher brand equity
benefits from the relationship. An experimental study by Kalafatis, Riley
and Singh (2014) demonstrates that the managerial evaluation process is
similar to that of consumers. Applying co-branding to the higher education
sector, Kalafatis, Ledden, Riley and Singh (2016) show that name-order
effects explain the higher perceived value of a dual degree between high-
and low-ranked universities compared to a dual degree between low- and
high-ranked universities. The research interest in co-branding has coincided
with its emergence as a promising strategy, and it is preferred by several
brand owners over brand extension.

However, co-branding is not without risks. Alliances may fail because of

the unique risks inherent in co-branding, such as differences in strategic
visions, legal and financial disagreements or incompatible brand synergies
(Blackett et al., 1999). The benefits of co-brand alliances are also tempered
by undesirable events that cause a partner brand to receive negative
publicity, and consequential negative evaluation. Ford received negative
media attention and was boycotted in the wake of Firestone’s tyre scandal.
Celebrity endorsement is a form of co-branding and this can go wrong. For
example, Accenture and Nike were affected by the controversy surrounding
Tiger Woods, and the US National Football League (NFL) received some
backlash from supporters as a result of the Ray Rice domestic abuse
controversy.
Exercise 3.2 Potential extensions
Pampers, Shell and the BBC are brands that have few extensions. Suggest extensions which
might be appropriate and successful for these brands. Give reasons for your suggestions. Why
have these brands not been extended into these areas before?
Section 3: Sales Losses by The Parent of A Line
Extension
Sometimes, a new entrant enlarges a market so that all the brands within it
gain. One example was when Pampers entered the disposable nappy (diapers)
market in South Africa (Broadbent, 2000). Its campaign helped the category to
become established as an alternative to cloth diapers and this helped Pampers’
competitors to sell more volume. But this is unusual. Normally, markets have
fairly static aggregate sales over the medium term so that the sales gained by a
new brand are at the expense of another brand’s share. When this applies, it is
important to know how a new entrant takes sales from the existing players
since this helps to decide a manager’s strategy. This sales effect is often large
when a line extension is launched, because existing lines can easily be
substituted by the new line: the new line cannibalizes the existing lines.
Sometimes, cannibalization is enforced by retailers who refuse to give
manufacturers additional shelf space for a new line and stipulate that some
other line in their portfolio must be abandoned in order to provide space. ‘Line
in, line out’ is the terse name for this retailer practice. When the new line can
compete with the previously established brands, there are two sales patterns
that can occur:

1. Existing brands may lose sales to the new entrant in proportion to their
market share. This is the basic effect that may be expected when there are
no special affinities between brands.
2. An extra loss of sales occurs among brands that are perceived to be
similar to the new entrant. With a line extension, the main similarity is
likely to be the common brand name, with the result that the parent loses
more sales than would be expected from market share alone. Extra losses
may also be incurred when there are similarities of formulation,
packaging, pricing, positioning, targeting, distribution and physical
proximity to other brands in the store.2

Some data on the British detergent market illustrate the way in which
consumers shift support from existing brands to a new brand. Table 3.1 shows
how, 20 weeks after launch, the first liquid detergent on the British market,
Wisk, was taking customers from other brands (right-hand column). Wisk was
a new brand and took share from other brands roughly in relation to their
market share (the correlation is 0.85), though Surf, with its value-for-money
positioning, seemed to resist loss better. Because the sales loss is proportional
to existing share, big brands lose more volume than small brands.

This effect was further investigated in a study by Lomax et al. (1996). Using
new data, this work confirmed that Wisk took share in relation to the market
share of the other brands. A second study by Lomax et al. showed the
cannibalization effect of the same brand name when a concentrated version of a
German detergent took disproportionately more sales from its parent than from
other brands. In a third study, Lomax et al. examined the gains of Ariel liquid
detergent that followed Wisk onto the British market. Here, there was a parent
powder brand and it was anticipated that this would be cannibalized by the new
liquid formulation, but this did not occur. Instead, Ariel liquid gained sales at
the expense of the whole powder section. Though unexpected, this finding was
consistent with a notion of cannibalization barriers introduced by Buday
(1989). Consumers apparently saw the new product in relation to its
formulation rather than its branding. In the language of mental representations,
there may be limited linkage between versions of the brand across
formulations. This seems to be an advantage but there is a danger that, when
this occurs, the new entrant will gain less benefit from the common brand
name. A fourth study examined the sales of Persil liquid, which was launched
after Ariel liquid; again, the impact on the parent powder was not
disproportionate.

This work shows that excess cannibalization of the parent brand is not
inevitable and is reduced by pricing, targeting and positioning. More generally,
it seems likely that it is reduced by pricing, targeting and positioning the brand
so that it is more similar to competitor brands and less similar to the
manufacturer’s existing brands.

A recent study by Singh, Scriven et al. (2012) examined the market

performance of 47 new extensions in a large-scale UK consumer panel. It
established that the successful new brands behave much like the established
ones within the first two quarters of their launch; the failing ones show declines
in both the number of customers and the repeat purchase rates from the third
quarter of their launch. This study gives a useful diagnostic framework for
assessing the performance of new brand extensions.
Summary
Memory can be represented as a network of interlinked nodes. Thinking
involves the activation of parts of this network. The whole network
represents long-term memory. Short-term memory is the currently activated
section of the network. Within this system, brand knowledge has two
aspects: brand image – the range of brand associations – and brand
awareness – the retrieval of the brand via the associations using recall or
recognition.

Brand equity is the value added to the basic product by branding. It is

thought that brand equity is greater when the attitude to the brand is more
positive, when the range of associations is large and strongly linked, when
awareness is high and where the brand is prominent, well marketed and has
a large market share. In practice, there is no coherent way of measuring and
aggregating all these factors, so either perceived quality is used as a proxy
for brand equity or it is measured by showing the difference in the value
created when branded and unbranded products are promoted.

Marketers want to know whether a category extension will succeed. This

has been approached by examining consumer judgements about different
extension propositions. Acceptance is greater when the proposed extension
fits the parent because it complements or substitutes for the parent product,
and when the parent is seen as suited to producing the offspring product.
Acceptance of a fitting product is increased when the parent is seen to be of
high quality. However, this approach has been criticized for ignoring
marketing expertise in the launch of an extension. Some fitting extensions
have failed and some non-fitting ones have succeeded, and it is likely that
success or failure owes a great deal to effective marketing.

When line extensions are launched, they often take a large proportion of
their sales from the parent. Sometimes, this is accepted as part of the
evolution of the product, but it is attractive to get extra sales from a line
extension. There has therefore been some interest in how a new entrant to a
market draws sales from existing players. Sometimes, the sales losses of the
parent are modest because consumers use formulation, price or another
factor to distinguish the new line from the parent.
Additional Resources
Branding research is becoming specialized according to the context, such as
political branding, Islamic branding, digital branding, sensory branding and
arts branding. For a comprehensive treatment, read Dall’Olmo Riley, Singh
and Blankson’s (2016) edited book of chapters on the latest aspects of
branding, entitled The Routledge Companion to Contemporary Brand
Management.
Notes
1 A well-known method of valuing brands in this way comes from the
Interbrand Group. Seven factors are considered: leadership, stability, market
stability, internationalization, trend, support and protection. Simon and
Sullivan (1993) describe a method that compares branded and unbranded
cash flows. Others have used the stock market to indicate the value of
brands by subtracting the value of fixed assets from the market valuation.

2 These affinity effects are also seen when consumers buy more than one
brand in a category. When they do this, their selection of brands may be
linked by a common characteristic such as brand name or product
formulation. This is discussed in Chapter 4.
4 Stationary Markets
Learning Objectives
When you have completed this chapter, you should be able to:

1. Describe the typical patterns of purchase found in mature, stationary markets.

2. Explain the role of stationary market models in evaluating brand performance.
3. Discuss the importance to sales of both light and heavy buyers.
4. Explain how market regularities set limits to marketing objectives.
Overview
Research on market patterns is done by analysis of the data provided by market research
companies. In particular, academic researchers use the findings of consumer panel
studies, described in Chapter 2, which are conducted by companies such as IRI, Taylor
Nelson Sofres (now in WPP’s Kantar group), GfK and Nielsen. Consumer panels record
the purchases of many hundreds or thousands of individual households for several years.
From such records we can see that most mature consumer markets are approximately
stationary (i.e. brand sales change little from year to year).

In order to judge how a consumer brand is performing in a stationary market, we need to

know the patterns of purchase that are commonly found in such markets. Then we can see
whether a brand is behaving in a normal manner, or whether there are exceptional aspects
to its sales. Marketing scientists have established elegant mathematical models that are
very effective at mimicking the patterns found in stationary markets that are revealed by
panel data. These models have been so successful that they now provide us with sales
norms that can be used to assess the performance of a brand. When brand performance
differs from the stationary market prediction, we can investigate why this is so. This work
applies primarily to frequently purchased repertoire markets (where consumers often buy
more than one brand in the category).
Section 1: Modelling Mature Markets
The Stability of Mature Markets
In this chapter, we are mainly concerned with established consumer markets
rather than markets for new goods or services or business markets.
Established, or mature, markets cover the majority of our purchases. An
important feature of these markets is that they usually do not change much
(in terms of the market shares of the major brands) and are therefore
described as near-stationary. Changes do sometimes occur in mature
markets: whole sub-markets may decline – for instance, the 1980s saw a
decline in the consumption of bitter beer and a corresponding rise in lager
drinking in the UK. Normally, such changes occur quite slowly over a
period of years. Only exceptionally do we see rapid changes that may then
become permanent for specific brands or for the whole category. Such
changes may occur when adverse publicity about a product damages its
reputation, or when an advertising campaign is particularly successful (e.g.
Stella Artois gained substantial share in the British lager market in the mid-
1980s following a very successful advertising campaign). Markets may also
change over short periods because of sales promotions, but usually these
gains are not maintained when the promotion finishes (Ehrenberg and
England, 1990; Ehrenberg, Hammond and Goodhardt, 1994a). Because
promotions run for short periods, the gains they produce have little effect
when averaged over several months, and often promotional gains are
counter-balanced by losses when competitors run promotions. As a result,
the market looks quite stable over a period of several months or a year.

One reason for the relative stability of markets has been explained in
Chapters 1 and 2. Individuals form habits of purchase that limit change. In
Chapter 2, we noted that a typical brand loses about 15 per cent of its
customers over a year, and that those customer losses are usually offset by
customer gains.
The Value of Mathematical Models
If a market does not change, brand performance measures, such as repeat
purchase, the relative number of heavy and light buyers and the pattern of
cross-brand buying in a category, will be much the same each time they are
measured for a period of the same length. An effective mathematical model
will let us predict these brand performance measures from other simple
brand statistics. If a model is routinely effective, it acquires diagnostic
value. When the observed brand performance does not fit predictions from
the model, we need to find out why this is so and we may have to adjust our
marketing support for the brand. The model used to predict the purchase
patterns for a single brand is the negative binomial distribution (known as
the NBD), while a more complex model used for predicting purchasing and
cross-buying for competing brands is the Dirichlet (pronounced: Dir-eesh-
lay).

Ehrenberg and his colleagues established that how often people buy a
product, and which brands or products they buy, is largely habitual, with
individual behaviour aggregating to measures of brand performance which
follows regular law-like patterns (e.g. Ehrenberg, Uncles and Goodhardt,
2004). This approach is based on the NBD-Dirichlet model of purchase
incidence and brand choice in established, competitive markets.1 The
empirical finding that most markets behave in a predictable ‘Dirichlet’
manner, leads to two main conclusions:

1. Loyalty (the propensity to purchase a brand again) at the individual

consumer level has multiple causes. However, it produces a common
effect at the brand level, which is captured by many different
measures.
2. Competing brands differ little in the levels of loyalty they enjoy.

The Dirichlet model is used to predict a range of brand output measures,

including several related to loyalty. With input information on the
penetration and purchase frequency of the total product field and a specified
length for the analysis period, the Dirichlet model can predict brand
performance measures such as the incidence of heavy and light buyers, the
share of category requirement and the percentage of exclusive buyers for a
brand.

Early research on brand modelling by Ehrenberg (e.g. papers in 1959, 1969)

was later brought together in Repeat Buying: Theory and Applications
(Ehrenberg, 1988, first published in 1972). This book challenged
conventional beliefs in marketing and caused a reappraisal of some of the
traditional ideas about brand loyalty, brand positioning, the effects of
advertising and the way in which sales grow. It was followed by work in the
USA investigating the mathematical properties of stationary market models.
Morrison and Schmittlein (1981, 1988), for example, gave detailed
attention to the precision of models and the modifications that might
improve this precision.

Mathematical models can also be applied to other forms of stable repetitive

behaviour. Goodhardt, Ehrenberg and Collins (1975, updated 1987) used
such a model to study television audiences. Another application has been to
store choice, with store groups being treated as brands (Kau and Ehrenberg,
1984; Wright, Sharp and Sharp, 1998). It is also possible to model other
category divisions such as a pack size (Singh, Ehrenberg and Goodhardt,
2008). Models can be used to predict the performance of brand
aggregations (e.g. all private label brands in a category, or a combination of
many small brands in an ‘other brands’ grouping).

Empirical research on stationary markets describes how consumers buy

mutually competing brands. Stationary market research, however, does not
explain why some people buy more than others and one brand rather than
another. Some critics argue that the lack of attention to such motivational
issues limits the application of these models, particularly when the marketer
is trying to induce change. What do you put in advertisements if consumer
motivations are unknown? Do those who buy more have different reasons
from those who buy less? Why do people avoid some brands? When
markets do expand or contract, these changes may reflect changes in
motivation, income or other household circumstances. But theorists such as
Ehrenberg do not claim to cover all the problems that arise in marketing and
specifically exclude motivation. What they do describe is the quantitative
form of stable markets, which is the nature of most markets, and the
numerical predictions from the model are usually very close to the
observations derived from panel data. On the other hand, when the market
is not stationary (e.g. a new category), the difference between the observed
facts and the model prediction is often instructive and may help us to
understand the way the market is changing.
Definitions
Before we examine the patterns of purchase found in mature markets, the
reader should be clear about the meaning of a number of terms. First, we
usually work with purchase occasions rather than sales. On a purchase
occasion, a buyer buys one or more units of a brand. In most markets,
consumers buy one unit at a time so that purchase occasions are
approximately equal to sales. Other important definitions are:
Market share = Total purchases of the brand Total purchases of the category

Penetration = The number buying the brand at least once The total number
of potential customers

Penetration: denoted by b, is the proportion of all potential buyers (in

the population we are studying) who buy a brand at least once in a
period. (Think: b for buyers.)
Purchase frequency: denoted by w, is the average number of purchases
made by those who purchase at least once in a period. (Think: w for
purchase weight.)
Mean population purchase frequency: denoted by m, is the number of
purchase occasions in the period made by an average member of the
population. (Think: m for mean.) When b is expressed as a percentage,
m will be the sales per hundred of the population.

The variables are linked by the sales equation:

m = bw

Thus, when the penetration of Persil over three months is 0.25 and the
purchase frequency is 4, m = 0.25 × 4 or 1. (In words: when a quarter of the
population buys Persil, on average four times, then the average purchase
occasion rate in the whole population is one.)

When people buy more than one unit per purchase occasion, we multiply by
a correcting factor to get the sales rate. For example, if people buy, on
average, 1.2 units of Persil per purchase occasion, then the mean population
sales rate ms will be given by:

ms = 0.25 × 4 × 1.2 = 1.2

Exercise 4.1 Applying the sales equation
1. In a stationary market, the penetration of Senso toothpaste is 0.07 over 24 weeks. Also
over 24 weeks, there are 21 purchases of Senso per 100 people in the population. What
is the purchase frequency?
2. How many purchase occasions per 100 consumers will there be in 48 weeks?
3. If the purchase frequency for the 48-week period is 4.6, what are the mean sales and
penetration?

Answers:

1. 0.21/0.07 = 3.
2. In a stationary market, you double the purchase occasions if you double the period: 42
per 100.
3. m = bw; therefore 0.42 = b × 4.6. So b = 0.42/4.6 = 0.09, or 9 per cent.
Section 2: Single Brand Purchase Patterns
The Impact of Recent Purchase
How does recent purchase experience affect the next purchase? In
particular, is there a bias towards purchasing the same brand as last time?
Consider two people who have both bought Persil and Ariel an equal
number of times over the last six months, as below:

Philip: Ariel, Ariel, Persil, Persil

Elizabeth: Persil, Ariel, Persil, Ariel

Both Philip and Elizabeth have bought Persil twice. Who is most likely to
buy Persil at their next purchase? If people learn more from their recent
experience, Philip is more likely to buy Persil next time. This is a first-
order explanation because it relates to the last purchase. A zero-order
explanation takes no account of the order of prior purchases and here there
would be no difference between Philip and Elizabeth in terms of their
likelihood of purchasing Persil. When the explanation is zero-order, we can
predict the likelihood of a future brand purchase using only the ratio of past
brand purchases.

Since people do occasionally switch brands, their most recent purchase

should be a slightly better guide to their next purchase than earlier
purchases. Kuehn (1962) found some evidence to support a first-order
effect; however, a study by Bass et al. (1984) showed that the majority of
purchases in most markets are zero-order. All studies have their weaknesses
and Kahn, Morrison and Wright (1986) argued that, because household
panel data were used by Bass et al., the first-order behaviour of individuals
might have been obscured. However, on balance, it seems likely that a zero-
order pattern of purchase is more common in stable markets and that habit,
rather than learning, provides the best way of thinking about repetitive
purchase.
Do Consumers Buy at Regular Intervals?
We have habits about what we buy, but are we also habitual about when we
buy? Purchase time habits would show up in panel data as an individual
tendency to buy a category once a week or once a month. Habits of this sort
would mean that a purchaser’s probability of buying rises sharply at
intervals. This pattern is also found for some frequent purchases, such as
newspapers or cigarettes. It also applies to shopping trips (Dunn, Reader
and Wrigley, 1983). Kahn and Schmittlein (1989) report that households
tend to be loyal to a particular day for grocery shopping, and East et al.
(1994) found that the majority of supermarket users were also loyal to
particular times of the day (Chapter 11).

Despite the routine timing of many shopping trips, brands are usually
bought at irregular intervals. There are several reasons for this. First, we
should note that most brands are bought quite infrequently. For example, a
typical US household buys a specific coffee brand about three times a year
and the category about nine times a year. This gives an average inter-
purchase interval between purchases of any brand of instant coffee of 5–6
weeks. Actual intervals differ because household consumption is varied and
shoppers may stockpile or run out. The prediction of when a specific brand
will be rebought is even more irregular because other brands in a category
may be bought instead. So, although there is a long-term average frequency
of brand buying, brand purchase occurs at irregular times. Mathematicians
describe this random pattern as a Poisson distribution.

However, people rarely buy a brand again immediately after purchasing it.
Because of this ‘dead time’ after purchase, the Poisson distribution does not
fit well for short periods (such as a week or less). Over the longer periods
covered in panel research, the fit of the Poisson assumption is close and
provides a basis for the mathematical models described later.
How Does Purchase Frequency Vary?
People differ widely in how much they buy of the category and of specific
brands. The range of purchase frequencies in a sample of buyers has a form
that is described by the Gamma distribution, like the one illustrated in Figure
4.1. This is a histogram of purchases of a frequently bought category in which
the largest number of buyers usually occurs at the lowest purchase frequency.

Few people buy heavily but those who do so are responsible for a large
proportion of a brand’s sales. Table 4.1 illustrates this with the purchases of
Kellogg’s Corn Flakes in the USA. In Table 4.1, you see that a sample of 100
households have bought 210 purchases of Kellogg’s Corn Flakes, 2.1 per
household, over three months. This average is based on the 55 per cent who
bought once, 22 per cent who bought twice, 8 per cent who bought three times
and so forth (a typical Gamma distribution). When we work out the sales from
these sub-groups we get the NBD distribution. This is the bottom row of Table
4.1 and it shows how important the few heavy buyers are for sales. Those who
bought six or more times – 5 per cent of all purchasers – were responsible for
20 per cent of sales.

Figure 4.1 Gamma distribution of purchase for a brand, showing that there are
many light buyers and few heavy buyers
In general, a substantial proportion of purchases are made by relatively few
heavy buyers; one rule of thumb, the heavy-half principle, is that the lighter-
buying 50 per cent are responsible for about 20 per cent of all purchases, while
the heavier-buying 50 per cent are responsible for the other 80 per cent. When
this rule applies, we find that the heaviest 20 per cent are responsible for about
50 per cent of sales. If you inspect Table 4.1, you will see that the 55 per cent
buying once are responsible for 26 per cent of purchases and the heaviest 23
per cent (buying three or more times) are responsible for 53 per cent of
purchases, so the data in Table 4.1 fit the heavy-half rule quite well. Other ratio
rules are more extreme. The best known is the 80:20 rule that 80 per cent of
purchases are made by the heaviest-buying 20 per cent of customers. The
precise ratio depends partly on the category. For example, if we investigated
those with savings accounts, we might find that very few heavy savers were
responsible for a large part of the total savings in a savings institution. The
ratio is less extreme when there is a natural ceiling on purchase within the time
period being studied, for example people rarely buy more than one newspaper
a day.

Ratios also depend on the period of time used to collect data. Because they buy
frequently, most heavy buyers will be sampled in any short purchase period.
Light buyers may not get around to buying in a short period, but as the period
lengthens more of them are captured. Therefore, if purchase data for instant
coffee are collected over a period of years instead of months, a greater
proportion of light buyers are recorded and the ratio moves from approximately
heavy-half to approximately 80:20. Ratio rules were first highlighted by Pareto,
an Italian economist, and they have been reviewed by Schmittlein, Cooper and
Morrison (1993).

The heavy-half principle shows that heavy buyers are an attractive segment in
many markets and marketers may therefore try to focus their efforts on them.
For example, promotions may give progressively more attractive benefits to
those who buy more and the frequent-flyer schemes that airlines run are
designed to benefit heavier users. Sometimes, it is possible to target heavy
buyers by using a particular distribution system. For example, a wine
warehouse, which sells wine by the case, may secure a larger proportion of
heavy wine buyers than a supermarket. Also, it may be useful to focus research
on heavy buyers since they are responsible for so much of the profit (e.g.
Hammond and Ehrenberg, 1995). In B2B marketing, key account management
has become a recognized speciality (the key accounts are the few big ones).

Ratio rules can apply to any phenomenon and Box 4.1 provides an interesting
example. Another application has been to the imprisonment of offenders. If
most crimes are committed by a relatively small group of offenders, crime will
go down if those offenders are imprisoned for longer.
 Box 4.1 Weight of consumption among
feline consumers (Churcher and Lawton,
1987)
An interesting demonstration of how consumption varies was provided by a study of what
the cat brought home to households in a Bedfordshire village. Over 70 domestic cats were
studied over a year and their tendency to kill and bring home sparrows, frogs, rabbits,
mice and so on was studied. One cat was responsible for 10 per cent of the total kill,
while at the other end of the distribution several cats brought back nothing in a whole
year.

The ratio of light to heavy buyers depends on the break point chosen, but
we can compare the top 20 per cent of customers that are typically
responsible for 50 per cent of sales with the bottom 80 per cent of
customers that are responsible for the other 50 per cent. On this basis, light
buyers are four times as numerous as heavy buyers. Because of their
numbers, light buyers in aggregate may offer more scope for sales gain than
the heavy buyers. However, when purchasers of a brand can be addressed
individually, it is clear that more attention should be given to the few heavy
buyers. This is partly because they could buy more and partly because their
loss could be very damaging. Mass communications such as advertising are
effective at reaching the large number of light buyers, but when marketing
to customers needs substantial resources it may be better to concentrate on
the heavy buyers.

It is easy to be confused by stationary market evidence. Although relatively

few heavy buyers are responsible for a substantial proportion of sales, this
does not mean that heavy buyers are responsible for most of the gain or loss
when sales change. First, as noted, this depends on the proportion of buyers
that are treated as ‘heavy’. If this is only the heaviest 20 per cent, any
change in total purchase volume may be greater for the more numerous
light buyers. Second, we find that light buyers tend to change their
purchasing proportionately more than heavy buyers. As a result, light (and
new) buyers are responsible for a large part of any gain in sales when a
brand improves its market share. But when the new pattern of sales has
stabilized, we are likely to still find that 20 per cent of buyers are again
responsible for 50 per cent of sales as a number of light buyers will have
been converted to heavy buyers. This analysis means that marketers must
not ignore the light buyers since their purchasing is more changeable, and
from their ranks some new heavy buyers may emerge. Further, as discussed
in Chapter 5, heavy buyers show some regression to the mean; that is, they
do not necessarily stay as heavy buyers.2 This evidence that sales gains
come mostly from new and light buyers may surprise those who are wedded
to relationship marketing with its emphasis on retaining customers. Two
books, How Brands Grow by Sharp (2010) and a sequel by Romaniuk and
Sharp (2016), have highlighted the fact that brands grow mostly by
acquiring new buyers.
How Does the Type of Product Affect Purchase?
Brands in product categories that differ substantially (e.g. food products and
household cleaners) may show similar patterns of customer purchase. For
example, if two brands in quite different categories have the same purchase
frequency and penetration, they will have much the same sole brand loyalty
and repeat purchase rates. Effectively, the purchase characteristics of a
brand are captured by purchase frequency and penetration, and a variety of
other brand performance statistics can be predicted from these measures.
This means that the specific brand or category need not be known in order
to predict these brand performance measures, provided that we know the
purchase frequency and penetration of the brand.
Repeat Purchase
If we compare any two adjacent sales periods, e.g. two quarters, we find that
many of the buyers of a brand in quarter 1 (Q1) return in Q2, particularly the
heavier buyers. But some do not return and these ‘lapsed’ buyers are replaced
by an approximately equal number of ‘new’ buyers. These ‘new’ buyers are
mostly light buyers of the brand, like those that they replace, and although they
did not buy in Q1, they have usually bought the brand before and are not really
new. At Q3 about the same proportion of Q2 buyers will drop out and be
replaced by others, including some of those who lapsed after Q1. This
intermittent pattern of purchase does not show loss of loyalty but instead
reflects the fact that many people buy a brand so infrequently that they often
miss quarters. The change of buyers at each quarter is explained mainly as a
probability effect, though a small part of this effect is due to more permanent
take up, and defection from the brand.

This analysis helps us to understand that repeat purchase rates depend mainly
on purchase frequency. A household that buys four times in one quarter is more
likely to re-buy in the next quarter compared with one that bought only once,
since the latter household may not need to buy again so soon. Because heavier
buyers are more likely to repeat, we find that the purchase frequency of repeat
purchasers is higher than the rate for the whole sample (by about 20 per cent).
In addition, new purchasers tend to be light buyers, and whatever the category,
their purchase rate usually does not rise much above 1.5 for any period.

Although repeat-purchase rates depend mainly on purchase frequency,

penetration does have a small effect, illustrated in Table 4.2. Compare the
baseline column with the next and you will see that a tenfold increase in
purchase frequency has a substantial effect on repeat purchase, raising it from
57 to 85 per cent; compare the baseline column with the last and you will see
that repeat purchase moves only three points when the penetration increases
tenfold. Note that it is the repeat purchase rate that is slightly affected by
penetration. The number of repeat purchasers is directly affected by
penetration; that is, as larger brands have more purchasers, they also have
commensurately more repeat purchasers, even if the repeat purchase rate is
little different.
Repeat purchase is an important diagnostic measure. If the purchase frequency
for a reporting period indicates a repeat purchase rate of 50 per cent under
stationary market conditions, then consistent deviations from this figure
indicate that the market does not have the normal stationary characteristics.
One application of this thinking was reported by Ehrenberg (1988: 97–8). A
brand launched 18 months before was being heavily advertised, but despite this
sales were constant. Two explanations were possible:

The advertising was ineffective and the brand was creating normal repeat
purchase.
The advertising was effective and consumers were trying the brand, but
they were not repeat purchasing so the brand was not gaining sales.

To distinguish these two, we must derive the normal repeat-purchase rate for a
stationary market and see whether this is what is found in the panel data.
Evidence that repeat purchase is normal favours the first explanation and
means that the advertising should be changed or stopped. Evidence that the
repeat purchase is below the normal level supports the second explanation. In
this specific case, the evidence suggested that the brand was weak, and that
sales would collapse when the pool of potential trialists was exhausted, so the
brand should be dropped before more advertising money was wasted. In a
study involving purchase data on more than 30 new brand extensions, Singh et
al. (2012) found that the successful new extensions were showing ‘healthy’
repeat rates within two or three quarters of the launch, but those new
extensions that eventually failed showed declining repeat rates during the same
period. This study shows that managers can use repeat purchase norms to
measure the performance of their new brands and to plan the investment of
resources accordingly.
Section 3: Patterns of Purchase in The Whole
Category
We now address the following questions:

How do penetrations and purchase frequencies vary in a product

category?
What changes do we see in penetration and purchase frequency when
market share changes?
Can we predict the buying frequency and penetration of a new brand if
it achieves a given market share?
When people buy more than one brand, how is their purchase
distributed between the different brands?
Does the evidence of cross-purchase support the idea of niche
positioning?
Is television watching like brand purchasing?
How is multi-brand purchase modelled mathematically?
Purchase Frequencies and Penetrations in a Product
Field

The double jeopardy ‘rule’

Table 4.3 presents data on the UK shampoo market, ordered by average sales.
You can see that bigger brands (shown by market share) have greater
penetrations and slightly greater purchase frequencies. This is typical of what is
found in most packaged goods markets. It means that changes in brand sales
will be seen mainly as changes in penetration. The low variation in purchase
frequency is not so surprising. Why should one brand of shampoo be used
more often than another by the people who like it?

The relationship between penetration and purchase frequency in a product field

fits the pattern known as double jeopardy (Ehrenberg, Goodhardt and Barwise,
1990). Double jeopardy (DJ) was described by the sociologist McPhee (1963:
133–40), who credits the original idea to the broadcaster Jack Landis. McPhee
noted how less-popular radio presenters suffer in two ways – fewer people
have heard of them, and among those who have heard of them they are less
appreciated. Applied to brands, DJ implies the pattern seen in Table 4.3, that
less popular brands are not only bought by fewer people (lower penetration)
but are also bought less often (lower frequency) by those who do buy them.
Ehrenberg et al. show that DJ is a ubiquitous phenomenon, occurring not only
in the purchasing of groceries, but also in such fields as the viewing of TV
programmes and the purchase of consumer durables, industrial goods and
newspapers.
Although DJ is easily seen, the explanation for the effect is less clear.
McPhee’s explanation took account of differential awareness. People who are
aware of less-popular presenters are usually also aware of more-popular
presenters and thus are more likely to ‘split their vote’ compared with those
who have only heard of the more-popular broadcasters. But the DJ effect is
seen in categories, such as supermarkets, where consumers are likely to be
fully aware of all the major brands in a market.

We see DJ as a statistical effect. Consider a board with 100 slots that can
receive counters. If you throw counters on to the board, the early ones will each
tend to get a slot on their own and each time that you do this, they raise the
percentage of slots with a counter (the ‘penetration’). As more counters are
thrown on to the board, they will increasingly land on slots where there are
already counters and this will raise the average number of counters in occupied
slots (the ‘frequency’). Figure 4.2 shows the theoretical relationship that
applies; further analysis of the theoretical double jeopardy line can be found in
Habel and Rungie (2005).

Notice that the relationship shown in Figure 4.2 is approximately linear for
much of the penetration range. The statistical relationship between penetration
and frequency will be disrupted in a number of ways in real markets. In
particular, there may be feedback effects so that, once consumers have bought a
brand, they may be more, or less, willing to buy the brand again. A second
empirical effect is that some consumers will never buy in some categories. For
example, those who do not own a cat rarely (if ever) buy cat food. This ‘out-of-
the-market’ effect will vary across categories and buying segments, and it
means that predictions should be corrected to take account of those who are in
the market.

Figure 4.2 Theoretical relationship between penetration and frequency

What else changes when sales change?

Table 4.3 is useful in showing the way in which penetration and purchase
frequency may be expected to change. Major changes over short periods of
time are rare, but looking at Table 4.3 if Vosene were to raise its share of sales
to that of Dove, it is very unlikely that it could do so by getting its existing
buyers to buy three times as much of the brand as they currently do. The
closeness of purchase frequencies across brands means that sales have to grow
mainly by increasing their penetration. Sometimes a gain in frequency may be
possible by persuading consumers to find new uses for a product, such as by
eating cereals at teatime, but this is a new use of the category which is likely to
raise the purchase frequencies of all brands.
Some evidence on what changes when a brand gains sales comes from
advertising cases. In 1982, advertising substantially increased the purchasing of
Curly Wurly, a chocolate-coated toffee liked by children in the UK. Sales then
remained approximately constant for the rest of the year (Channon, 1985: 168).
From the published data, it appears that among children of 7–11, the two-
monthly penetration increased by 60 per cent while the purchase frequency
increased by a mere 6 per cent. Among adult purchasers, the main gain was
also in penetration. Thus, sales gains came mainly because the advertising
attracted new buyers, in line with the DJ pattern. Other evidence comes from
the sales of Hellman’s mayonnaise (Box 4.2).
 Box 4.2 Attempts to raise purchase
frequency
There are some situations where increased consumption of a brand might be expected to
come from more frequent use by existing buyers rather than by increased penetration.
One such situation arose in the case of Hellman’s mayonnaise when it was new to the
British market. The manufacturers found that the British used Hellman’s mayonnaise on
little else but salads, a habit they had probably learned from their earlier experience with
salad cream, which looks rather like mayonnaise. The advertising for Hellman’s was
therefore designed to expand the number of situations in which the product could be used.

The case report (Channon, 1985) shows that, following the advertising, 40 per cent of
users were trying the product in the new ways suggested in the campaign. However, the
sales increase that was noted was still largely due to a gain in penetration in line with the
double jeopardy rule.

The relative constancy of purchase frequency is also important to those who

are hoping to break into a market with a new brand. With enough
advertising money and a sound product, marketers may occasionally
achieve a high market share. If they succeed with the new brand, the DJ
pattern will define how much of that success will come from penetration
and how much from purchase frequency. Supporting this, Wellan and
Ehrenberg (1988) found that, after rapidly establishing leadership in the UK
market, a new soap called Shield registered a purchase frequency
appropriate to its new position, i.e. slightly more than its competitors.

For the reporting periods used in market research, most sales changes affect
penetration, but Treasure (1975) pointed out that when sales are aggregated
over longer periods, changes in purchase level appear to be based more on
changes in purchase frequency and less on penetration. This difference
arises because of the way in which light buyers are counted over short and
long periods. Over short periods, an infrequent buyer tends to be a non-
buyer in the reference period and is therefore registered as a penetration
gain when he or she buys in a later period. Over longer periods, the
infrequent buyer is more likely to purchase in the reference period and
therefore any gain in purchase in later periods will be recorded as a
purchase frequency gain. There is a danger of too rigid an interpretation of
the double jeopardy pattern; the important matter here is that it is light
buyers who produce most of the change in brand purchase.
Patterns of Multi-brand Purchase
As we saw in Chapter 2, most buyers buy more than one brand in a category. In
many grocery markets, the average share-of-category requirement (SCR) is
around 30 per cent; this means that for every three purchases of a given brand,
the average buyer makes seven purchases of other brands. Ehrenberg (1988:
190) remarks that ‘buyers of one’s brand are perhaps mostly heavy buyers of
some other brand or brands, who occasionally also buy your brand’.

Is there a pattern to this multi-brand buying? There are two possibilities. One is
that certain brands are mutually substitutable. If most buyers see brands X and
Y as almost the same, we would expect a higher than average purchase rate for
Y among purchasers of X and vice versa. Such cross-purchasing of brands
would create purchase sub-sets or market partitioning. Ehrenberg and
Goodhardt (1979) demonstrate that this effect does occasionally occur, such as
in children’s cereals where those who buy one sweetened cereal are more likely
to buy another sweetened cereal rather than an unsweetened brand. An
alternative pattern is that the purchase of one brand is unrelated to the purchase
of other brands, and purchase rates simply reflect the penetrations of the other
brands. We find both effects in grocery markets. The basic pattern is that other-
brand purchasing is directly proportional to the penetrations of the other
brands. Ehrenberg (1988) calls this the duplication of purchase law.
Superimposed on this pattern are some cases of market partitioning.

Some people expect more market partitioning because they see some brands as
close alternatives. Yet while many of us will have our own personally preferred
groupings of brands, there may be little agreement between individuals. For
example, in the toothpaste market, during the course of a year one person may
buy Colgate and Aquafresh, another person Aquafresh and Macleans, a third
Macleans and a retailer’s private label. When these diverse combinations are
put together, the different individual cross-preferences will average out so that,
usually, there is not much evidence of market partitioning at an aggregate level.
When partitioning does occur, it can usually be connected with distinct product
features such as price, product form (e.g. flavour, pack type) or a common
brand name, rather than with the less tangible claims of the brand that may be
identified in advertising. Collins (1971) has reviewed this issue.
Table 4.4 shows how buyers of one brand of toothpaste also bought other
brands. These brands are arranged in market share order by column and row.
The bottom row shows the average cross-purchase (or purchase duplication –
the mean percentage of people buying a ‘column’ brand in addition to their
‘row’ brand). Table 4.4 shows some partitioning. There is a higher level of
cross-purchase between Colgate GRF and Colgate gel than is implied by
average cross-purchase. Compare the 32 per cent for Colgate gel bought by
Colgate GRF buyers with 24 per cent average duplication, and compare the 53
per cent for GRF bought by Colgate gel buyers with 35 per cent average cross-
purchase. This effect is common where two products with the same brand
name compete in the same field; it provides a basis for line extension since
some of the tendency to buy a brand seems to pass to the new line (see Chapter
3).
Cross-purchase and Positioning
Positioning is heavily emphasized in marketing. It is the set of beliefs about
the brand that the manufacturer and advertising agency seek to establish in
the minds of potential purchasers: in other words, it is an intended brand
image. One implication of the cross-purchase evidence is that new brands do
not need to have some unique formulation or brand image to succeed, a
finding which has worried those who attach strong importance to brand
positioning. If cross-purchase between brands depends largely on
penetrations, with exceptions relating only to price, formulation or brand
name, we must conclude that positioning, with its implication that each
brand occupies a distinct niche in the minds of consumers, is poorly
supported. When advertising does succeed, it may be because it has been
effective at producing high brand awareness and not because the brand is
seen as subtly different from other brands in its category. Indeed, many
successful brands may do well because they are perceived as typical of other
brands in the category, rather than as different from those brands. Thus,
manufacturers often focus on the strategy which Ehrenberg (1988) called
‘me-too’, i.e. copying the formulation and appearance of existing successful
brands and thereby trading on established purchase habits. This is a strategy
much used by retailers when they offer private label brands.

Positioning assumptions are well established in marketing. It is hard to

abandon the idea that a brand has a unique selling proposition (USP) which
is appreciated by consumers. This approach leads to the common idea of a
niche brand, one that is appreciated by some consumer segments but not by
others, even though they could buy it. This sometimes happens in fashion
(see Box 4.3). Of course, advertisers do need to show that their offering fits
consumer expectations but these expectations may relate more to the whole
category or to variants. (As we saw in Chapter 1, there are more differences
between the variants of a brand than between brands.) Also against the idea
of positioning, Uncles et al. (2012) show that the user profiles of directly
competing brands seldom differ. This work directly challenges the long-held
belief of marketers that brands are differentiated and therefore need to be
customized for distinct target audiences. Lack of segments is also observed
in studies on product variants by Singh et al. (2004, 2008). Such work
questions received wisdom and marketers should take note of this evidence.
 Box 4.3 Niche brands
Niche brands are those that are bought by one section of the population but not by other
sections that could afford to buy them (in this sense, expensive brands like Porsche are
not niche brands). To identify niche brands, it is necessary to compare the consumers who
buy different brands to see whether they differ in terms of demography or beliefs (e.g. age
or politics). One study compared the demographic profiles of brand buyers in different
grocery categories (Hammond, Ehrenberg and Goodhardt, 1996). This work showed very
little difference between the buyers of different brands, except in the case of cereals
where certain brands were bought only when there were children in the household. Given
this evidence, the realistic assumption is that there is normally little difference between
the buyers of any two brands in a packaged goods category. This is not so surprising
when we remember that the buyers of one brand are often the buyers of another brand.

But niches can occur. One interesting case arose when the fashion designer Burberry
launched a wide range of lower-price brand extensions which were taken up by ‘chavs’.
In the UK, this is a term used for those perceived as ‘uneducated and uncultured people’
(Wikipedia). Their endorsement of Burberry was unwelcome for the fashion house
because it could lead other potential buyers to avoid the brand.

The close association between other-brand purchase rates and penetration

does not hold when brands lack a common distribution structure, since the
availability of brands obviously affects choice. Thus, local variation in
distribution will produce local variation in other-brand purchases. This can
occur with beer brands since many are regional and are not available in
certain parts of the country. Uncles and Ehrenberg (1990) have shown this
effect with stores; they found that, in the USA, the cross-purchase between
two chains (Safeway and Lucky) was higher than predicted because these
chains tended to have stores in the same areas.
Watching Television
Television viewing has been shown to have similarities to brand purchase
(Goodhardt, Ehrenberg and Collins, 1987). When the programme is a serial,
viewing the next episode is like repeat purchase and we can ask how much
programme loyalty exists, as measured by repeat viewing. The evidence
shows that there is loyalty, particularly for serials with very high ratings. In
the UK, about 55 per cent of those watching a serial will repeat-view in the
following week but this figure is derived from several different loyalties.
Some people watch more television than others (i.e. they are loyal to the
medium) and this raises their chance of being a viewer of the next episode;
some people are channel loyal so that serials on favourite channels have a
better repeat-viewing chance; and some people watch more at particular
times of the day so that time loyalty may enhance repeat viewing. These
three loyalties ensure that people who have been watching a serial are quite
likely to be tuned to the same channel at the same time of the week after the
serial has ended. The difference between this end-of-serial viewing and the
repeat viewing when the serial is running indicates the true loyalty to the
programme. Such partitioning of loyalty can be applied to other categories,
for instance to the way in which loyalty to a car is divided between brand
and model.

Goodhardt, Ehrenberg and Collins (1987) also found some programme

partitioning, for example those who watch one sports programme are likely
to watch other sports programmes, but they also note that people watch a
wide variety of programmes so that cross-viewing has limited partitioning.
More recently, Lees and Wright (2012) found similar effects for radio
audiences, with some limited partitioning between the ‘talk’ and ‘music’
formats.

Terrestrial TV is a broadcast, not a narrowcast, medium and this makes it

difficult to target specific social groups accurately using the main television
channels. However, television transmission by cable, satellite and
broadband has increased the degree of partitioned viewing. Those interested
in particular genres, such as sport or children’s programmes, are able to
access more of these channels, leading to an increased duplication of
viewing between these genre-specific channels. Collins, Beal and Barwise
(2003) offer evidence on this point.
The Dirichlet Model
The Dirichlet is a mathematical model that predicts brand performance
statistics for all the brands in a product field. It was described in work by
Chatfield and Goodhardt (1975) and developed by Bass, Jeuland and Wright
(1976). It was presented in a comprehensive form by Goodhardt, Ehrenberg
and Chatfield (1984). The assumptions are similar to those for the single-brand
NBD model, but in addition it is assumed that the market has no partitioning.3
The model does not apply if there is appreciable evidence of brand clustering
on any other basis than penetration.

Programs running Dirichlet analyses can give predictions of penetration,

purchase frequency, sole buyers, sole-buyer purchase frequency, proportions of
buyers at different frequencies and the repeat-purchase rates of those buying
with different frequencies. Table 4.5 shows the real data for the US instant
coffee market together with the Dirichlet norms. The fit is fairly close, with the
exception of Maxim and Brim which may be a sampling error effect since
these are small brands.
The Value of Models in Marketing
Ehrenberg, Uncles and Goodhardt (2004) point out that mathematical
models provide norms against which real markets can be assessed. One
management application of such norms is to set out the realistic options that
are open to those who want to improve the share of their brand or wish to
launch a new brand on the market. Table 4.5 shows that the penetration and
purchase frequency of a brand is anchored in its market share; brand share
changes will relate mainly to changes in penetration rather than purchase
frequency. Models also provide norms for cross-purchase in an
unpartitioned market and show the long-run propensity to buy a repertoire
of brands. When markets are partitioned, for instance powdered and liquid
detergent, we can see the extent to which this partitioning affects cross-
purchase. In some cases, the market analysis, coupled with Dirichlet norms,
may help to show how much of a market is accessible to competition. For
example, if a new environmentally friendly detergent is being
contemplated, does it compete with all detergents, or all detergents that
make environmental claims, or some combination thereof?

More than 50 years of empirical evidence on brand performance metrics has

led to the derivation of a set of generalizable patterns. Ehrenberg et al.
(2004) summarize the main Dirichlet-type patterns as:

A. Brand penetrations vary within a product category and are much

lower for smaller brands.
B. Average buying frequencies do not vary much amongst brands in
a category. Underlying these averages, some individuals are heavy
buyers and others light buyers.
C. Smaller brands not only have fewer buyers than larger brands,
buyers of smaller brands also buy the brand slightly less often than do
buyers of bigger brands – the double jeopardy effect.
D. 100%-loyal buyers are relatively rare – of the buyers of a typical
brand in a year, almost all are multi-brand buyers and divided in their
loyalty.
E. Levels of loyalty are higher in shorter-length periods, mainly
because there are fewer opportunities to purchase different brands in
 shorter periods compared to longer periods.

These patterns are observed in the raw purchase data and are also closely
predictable from the Dirichlet model. Major exceptions from the Dirichlet’s
predictions are noticeably uncommon, but there may be minor exceptions or
deviations from the model’s predictions. In many cases, the exceptions are
simply explicable as one-off events (e.g. a serious stock-out or disruption
because of weather conditions, a strike or a fire in the warehouse) or
statistical anomalies (e.g. with small brands, the underlying customer
sample might be small, leading to a statistical sampling effect).
The Dirichlet Predictions
Table 4.6 shows the Dirichlet model predictions for the performance measures.
The predicted values (denoted by T for ‘Theoretical’) are derived from a few
inputs – brand share, penetration and average purchase frequency. The model
input data should apply to a specific time period – in Table 4.6 the data are for
a year – but note that the predictions can be for time periods of different length
(an average week, an average month, a year, two years, and so forth).

The model predictions in Table 4.6 are close to the observed values for almost
all measures. The closeness of the fit is clearly demonstrated in terms of low
mean absolute deviations (MADs) between the observed and the theoretical
values at the bottom of the table. There are no major deviations. The slight
under-prediction by the model for the penetration of 100%-loyal buyers is
possibly due to low sample sizes. Another possible deviation is for heavy
buyers of Crest Complete, which may be due to its unique product properties,
or it could be another statistical sampling effect because the brand is small and
therefore sample sizes are small.

Note: MAD = mean absolute deviation

Implications of the Model Fit
The Dirichlet theory confirms that customer behaviour towards a brand can
be described by a set of brand performance measures. Based on the brand’s
penetration and purchase frequency, other performance metrics, such as
SCR, percentage of 100%-loyal buyers, percentage of light and heavy
buyers, etc., are routinely predictable by the model. The main implications
of the finding that a wide range of markets behave in a ‘Dirichlet’ manner
are:

Individual brands do not need to be identified in order to predict their

repeat-buying and brand-switching properties, since these
characteristics are predictable for any brand in any time period, simply
from the brand’s market share. A brand’s health can be gauged simply
on the basis of its market share, penetration and purchase frequency.
This reduces the need for numerous parameters for the purpose of
decision-making.
In order to describe patterns of buyer behaviour and make aggregate-
level predictions, a focus on the individual consumer is unnecessary.
However, the model takes account of the revealed behaviour of
individual consumers in its distributional assumptions.
The routinely predictable loyalty-related measures provide
benchmarks that can help in identifying exceptionally high or low
performance levels for particular brands, and therefore highlight cases
that may deserve closer scrutiny.
The model tends to slightly under-predict the percentage of 100%-
loyal buyers; model discrepancies such as this can be further
investigated to see whether they hold in other situations and under
what conditions.

There is also increasing evidence to suggest that these patterns persist over
more extended periods of time – that, for example, patterns described for
one year are likely to be evident in the following years. There may be
volatility in the market shares of specific brands, but the underlying patterns
of buyer behaviour persist (Uncles et al., 2010b). Some of the discrepancies
are likely to recur, as has been found for the excess brand loyalty that has
been described for high market-share brands (Pere and Dawes, 2012).

A further value of this work lies in management education. An

understanding of stationary markets helps managers to read their own brand
statistics and thus understand the ways in which change may, or may not, be
brought about. When change does occur and the market stabilizes again, the
new brand performance statistics will fit the Dirichlet norms.
Summary
Over reporting periods of three months to a year, the sales of most
established consumer brands are approximately stationary: short-term
fluctuations due to promotions are averaged out and longer-term trends are
too slow-acting to have much effect. The steady state of such markets arises
because most buyers in a category maintain their propensities to buy the
same group of brands for long periods of time.

The penetration, purchase frequency and market share of a brand are key
statistics. These measures encode most of the buying propensities of
consumers so that there is no need to know anything more about the brand
in order to predict other brand performance statistics. This makes
mathematical modelling possible.

Brands have few heavy buyers but these are responsible for a large part of
the sales; the heaviest 20 per cent of buyers typically account for about 50
per cent of the purchases. However, because there are many light buyers
and some new buyers, a change in the purchasing of these groups can have
a substantial effect on sales. Over shorter periods (3–12 months), sales
changes are mostly seen as a change in penetration, as occasional buyers
return to the category, with only a small change in purchase frequency in
line with the rule of double jeopardy.

The NBD and Dirichlet models rest on assumptions that purchase incidence
is Poisson, that purchase rates in a population of buyers are Gamma, and
that (in the case of the Dirichlet) there is no market partitioning. The
predictions from such models usually fit the data derived from panel
research. When the fit is poor, the model provides benchmark norms for
interpreting the unusual brand performance.

Analysis of cross-purchase suggests that there is limited market

partitioning. The absence of substitution patterns between specific brands
indicates that the different brands in a category are seen by consumers in
much the same way. In these circumstances, a positioning strategy based on
brand differences may have little relevance.
Additional Resources
A clearly written but technical account of the NBD can be found in
Morrison and Schmittlein (1988), while the phenomenon of double
jeopardy is well explained by Ehrenberg, Goodhardt and Barwise (1990).
For a useful review of the Dirichlet and its applications, see Ehrenberg,
Uncles and Goodhardt (2004). An exercise designed to bring home the
features of stationary markets is available in Ehrenberg, Uncles and Carrie
(1994b). Two books designed for practitioners bring home the implications
of research on stationary markets; these are How Brands Grow (Sharp,
2010) and How Brands Grow: Part 2 (Romaniuk and Sharp, 2016).
Another easy-to-follow practitioner-focused account of the stationary
markets theory can be found in the chapter ‘Measuring the Market
Performance of Brands: Applications in Brand Management’ by Singh and
Uncles, in The Routledge Companion to Contemporary Brand Management
(Dall’Olmo Riley, Singh and Blankson, 2016).
Notes
1. A detailed discussion on the model (including mathematical equations)
can be found in Goodhardt, Ehrenberg and Chatfield (1984).

2. Firms sometimes conduct Pareto analyses and delete the worst-

performing brands or discourage the customers who buy the least. It is
important to consider whether such brands and customers could change. For
example, a customer at a DIY store could buy little for several years but
become a heavy buyer after moving house.

3. Within these assumptions, it is possible to derive an expression for the

probability of making r purchases in a period, pr. For those interested in the
technicalities: pr = [1−m/(m+k)]−kwhere k is a parameter that is estimated
from the purchase frequency and penetration. Expressions of the form
(1+x)n are called binomial; the equation for pr is a negative binomial
because the exponent is negative. The calculation of the NBD requires the
solution of the equation 1−b = (1+m/k)−kto obtain the parameter k.
5 Market Dynamics
Learning Objectives
When you have completed this chapter, you should be able to:

1. Sketch a 52-week sales curve showing seasonality and sales promotions.

2. Discuss the role of product line expansion, increased distribution, advertising and price
promotion in changing the baseline of sales of a mature brand.
3. Explain what will happen if there is an imbalance between customer defection and
customer acquisition.
4. Explain the social basis of diffusion theory.
5. Describe/sketch the technology substitution model, the Bass model and the typical trial
growth curves for new frequently bought products.
Overview
Chapter 4 described the regular patterns of purchase found in mature, stationary markets.
But what if a market is not stable? Sales and market share do sometimes change and new
products sometimes gather sales and become established. In this chapter, we explore
aspects of this change.

In the first section, we discuss how seasonality and sales promotions lead to fluctuations
in sales for mature markets, and how the baseline level of sales is affected by different
elements of the marketing mix. The second section describes the dynamic patterns in
loyalty underlying stable markets. We document these, show how they lead to dynamic
equilibrium and discuss the effects of disturbing this equilibrium. In doing so, we extend
some of the material from Chapter 4. In the third section, we consider the launch of major
innovations. These may create new markets or lead to the complete substitution of an old
way of doing things. We draw on the theory of innovation diffusion, as described by
Everett Rogers, introduce the technology substitution model and give an overview of
Frank Bass’s model of new product adoption. Finally, we consider innovation in
frequently bought categories, such as grocery products. These categories exhibit recurrent
minor product innovation, but arguably with a different social dynamic from that found in
the work of Rogers and Bass. We document the growth of first purchases for these minor
innovations and discuss the development of loyalty to such new products.
Section 1: Changes in Aggregate Sales
Variations in Demand – Seasonality and Sales
Promotions
Most of the sales fluctuations seen in mature markets are transient, and do
not affect the baseline level of demand for a brand or category. Businesses
typically experience large swings in demand that simply relate to the time
of the year, holidays, sales promotions and other events. More butter and
soup are sold in winter. There is little demand for Easter Eggs at Christmas.
Americans buy barbecue sauce in summer, with demand skyrocketing for
the 4th of July, assisted by heavy sales promotion. Similar patterns can
occur in durables and business-to-business markets. The weather, holidays
and tax refunds all nudge consumers towards buying certain products at
particular times. Even sales targets and discounting policies can lead to
seasonality in demand (see Box 5.1).
 Box 5.1 Discounts create irregular sales
Early in his career, one of the authors worked for a large American computer company
where the need to meet sales quotas led to a crescendo of effort at the end of the year. If
sales were slow, management would authorize company-wide promotions in the final
quarter, in the hope that they could still achieve their targets. Some customers responded
to this by adjusting their capital purchase cycle. When pressed about low sales mid-way
through the year, sales staff would tell management that their customers ‘liked to buy at
the end of the year, when the discounts are offered’.

It is important to be familiar with such temporary variations in demand.

Figure 5.1 provides an example. The graph covers two and a half years of
weekly data, and is representative of sales of widely used table condiments
in a major western economy. The x-axis shows intervals of 13 weeks, or a
quarter, which is a common management reporting period. The base of the
y-axis is zero, so differences in the height of the graph are directly
proportional to percentage differences in demand.

This figure shows spikes that increase sales by as much as 150 per cent
above the baseline level. They tend to be about four quarters apart. In other
words, the spikes occur at the same time (or season) every year. Why are
these sales spikes so large? Because sales promotions are often timed to
match holidays or seasonal upswings in demand, giving a large combined
effect. As mentioned in Chapter 4, despite the magnitude of these changes
there is usually little long-term increase in average yearly sales or in
customer loyalty. This is because most of the extra sales come from a
brand’s existing users (Ehrenberg, Hammond and Goodhardt, 1994a; Gupta,
Van Heerde and Wittink, 2003).

Many managers are surprised when they first see such period-to-period
sales variations, so it is important to be aware of them. The size of these
variations can cause supply problems and managers should take steps to
avoid stock-outs during periods of heavy demand, such as sales promotions.
Measuring Long-Term Sales Performance
These transient changes in demand make it difficult to identify baseline
demand or the underlying trend in sales. Figure 5.1, for example, shows an
upward trend over time, although this is hard to pick out at first glance. One
reason why market share is a popular measure of performance is that it
controls for some of these fluctuations; Seasonality and holidays should
affect all competitors equally, so market share is a more stable measure of
performance than raw sales data. In addition, many companies measure
market share using statistical techniques that further reduce the period-to-
period variation in the data and enable a clearer picture of trends in sales to
emerge.

Figure 5.1 Category volume for a grocery product

Some companies may apply seasonal decomposition to remove seasonality

effects, or exponential smoothing to reduce the impact of random variations.
Advanced econometric techniques, such as ARIMA (autoregressive
integrated moving average), may combine both methods, include estimates
of trend lines, and use dummy variables for holiday effects. By using
statistical models to identify the seasonal, promotional and holiday effects,
the trend in baseline demand is more clearly revealed.

Some researchers have gone further by using advanced forms of regression

analysis to tease out the effects of different parts of the marketing mix on
baseline brand sales. This is a tricky procedure that requires high-level
analytical skills and large data sets to deal with the seasonality, random
fluctuations, interactions between elements of the marketing mix, and
reverse causality that is common in these data. By reverse causality, we
mean the kind of phenomenon described earlier for barbecue sauce in the
USA, where the sales spike around the 4th of July causes the decision to run
a sales promotion, rather than the decision to run a sales promotion being
solely responsible for the increase in sales.

Ataman, Van Heerde and Mela (2010) used such advanced procedures to
analyse an extensive dataset from France. They found that, for mature
brands, product line expansion was most effective for increasing baseline
demand, followed by increases in distribution, increases in advertising and,
lastly, price discounting. This analysis did not consider the costs of these
initiatives and so the conclusion relates only to effects on sales and not
profit.

Srinivasan, Vanhuele and Pauwels (2010) reported a similar analysis, also

using French data. They did not consider the length of the product line but
otherwise showed consistent findings. That is, increases in distribution had
larger effects than increases in advertising or price promotions. Srinivasan et
al. (2010) also examined customer mindset metrics (advertising awareness,
brand liking and brand consideration) and found these to be strongly
associated with changes to baseline sales.

Other research confirms the strong association between distribution and

brand sales, and shows that it is non-linear with increasing returns to scale
(Wilbur and Farris, 2014). In other words, while the distribution effect is
strong, it shows double jeopardy with smaller brands having lower levels of
distribution that are also less effective in achieving sales.
Identifying the dynamic effects of marketing mix elements on brand sales is
a complex area and the work reported here is only preliminary. It shows that
dynamic change in mature markets is most strongly associated with changes
to the length of the product line, followed by breadth of distribution and
customer mindset metrics. However, much remains to be done, including
checking these initial results in other countries and on other products and
services.
Section 2: Dynamic Effects and Brand Loyalty
The Changes Underlying a Stable Market
Section 1 outlined the fluctuations in total sales that occur in stationary markets
together with some factors that underlie changes to baseline sales for mature
brands. In this section, we examine fluctuations in individual consumer
purchases. Even in a stationary market, individual buying will fluctuate. Some
of the commonly observed patterns include regression to the mean for heavy
buyers, long-term erosion of loyalty, and churn (turnover in the customer base)
as customers cease buying and are replaced. Customer churn is sometimes
mistaken for a permanent loss of loyalty that presages a drop in market share; it
is important to differentiate between a genuine erosion of the customer base
and the normal customer churn that is present for all mature brands.

We know from Chapter 4 that the NBD model describes the normal patterns of
light buying, heavy buying and repeat purchase. We can expand on this, using a
technique called conditional trend analysis (CTA). Conditional trend analysis
was developed by Goodhardt and Ehrenberg (1967) to extend the NBD to
different classes of buyers, so that repeat purchase can be predicted for light
versus heavy buyers, or indeed for any group of buyers who made a specified
number of purchases in the first period analysed. Managers should be
interested in this, because of the sales importance of heavy buyers, the desire to
increase the purchase rate of light buyers, and the need to convert some
proportion of zero buyers into regular customers. In particular, managers are
often concerned that a loss of loyalty from heavy buyers presages a decline in
brand sales and share. CTA reveals the level of repeat purchase that should
usually be expected for each group of buyers – heavy buyers, light buyers –
and even the purchase rate to be expected from people who were zero buyers in
the previous period.

Consider a brand that is bought by 15 per cent of category purchasers, on

average 1.6 times in a 13-week period, a fairly typical pattern. The NBD
analysis in Chapter 4 tells us that the expected repeat purchase rate in the next
13-week period is about 50 per cent (i.e. half of the people who bought the
brand in the first period will buy again in the second). But the NBD does not
tell us how repeat purchase is distributed between light and heavy buyers. To
compare actual repeat purchase with the norms for heavy buyers, we need to
undertake CTA. Table 5.1 reports some results from both normal NBD analysis
and CTA, broken down by the number of purchases made in the first period.

First, look at the set of buyers who made a single purchase in period 1. They
are important, as they make up 65 per cent of all buyers. Given that they
bought once in period 1, a manager may expect them to continue buying at the
same rate. It may be a surprise to find that this set of buyers makes just 0.6
purchases (on average) in the second period. Likewise, the set of buyers who
bought twice in the first period drops to 1.2 purchases in the second period and
the heaviest buyers have gone from an average 4.8 purchases down to 2.7.

It would be easy to interpret this as an erosion of loyalty, and to conclude that

you need to find the source of the problem and recruit replacement customers.
That would be a mistake, though not an uncommon one. It is what Andrew
Ehrenberg calls the ‘Leaky Bucket Theory’ (most recently, in Ehrenberg,
Uncles and Goodhardt, 2004) – the idea that customers are steadily leaking
away, requiring replacement by freshly recruited buyers.

In fact, the values in Table 5.1 are theoretical norms, rather than observed
values. They are just what we would expect in a normal, mature, stationary
market. Period 1 shows familiar NBD patterns; most buyers make only one
purchase, and 35 per cent of buyers account for 59 per cent of sales. In period
2, we simply see how the NBD plays out over time. We see how the 50 per cent
repeat-purchase rate is distributed across heavier and lighter buyers, and that
the average number of purchases for each group declines in period 2. Instead of
thinking of this as undesirable, we should recognize it as the typical pattern of
purchasing over time. Heavier buyers show higher repeat rates simply because
they are heavier and more regular buyers to start with. Despite this, their
purchases decline in the second period, exhibiting the familiar statistical
phenomenon of regression to the mean.
Total sales remain stable because those who do not repeat purchase in period 2
are balanced by an influx of other regular buyers. These are households who
did not purchase in period 1, but are nonetheless established buyers of the
brand. Similarly, a reduction in purchase frequency by some buyers will be
matched by an increase in purchase frequency by others. Rather than a ‘leaky
bucket’ in need of constant replenishment, this is the normal pattern of buying
in repertoire markets.

Some companies do monitor erosion of loyalty by examining repeat-purchase

rates for heavy buyers. This is sometimes called buyer flow analysis and data
for this are available from consumer panel companies. If you want to undertake
buyer flow analysis, be sure to use the NBD and CTA for a benchmark. You
will find that much apparent erosion is simply the normal pattern of occasional
purchasing.

The CTA model does slightly overstate the sales importance of repeat
purchases by period 1 buyers, so Trinh et al. (2014) proposed the Poisson log-
normal model as a more accurate alternative. Unlike CTA, the Poisson log-
normal model requires highly specialized statistical methods, and the
improvement is incremental rather than being a fundamental change. CTA
therefore continues to be a useful practical approximation for buyer flow
analysis.
Erosion of Loyalty in Repertoire Markets
The previous section demonstrated that much apparent erosion in loyalty is
simply due to customer purchase rates fluctuating from period to period.
That is not to say that there is no true erosion of repeat-purchase loyalty.
There is, but it is smaller than implied by a simple buyer flow analysis. True
erosion is revealed by examining the decline in repeat-purchase rates for the
whole customer base. For example, East and Hammond (1996) examined
repeat-purchase rates for supermarket products and found that they declined
from around 55 to 47 per cent over a year. This is eight percentage points,
and 8/55 is 15 per cent of buyers. Similar observations can be found in
Zufryden (1996). In a study of four brands in two segments, he observed
year-to-year retention of first-brand loyalty ranging from 79 to 86 per cent.
Thus, erosion ranged from 14 to 21 per cent. These values are a little higher
than those of East and Hammond (1996), perhaps because a loss of first-
brand loyalty does not necessarily represent a complete switch. It may
simply be a downgrade to a lower position in a customer’s repertoire, such
as moving from a favourite to second-favourite brand.

So, there is erosion but it occurs slowly. The leaky bucket theory is not
wrong, but the holes are quite small.
Defection and Churn in Subscription Markets
Leakage is often more obvious in subscription markets, where consumers
have only one supplier, such as insurance, banking, utilities and mobile
phone airtime. These markets may be based on an annually renewable
contract, as in the case of insurance, or they may be a ‘tenure’ contract in
which the subscription continues until terminated, as with banking or
utilities. Some markets, such as hairdressing or the family doctor, are
thought to show the same kind of persistent subscription-like behaviour,
even when choice is not constrained by a contract (Sharp, Wright and
Goodhardt, 2002).

In subscription markets, a loss of loyalty often represents a complete loss of

revenue from the customer concerned. Thus, the rate of defection is a key
performance metric and a leading indicator of a brand’s fortunes.
Sometimes, a loss of loyalty may represent a downgrade rather than a
complete defection, but may nonetheless be treated as if it were a full
defection. Banking is a prime example of this. Customers often have
relationships with more than one financial institution, and these
relationships tend to be long-lasting and valuable. Banks are interested in
whether they are a customer’s main bank, with the largest share of wallet,
and therefore the best opportunities for cross-selling. So, main bank
defection rates are treated in a similar way to other types of defection.

Annual defection rates in subscription markets typically range from 4 to 20

per cent of a brand’s customer base. Lees, Garland and Wright (2007)
reported a figure of 3.6 per cent for a main bank. Gupta, Lehmann and
Stuart (2004) found figures of 5 per cent for online stock trading and 15 per
cent for a credit card company. Wright and Riebe (2010) calculated, but did
not report, category switching rates in both consumer and business-to-
business markets. However, these were 4.1 per cent for the main bank and
20 per cent for annual industrial pipe contracts. Reichheld (1996b) claimed
that defection varied quite widely but averaged about 15 per cent across a
range of services.
Turnover in the customer base, often called churn, is not always due to
defection to or from another supplier. First-time market entry and market
exit both play a part. People may enter some markets as they reach
independence or form new households, and leave them as they cease to
support a family. There may also be churn due to upgrades or downgrades
in the product used or because of competitive pricing.1

In some fields, change is linked strongly to lifestage. Consider banking.

Over a lifetime, an individual may move from a child’s savings account, to
an overdraft, to a mortgage, to an on-call cash management account. People
may move house or be affected by a major life event such as their first job,
marriage, birth of a child, and so forth. This can affect the types of product
and service they require and thus their loyalty to previously purchased
brands. Some car models are suitable for young people, others for
executives and others for families. Similarly, there are regional banks in
many countries that cannot offer a full range of services in large cities. A
young family that moves from a regional centre to a large city may be
forced to change their bank.

Some evidence on the reasons for switching banks comes from Lees,
Garland and Wright (2007). They found that, in New Zealand, a change of
main bank was due to better offers 32 per cent of the time, to product or
service failures 31 per cent of the time, to reasons beyond the bank’s control
(such as moving house) 22 per cent of the time, and to a combination of
these reasons 15 per cent of the time. Bogomolova and Romaniuk (2009)
examined brand defection for a business-to-business financial service and
found that 60 per cent of defection was for reasons that managers could not
control. Chapter 2 carried evidence of defection in other categories.

Consequently, as with buyer flow in repertoire markets, there is a certain

level of inevitable turnover in the customer base. When interpreting
defection figures, theoretical norms should be used to determine how much
defection is normal, and how much is unusual. Wright and Riebe (2010)
provide a method of calculating such norms for subscription markets.
Change from Market Imbalance
If erosion, or switching, from a brand is exactly matched by customer
acquisition, a dynamic equilibrium will exist in which market share for that
brand remains constant. An imbalance between the two will lead to a
change. For example, reducing a brand’s defection rate while maintaining
the rate of customer acquisition will lead to an increase in market share.
Conversely, if the defection rate increases while acquisition remains stable,
market share will decrease. Defection reduction has been recognized as a
possible source of growth for some time (Reichheld and Sasser, 1990).
Reichheld (2003) further pursued the idea of growth coming from dynamic
forces through his ‘Net Promoter Score’, which takes account of the
propensity to give positive word of mouth (see Chapter 12). Other work has
also integrated defection rates into an understanding of customer lifetime
value (Gupta, Lehmann and Stuart, 2004) and compared the relative returns
of customer acquisition and defection reduction efforts (Reinartz, Thomas
and Kumar, 2005) to support an optimal allocation of the marketing budget.

However, work in this area tends to overlook the fact that defection
reduction is bounded. That is, defection rates are subject to double jeopardy
in the same way as other loyalty metrics (Wright and Riebe, 2010) and so
are not easily changed without a substantial accompanying change to
market share. Thus, if a brand has 6 per cent defection, this can only be
reduced a little without violating the principle of double jeopardy.
Furthermore, no matter how superior the product or service, some loss of
customers will occur due to market exit, changes in customer requirements,
or other matters beyond the control of managers.

Recent work has compared unusual defection and unusual acquisition as

sources of dynamic change in market share (Riebe et al., 2014). The
evidence leads to the conclusion that acquisition rates are much more
important than defection rates in explaining changes in market share for
both growing and declining brands. In other words, when there is a
deviation from the double jeopardy norm, it tends to occur for acquisition
rates rather than defection rates.
Exercise 5.1 Defection reduction and share
growth
Question: Imagine you have a brand with a 3 per cent market share and an annual defection
rate of 10 per cent. How long will it take to increase your market share from 3 per cent to 10
per cent, simply by halving your defection rate?

Answer: Assuming your market share has been in a dynamic balance and that all customers
buy at the same rate, your 10 per cent annual defection rate will have been matched by a 10 per
cent annual customer acquisition rate. If your defection rate halves to 5 per cent while your
customer acquisition rate remains constant, you will grow by 5 per cent of your customer base
per annum. That is, your market share after T years can be found by a simple compounding
formula, as follows:

3% × (1 + .05) ^ T

We suggest you conduct a simple spreadsheet analysis to find the answer. We did this and
found that market share will rise above the 10 per cent threshold in the 25th year. However,
defection reduction can be an expensive exercise so there is a risk that market share growth
will not be matched by profit growth.

Comment: This example is somewhat similar to that of MNBA, outlined in Reichheld and
Sasser (1990). Reichheld and Sasser were more concerned with customer lifetime value than
with growth. However, we hope our example makes it clear that brand growth requires a
broader explanation than just defection reduction, especially when brands have a relatively low
market share.

Ask yourself the question: what else could lead to brand growth?
Observing Market Change
One consequence of occasional buying and relatively low levels of erosion
is that marketing actions can take a long time to bear fruit. Exposure to
advertising, for example, cannot affect brand choice until the next purchase
occasion. Similarly, defection-reduction programmes seek to plug a leak
that may be fairly small to start with, while attempts to acquire customers
will be limited by the number entering the market or seeking to make a
change. So it may take some time for marketing initiatives to play out in
changes to sales. An exception is sales promotion, which tends to have a
large effect during the period of the promotion, leading to the spikes in sales
seen in Figure 5.1.

Srinivasan et al. (2010) identified wear-in time before a change to the

marketing mix became effective. The wear-in time for advertising and
distribution was about two months, for price about 1.6 months, and for
promotions just one month. Ataman et al. (2010) similarly found that the
long-term effects of marketing mix changes were much greater than the
short-term effects, except for discounting where most of the effect was
immediate.

Similarly, market problems, such as a drop in perceived quality, may not be

immediately noticeable because it may be some time before affected
customers are ready to purchase the category again. By that time, quite a
weight of dissatisfaction may have accumulated so that many will switch on
the next category purchase. This makes it important to include satisfaction,
perceived quality and other customer mindset metrics when evaluating
brand performance. These more qualitative data may highlight potential
problems, allowing a manager to respond before the problems gather too
much momentum.
Section 3: The Dynamics of New Product
Adoption
So far we have examined change in mature markets. What about the more
fundamental changes that result from the launch of new products? Whether
they satisfy a previously unmet demand, or replace a previous technology,
new products can result in more rapid and enduring change than is seen in
mature markets.

How are such innovations adopted by a population of consumers? While

there is a large literature on this topic, we will restrict ourselves to
examining three approaches of particular importance to marketers. These
are Rogers’ innovation diffusion curve, Fisher and Pry’s technology
substitution model and the Bass model.
Rogers’ Approach to Innovation Diffusion
Most marketers have some awareness of work on the diffusion of
innovations. The most well-known author in this area is Everett Rogers
(1962, 2003), whose book on the subject is a widely cited classic. Rogers
brought together studies from many disciplines, but gave emphasis to
sociology. He followed influential work by: Gabriel Tarde (1903), who noted
that cumulative adoption followed an S-shaped curve and saw adoption or
rejection as a critical decision; George Simmel (1908), who introduced the
idea of a social network; and Ryan and Gross (1943), who undertook a
landmark study into the adoption of hybrid corn. Later work incorporated
Granovetter’s (1973) finding that the spread of an innovation through a
social system was helped if people were loosely bonded to many different
groups.

In Rogers’ work, these ideas are more fully developed and integrated, with
diffusion defined as ‘the process in which an innovation is communicated
through certain channels over time among the members of a social system’
(Rogers, 2003: 5). Rogers believed that adoption follows a normal
distribution curve, with time as the x-axis and number of adopters as the y-
axis. Figure 5.2 shows this normal curve. There are two versions: the
standard normal curve of (non-cumulative) adoptions in each period and the
S-shaped curve of total (cumulative) adoption. Historical examples of new
products that become widely established include black and white televisions,
computer chips, air conditioning units, facsimile machines and first-
generation mobile telephones. Modern examples might include rooftop solar
power generators, ultra-fast broadband and various forms of wearable
computing such as the Apple Watch.

Figure 5.2 A normal adoption curve

Characteristics of an Innovation
Innovations do not always follow this curve. They may fail. Or they may
diffuse so slowly that the left tail, or leadtime, of the adoption curve extends
for many years before rapid growth kicks in. Rogers cites as examples the
Royal Navy’s reluctance to use citrus juice as an anti-scurvy agent,
resistance to the use of boiled water in Peru and the failure to displace the
QWERTY keyboard on which this book has been rather inefficiently typed.
He became interested in the characteristics affecting the innovation adoption
rate and posited that it depended on:

The relative advantage that an innovation has over previous methods of

meeting the same need. Some innovations, such as text messaging and
social networking tools, also have strong network effects, so that
relative advantage increases with the number of adopters.
 Compatibility, or the consistency of an innovation with the existing
experiences, needs and values of the adopting population. Cultural and
religious incompatibilities may be particular risks for the adoption of
innovations.
Complexity, or how easy an innovation is to use and describe to others
and whether it requires specialized expertise or substantial learning.
Trialability, enabling experience to be gained with the innovation
before purchase or full adoption.
Observability, increasing both awareness and social influence, as we
are influenced by seeing others use new products and tend to follow the
consumption habits of the communities we live in.

There is also the perceived risk of an innovation. The higher the price, the
less the confidence in after-sales service, or the harsher the returns policy,
the more reluctant people will be to adopt. Rogers’ adoption characteristics
also have an indirect impact on perceived risk, which will be reduced by
trialability and observability but increased by problems of compatibility or
complexity.

Adopter Categories and Innovativeness

Perhaps the most famous part of Rogers’ work is the division of the normal
curve into adopter categories. People in the first 2.5 per cent of the normal
curve in Figure 5.2 are labelled the innovators and the next 13.5 per cent the
early adopters. Then we have the early majority (34 per cent), late majority
(34 per cent) and laggards (16 per cent). These adopter categories are
reproduced in most consumer behaviour textbooks, together with comments
about their typical characteristics.

This approach has come in for some serious criticism. Bass (1969) famously
described Rogers’ approach as ‘largely literary’. Wright and Charlett (1995)
pointed out that Rogers’ adopter categories were a post hoc tautological
classification system, of no value for forecasting. This is because Rogers’
categorization depends on the standard deviation from the mean time to
adopt. Innovators are all those up to 2 standard deviations before the mean
time to adopt; early adopters are between 2 and 1 standard deviations from
the mean time to adopt; and so on. Yet neither the mean nor the standard
deviation can be calculated until the diffusion of the innovation is complete,
at which point the adopter categories have little managerial value.

A counter-argument is that innovativeness is a normally distributed trait,

associated with other consumer characteristics in a predictable manner,
allowing innovators and early adopters to be identified and targeted. Rogers
claimed that there were 26 characteristics that varied between adopter
categories, including socio-economic variables, personality values and
communication behaviour. However, empirical findings do not always
support this claim. Taylor (1977) found that a striking characteristic of
‘innovators’ in grocery products was that they were simply heavy buyers of
the category, a finding that was recently confirmed for pharmaceuticals
(Stern and Wright, 2016). This hardly suggests an enduring personality trait.
A more popular modern view is that innovativeness varies by product
category.

The long leadtime that can occur before an innovation takes off is sometimes
explained by the characteristics of the most innovative people. Rogers notes
that innovators are often ‘deviants’, somewhat apart from the rest of the
social system: they might be metaphorical hermits in the wood, dwelling
apart from the community, or perhaps the archetypical computer geek. While
such people may be innovative, they could have little social influence, so
their behaviour has little effect on the broader community. Diffusion will
only take off when more influential people, well connected within the social
system, adopt the innovation. In a tribal village, this might be the chief or the
chief’s wife. In western social networks, it may be opinion leaders or
celebrities, or people who offer marketplace advice to many others, such as
Feick and Price’s (1987) ‘market mavens’ (if they can be found; see Chapter
12). Rogers places such people in the early adopter category.

While these ideas offer insights into ways of speeding up adoption within
particular social systems, they are not necessary to explain the S-shaped
curve that we see in practice. A long lead-time can be explained through the
cumulative effects of social forces, as we shall see later. Nonetheless,
Rogers’ work is very useful. The concept of an idea spreading through a
social system via specific communication channels underlies much
subsequent work on innovation diffusion. The insight that adoption follows a
normal curve has been borne out in many product categories.
Fisher and Pry’s Technology Substitution Model
Rogers’ adopter categorization exploits the properties of the S-shaped
normal distribution curve, but is not helpful for forecasting. Marketers need
to make forecasts to assist launch decisions on new products, set targets
against which performance can be assessed, and allocate marketing
expenditure.

Fisher and Pry (1971) showed how to use an S-shaped curve to forecast the
replacement of an inferior technology with a better one that meets the same
need. This process is called technology substitution. They dealt with the
problem of an indeterminate leadtime by assuming the substitute technology
must achieve a several per cent market share before complete substitution
could be guaranteed to occur. They then treated substitution as a logistic
function of time.2 This has the practical effect of transforming the S-shaped
cumulative adoption curve into a straight line. Extrapolation through
regression then yields a forecast. Fisher and Pry validated this model over 17
diverse data sets and found surprisingly accurate results.

Figure 5.3 is an example of the raw pattern of technology substitution, using

some previously published data on diesel and steam locomotives in the USA.
These data are expressed in relative percentages of the total number of
locomotives. You will see that the growth curve for diesel locomotives is
similar to the cumulative normal adoption curve shown in Figure 5.2.

Although quite old now, Fisher and Pry’s (1971) model is likely to continue
to be of great importance as we undergo accelerating technological
substitution in areas such as telecommunications, computing and home
electronics. This model has the useful feature of forecasting the dynamic
decline of the old technology as well as the growth of the new one. This is
potentially very helpful to companies in fields such as telecommunications,
where there are substantial but declining revenue streams from products such
as fixed-line telephone connections.

By way of example, consider the following incident from the period in

which broadband was displacing dial-up Internet connection throughout
Australia (Box 5.2). We use the Fisher and Pry method and describe the
workings so you can see how easily the model is applied. Note that the date
had to be adjusted to an equal interval scale to allow the statistical analysis
to take place. Figure 5.4 graphs the example in Box 5.2, in this case in
fractions rather than percentages.

Figure 5.3 Cumulative substitution of locomotives

Source: Interstate Commerce Commission, Statistics of Railroads 1925–1960, as cited in Moore and
Pessemier (1993: 79)

Figure 5.4 Cumulative substitution of Internet access

 Box 5.2 The growth of broadband
We applied the Fisher and Pry model to forecasting the decline of Australian dial-up
Internet access and, conversely, the growth of broadband in that country. Using Nielsen
net ratings figures published at www.dcita.gov.au, we converted the proportions of
Internet users that had dial-up and broadband access to a log odds ratio (LOR) and ran an
OLS regression. The OLS regression predicted that LOR = 3.214 − 0.068 × adjusted date.
(We had to use an adjusted date to ensure an equal interval scale, as the time between
reports varied.) We put future dates into this equation to predict changes in LOR and then
converted these changed LORs back to predicted proportions of dial-up and broadband
connections. Figure 5.4 shows our forecasts. Forecasts like this give useful long-range
expectations about the size of the customer base and the support infrastructure needed
from capital equipment to call-centre staff. They are also useful to help forecast the
market potential for Internet services relying on a broadband connection, such as Internet
TV.

Postscript: The forecasts were based on data up to June 2006, at which stage 71 per cent
of Internet users had broadband. This was forecast to grow to 95 per cent by the end of
2008. In 2009 a press release from Nielsen online revealed that the actual figure for
broadband penetration in 2008 was 97 per cent.

The technology substitution model gives impressive forecasting ability, but

it does not offer much explanation. It is also constrained by the simplicity of
the assumptions. In their original article, Fisher and Pry (1971) applied their
model to the substitution of margarine for butter, water-based for oil-based
paints, and artificial for natural fibres. We now know that those
substitutions are partial – there is a limit on their potential that is less than
the total usage of yellow fats, paint and clothing fibres respectively.
Sometimes, silk stockings are still preferred to nylon.

It would be ideal to have a model of innovation diffusion that offered the

forecasting ability of Fisher and Pry, the explanation of Rogers, and a
variable ceiling on the number of adopters. As it happens, there is such a
model in marketing – the Bass model.
The Bass Model
Frank Bass’s (1969) approach to diffusion of innovation is one of the most
well-known models in marketing. Bass drew on epidemiological theory –
the theory of the spread of diseases – to examine the spread of an
innovation throughout a connected social system. He modelled the effect of
both innovative and imitative forces in promoting adoption up to a
saturation point, and saw innovative forces as those external to the social
system, such as advertising and personal selling. Innovation therefore acts
throughout the diffusion process rather than being limited to the first 2.5 per
cent of adopters. Conversely, imitative forces reflect social influence, which
is internal to the social system. Therefore, imitative forces are exerted by
previous adopters as they model and recommend the innovation to others.
For example, you might buy an iPad simply because you see an
advertisement (innovation), but your purchase becomes more likely the
more you see people using one, the more your friends and family own them,
and the more they tell you about it or show you how it works (word of
mouth). Unlike earlier researchers, Bass found a way to quantify these
social influence variables within a mathematical model. An early success of
the Bass model is illustrated in Box 5.3.
 Box 5.3 A strong test of the Bass model
The Bass model is lauded for having passed the key scientific test – prediction. Bass used
his model to predict the peak of the colour television adoption curve in the USA. He was
roundly criticized at the time by industry figures for his ‘pessimistic’ forecasts. However,
when actual sales figures became available, he was proved right. Ignoring his forecasts led
the consumer electronics industry to build too much production capacity, at considerable
cost.

Parameters, Shape and Equations

Bass’s model is elegant. It requires only three parameters: innovation (p),
imitation (q) and the eventual number of adopters, known as market
potential (m). The dependent variable is either cumulative or non-cumulative
adoption by time period – Y(T) or y(t) respectively. Sales data can be equated
with adoption data if the purchase is a high-value durable or service that is
bought once and does not need replacement for some time.

In the first period (usually a year), the probability of adoption, given that no
adoption has yet taken place, is simply p, and the predicted number of
adopters is p × m. However, as more of the population adopts, social
influence (q) increases and the probability of adoption becomes p + q ×
[Y(T)/m]. This basic equation can be expressed in a variety of forms, both
algebraic and probabilistic and for either cumulative or non-cumulative
adoption.

The mix between internal and external influence can be seen in the Bass
curve in Figure 5.5, estimated on data for first-time use of a telephone
enrolment system at Massey University in New Zealand. The parameters of
this curve are p = 0.10, q = 0.47, and m = 5,788. This p value is quite high
for Bass modelling, so the adoption curve starts at a point well above zero,
peaks early, and is over quickly. This shows the extra flexibility of the Bass
model compared to a simple normal distribution curve.
Validation, Replication and the Extension of the
Bass Model
Box 5.3 notes a famous case of forecasting success for the model but Bass
may have been lucky on this occasion. More cases are needed to assess it. In
a review of early replications, Wright, Upritchard and Lewis (1997) noted
that a poor fit was sometimes found for the model outside North American
and European settings. They also found that its predictive ability was limited
by variations in the early data. As early sales (or adoptions) are quite few,
fluctuations are proportionately larger and so have a larger effect on long-
term forecasting accuracy. Putsis and Srinivasan (2000) subsequently found
that Bass model forecasts generally do not stabilize until the peak of the
adoption process has passed. Also, the assumption of constant social
influence may be an over-simplification. As we shall see in Chapter 12,
word-of-mouth production may vary with the period a person has owned a
product.

Figure 5.5 Adoption of an interactive telephone enrolment system (non-

cumulative)
 Source: Model parameters from Wright, Upritchard and Lewis (1997)

Notwithstanding this, there has been a wealth of successful research on the

Bass model, including applications to other historical data sets,
methodological improvements and many useful extensions, such as allowing
population size to vary or modelling the successive generations of new
technologies. Bass models have accurately described innovations in areas as
diverse as consumer durables, consumer electronics, education, agricultural
services, telecommunications and even United Nations membership. In these
areas, only one purchase is normally made; thus, sales (or membership) data
give a direct measure of diffusion, up to the point where replacement
purchases commence.

Using the Bass Model to Predict

To predict the sales of an innovation using the Bass model, we need to know
the values of p, q and m. The usual way to estimate these is through non-
linear regression against sales data for the first few periods. However, this
requires several years of such data and some degree of statistical skill.
Furthermore, as noted earlier, the results tend to be unstable until the data
include the peak of the adoption process.

It would be much more helpful to apply the model before the launch of a
new product. Then it could be used for initial sales forecasting, launch
decisions, capacity planning and marketing budgeting. Two traditional
approaches to doing this are to forecast by analogy or to use management
judgement.

To forecast by analogy, known values of p and q from other innovations are

used. Sultan, Farley and Lehmann (1990) reported the results of a meta-
analysis of these parameters, and tables of these have been published
elsewhere (e.g. Lilien and Rangaswamy, 2002). However, Rogers’ theory
tells us that the characteristics of the innovation and the characteristics of the
social system are the key drivers of adoption, so it is questionable whether p
and q values are generalizable unless they are from a similar innovation in a
very similar social system. For the m parameter, the value cannot be derived
by analogy, but may be estimated through market research.

The techniques for applying managerial judgement are not made very clear
in the literature and should generally be seen as a last resort. Managers may
not understand the social basis or typical values of the Bass model
parameters p and q and will therefore have little basis to apply their
judgement. Also, Armstrong (1985) has evaluated evidence on the
performance of subjective estimates – managerial judgement – in
forecasting. He found that they do not perform well compared to forecasts
from objective methods and that integrating objective information with
managerial judgement typically provided substantial improvements over
managerial judgement alone (Armstrong, 1985: 387–420). So it is doubtful
whether management judgement will show much accuracy unless the judge
already has substantial experience with the Bass model and takes steps to
combine this judgement with other, objective, information.
Lessons for Market Dynamics from the Bass
Model
What lessons does the Bass model offer us in understanding market
dynamics? First, consider the domain of the model. It concerns new
products, not brands. It is typically applied to high-value products or
services that have long inter-purchase times. These may be completely new
or substantial technological improvements (although it can be applied to
other behaviours if true adoption data are available, as in Figure 5.5). Bass
modelling has shown that sales curves following the introduction of such
products and services will be roughly normal or S-shaped.

Second, the exact shape of the curve will depend on social factors relating to
the social system involved. Understanding the relative importance of these
external and internal influences will give richer insights for marketing
planning and more accurate forecasts of product sales growth.

Third, a long lead-in time can be explained simply by the values of the
parameters. It will occur if the innovation parameter, p, is relatively low.
Rapid acceleration of adoption will then occur if the innovation parameter, q,
is relatively high.

Therefore, knowledge of the Bass model gives a good understanding of

market dynamics following the introduction of a new product, at least for
durables or high-involvement services. However, the Bass model has little to
say about frequently bought products or new brands in established markets.
We address this in the final section of this chapter.
Exercise 5.2 The flexibility of the Bass curve
The following graphs show how changes to the parameters affect the shape of the Bass curve:

Figure 5.6 Four cumulative Bass adoption curves

Examine each one of the four examples and then consider the questions that follow:

For each curve, identify a social system and innovation it could be describing. (Hint: think
about the level of innovation and the extent of word-of-mouth recommendation.)
Consider Rogers’ innovation characteristics. How might they affect the shape of the
cumulative adoption curve?
Which do you think are the most and least common shapes? Justify your answer.
Section 4: The Sales Dynamics of Frequently
Bought Categories
Low-Innovation Products
Section 3 described what we know about sales changes that occur when
new categories are introduced, or when technological substitution causes a
wave of change in an existing durable category. Yet most new products are
much more mundane and involve incremental ‘me-too’ products in familiar,
frequently bought, low-value and low-involvement categories. We cannot
expect that the same dynamic processes will necessarily apply in such
different circumstances. Rather, new products in these frequently bought,
low-involvement categories rely on familiarity and existing memory
structures, as discussed in Chapter 3, and most sales come from repeat
purchases rather than from the initial adoption decision. Occasionally, there
may be a major innovation, such as the first olive oil spread or low-fat
calcium-enriched milk. However, even these are typically presented as new
brands launched into a familiar category (yellow fats and milk,
respectively).

Nonetheless, diffusion theory offers some useful insights. In general, we

can expect that trial and adoption will be affected by influences external to
the social system (advertising, promotion and personal selling) and
influences internal to the social system (observation of others and word-of-
mouth recommendation). We expect that high-value durables will be more
visible and will tend to be a topic of conversation, because of their novelty
or because the high price level leads people to seek pre-purchase
recommendations. The result is that social influence increases with the
number of previous adopters, leading to the S-shaped cumulative adoption
curve.

Now, consider a low-value, frequently bought item. The low value means
lower risk. Also, it may be one of many dozens of regularly purchased
product categories. The category is less likely to be a topic of conversation
– who talks much about toilet paper, bleach, milk, baking powder, flour,
light bulbs, butter and the like? Of course, there will be exceptions – a
cooking club may talk a great deal about a new type of flour. However, such
exceptions represent a small fraction of the buyers and we are not aware of
any corresponding light bulb-changing or milk-drinking clubs.
Rogers’ adoption characteristics can easily be applied to such low-value
incremental innovations in frequently bought categories. We suggest that
such innovations are familiar (compatible), simple (not complex), cheap
(trialable) and widely present on the shelf (observable). So there are few
barriers to the adoption of such a new product, provided it has a relative
advantage. Conversely, such products are less likely to be topics of
conversation, so there will be relatively little social influence.3 The
probability of adoption remains constant over time rather than increasing as
social influence increases; however, the number of adopters falls with each
period as the pool of those who have not yet adopted shrinks. The result is a
curve whose shape is known as exponential or an exponential decline.
Figure 5.6(d) in Exercise 5.2 shows the pattern of cumulative adoption for
such an exponentially declining adoption curve.
Exercise 5.3 Sketch an exponential decline
curve
It’s easy to see for yourself how an exponential decline curve works. Imagine that a population
of 100 people has an adoption rate of 10 per cent. In the first period, 10 will adopt; in the
second period, 9 (10 per cent of the remaining 90 who had not yet adopted); in the third period,
8.1 are expected to adopt (10 per cent of the remaining 81); and so on:

In Excel, calculate the number of adopters for each of the first 20 periods. Assume a
population of 100 and an adoption rate of 10 per cent. (Hint: you might find it helpful to
have columns for those yet to adopt, adopters in the period, and total adopters to date.)
Graph both the cumulative and non-cumulative adoption curves.
Repeat the exercises for adoption rates of 30 per cent and 5 per cent.
Empirical Patterns in First Purchases
The shape of adoption curves for frequently purchased products has been
studied empirically. Wright and Stern (2015) examined 12 national product
launches and 19 controlled test market product launches for a variety of
frequently bought categories. Figure 5.7 sketches some of their findings. It
is noteworthy that the individual product launches showed similar patterns,
despite varying degrees of innovativeness in the product or brand being
launched.

These curves are not obviously S-shaped and so show little evidence of an
increasing word-of-mouth effect (although see note 3). They show an
exponentially declining pattern of growth in first purchases. There is a
slight bend in the curve for national launches, although this may be due to
distribution growth. Yet, even with this, the curve still generally follows the
declining exponential pattern. Nonetheless, it is worth remembering that
changes to distribution are a primary cause of changes to baseline sales;
availability is a necessary precursor to purchase. This is confirmed by the
work of Ataman, Mela and Van Heerde (2008), who show that distribution
plays the greatest role in the success of new brands in repeat-purchase
markets.
After the First Purchase
For durable products, adoption and sales are measured by the simple act of a
purchase. However, the picture is not so clear for frequently bought items
such as groceries. Exponential models of cumulative first purchase do not
assume that the first purchase is an adoption, or tell us what the long-term
purchase frequency will be. We might instead expect buyers of these new
products to make one or two trial purchases before choosing either to drop
the new product or to include it in their repertoire. This is the traditional
view in most consumer behaviour texts and it is consistent with Rogers’
theories on innovation diffusion, with exposure being followed by evaluation
and then an adoption or rejection decision.

Figure 5.7 Cumulative first purchases for 52 weeks from launch

This traditional view turns out to be surprisingly controversial. Light buyers

may only buy the category two or three times a year, so trial will continue
for some time and it is hard to say when a light buyer will have completed
their evaluation, or even whether such a question is meaningful. If somebody
makes a low-involvement purchase of a brand once or twice a year, it is not
clear that this involves much cognition or post-purchase evaluation, as is
expected by traditional theory.
Evidence on this point was provided by Ehrenberg and Goodhardt (2001),
who found that the long-term average purchase rates for a new brand are
achieved very quickly. In a study of 22 grocery product launches, they
compared average purchase frequencies for new brands with those of the
existing brands. In the first quarter, the new brands had an average purchase
frequency of 1.4 – a little lower than established brands, which averaged 1.9.
In subsequent quarters, the new brands showed a purchase frequency of 1.8,
1.9 and 2.0, virtually identical with the established brands. Wright and Sharp
(2001) found similar results for an Australian grocery product launch. These
findings are inconsistent with the traditional idea of a trial purchase followed
by an adoption or rejection decision.

How could such near-instant loyalty come about? One explanation is that
new brands in low-involvement categories are immediately included in the
consumer’s repertoire. The trial curve may then just reflect distribution
growth, the occasional nature of category purchases, and the shuffling of
purchases around the consumer’s repertoire.

This explanation remains somewhat unsatisfying. Intuition tells us that

consumers sometimes simply do not like a new product. They may buy it
once and then decide it tastes awful or does not work for their household.
The reliance of the near-instant loyalty hypothesis on average purchase
rates, rather than repeat purchase rates, excludes this effect altogether.

Repeat-purchase rate is therefore a more direct measure of loyalty, and has

greater face validity. Yet there is surprisingly little evidence on patterns in
repeat-purchase rates for new products. Eskin (1973) sought an early
indication of final repeat-purchase rates by separating out heavy buyers and
examining their repeat-purchase rates. Yet we know from Chapter 4 that the
top 20 per cent of buyers only account for 50 to 60 per cent of sales, and
CTA leads us to expect that heavy buyers will have different repeat-purchase
rates from light buyers, and that these purchase rates will regress to the mean
in subsequent periods. So Eskin’s approach, which has been highly
influential, seems incomplete.

A further difficulty is that early buyers are more likely to be heavy buyers;
Taylor (1977) demonstrated this, Fourt and Woodlock (1960) made a similar
assertion and Stern and Wright (2016) have confirmed it. These heavy
buyers may complete trial purchases and evaluations quickly. Therefore,
initial repeat-purchase figures may be misleading, as they combine heavy
buyers, whose repeat-purchase rate has settled down, with lighter buyers
who may still be making evaluative purchases. So the overall repeat-
purchase figure will keep changing until light buyers have finished making
evaluative purchases.

Fader, Hardie and Huang (2004) took a different approach. They adapted a
standard trial-repeat model to allow consumers’ purchase rates to be revised
after experiencing the new product. The result was an extremely accurate
forecasting performance for the single brand that they examined. This
suggests that their assumption, that purchase rates are revised slightly
following experience, is a reasonable one, although replication across other
data sets would give more confidence in their conclusions. Some indirect
support for Fader er al.’s approach is provided by Singh et al.’s (2012) study
of sales growth for 47 brand extensions in the UK. They found that, in the
third quarter after launch, those with a positive sales trend had an average
repeat purchase rate of 32 per cent, while those with a negative sales trend
had an average repeat purchase rate of 22 per cent. This difference is
consistent with a post-experience revision of purchase probabilities
suggested by Fader et al. (2004).

More work is needed on the evolution of repeat-purchase rates, for both

successful and unsuccessful new packaged goods. As a practical matter, it
would be helpful to have empirical studies that identify the typical evolution
of repeat-purchase rates across a wide range of conditions. This would
enable managers to benchmark the repeat-purchase performance of their new
brands against an existing knowledge base.
Summary
Markets undergo sales change for many reasons. First, time of year,
holidays, sales promotions and other events have a temporary effect.
Second, managers of mature brands may expand their product line, increase
distribution, or improve customer mindset metrics to achieve increases in
baseline sales. Third, a dynamic equilibrium underlies stable markets, with
buyers churning between brands or migrating between heavy and light
buying. Disturbances to this equilibrium may result in changes to market
share. However, changes often occur slowly because of the low rate of
erosion or switching and the long inter-purchase times of occasional buyers.

Fourth, the launch of new products is a major source of more rapid and
enduring change. Sociology helps us to understand the underlying forces,
while models such as the technology substitution model and the Bass model
provide tools for analysing and forecasting new product adoption. Fifth,
social forces may be different for low-value, frequently bought items. We
can capture the first purchase process in these markets with exponential
models (which assume little social influence). For repeat purchase, the
analysis is a little more complicated, due to early trial by heavy buyers, the
large number of light buyers and the unclear role of evaluative purchases.
One approach is to assume near-instant loyalty. Another is to use statistical
models that allow purchase rates to be revised after experience with the new
product. Recent evidence shows that differences in repeat-purchase rates for
succeeding and failing brand extensions can be detected in the third quarter
after launch.

In studying or managing markets, you need to know what fluctuations can

be expected, even in stable markets, and also how genuine change will play
out over time. From this chapter, we hope you have developed a good
understanding of the mechanisms that underlie changes in sales figures.
Additional Resources
The classic text on the diffusion of innovations is Rogers (2003). This gives
a good review of the diffusion literature and offers many interesting
examples, although it has little on alternative models. For an appreciation of
the technological substitution model, go straight to Fisher and Pry (1971),
which remains readable and interesting and has a number of case studies.
An overview of the Bass modelling literature can be obtained by reading
Bass (1995), Mahajan, Muller and Bass (1990) and Sultan, Farley and
Lehmann (1990). Finally, if you wish to know more about new product
development, there are many excellent textbooks in the area, such as
Crawford and Di Benedetto (2014).
Notes
1 The term churn is widely used in industry but, if the defection and
recruitment rates differ, which of these is the churn? We prefer to use
‘churn’ to describe the average customer replacement rate for all brands in a
category. Average defection and acquisition will be the same across the
category if the category itself has stationary sales.

2 A logistic function is a function of logits, or log-odds ratios. Odds ratios

are simply the ratio of possible outcomes. If a football team is expected to
win 80 per cent of the time, the odds of victory are 8:2 (or 4:1), and the
natural logarithm of this is ln(4), which is 1.39. The logit is widely used in
some other types of market modelling, such as the multinomial logit choice
model.

3 An alternative view is put forward by Du and Kamakura (2011), who

argue that weak contagion effects can be observed in new consumer
packaged goods, provided the modelling of such effects properly controls
for potential sources of bias.
6 Consumer Group Differences
Learning Objectives
When you complete this chapter you should be able to:

1. Identify consumer differences that are relevant and those that are less useful to
marketers.
2. Categorize cultures along the dimensions proposed by Hofstede.
3. Give examples of how these dimensions have been used in consumer research.
4. Understand the criticisms of Hofstede’s approach.
5. Describe how values can be used to understand fundamental consumer motivations.
6. Explain how gender and age are related to consumption.
Overview
Marketers need to take account of group differences when they design and promote
products, since what may appeal to one group may be of no interest to another. Group
differences may reflect local conditions or history and may be related to ways of viewing
the self and others in society. Some differences, such as individualism versus
collectivism, can be seen as orientations shaping consumer consciousness; these have
interested social scientists who seek to characterize pervasive features of whole societies.
Other differences relate to divisions within society and we focus on age and gender since
there are new developments here.

We start by asking what differences are relevant in consumer research and note that many
personality differences are hard to use because it is difficult to target people with a
particular disposition unless this is associated with physical segregation or differential
consumption of media. Deep-seated cultural differences are approached via Hofstede’s
(1980, 2001) five dimensions and Rokeach’s (1973) value system. We look at the use of
these ideas in consumer research and show that it may be quite difficult to derive
predictions from them. We also note that there may be simple social differences in
consumption that are best revealed by market research rather than academic study.
Contrasted with consumer group differences is the idea of convergence – that a process of
globalization is ongoing and that societies, at least in market practice, are becoming
similar. We explore this idea and look in particular at China, comparing it with the West.
On age, we focus particularly on the older members of society. The proportion of over-
65s is growing rapidly and represents a considerable burden on more productive, younger
groups. Though many are poor, much wealth is held by the older generation and recent
work has focused on the relatively conservative decision-making of older consumers. On
gender differences, there is evidence that women form more interpersonal bonds than
men, which affects their loyalty patterns.

In this book, we do not cover segmentation by brand. This is because there is a body of
robust evidence showing that there is generally little difference between the customers of
any two brands in a category. Brands do not usually differ enough to appeal to different
sub-groups in the population. Uncles (2011) summarized the empirical evidence on
comparisons of user profiles of directly competing brands. Through a series of replication
studies and extensions, it was confirmed that user profiles of directly competing brands
seldom differ. The study used data spanning 25 years, across 50+ categories and 60 data
sets, and confirmed lack of user segments for brands in emerging markets, private labels,
variants and composite segments. Similar findings are reported in other studies. For
example, in a study on Japanese buyers’ purchase intentions, Singh, Dall’Olmo Riley, et
al. (2012) showed demographic similarity across the users of top Japanese grocery
brands. Similar findings were also reported by Uncles and Lee (2006) amongst Australian
customers and in the UK by Singh, Ehrenberg and Goodhardt (2008).
Section 1: Relevant Differences for Consumer
Research
Segmentation, a fundamental notion in marketing, means grouping
consumers into discrete segments according to their similarities or
differences. The segments may differ in needs, consumption habits,
responsiveness to marketing actions, etc. Although it is useful to diagnose
differences between consumers concerning a specific product or issue with
dedicated market research, understanding the fundamental drivers that
account for differences across a broad range of behaviour may have more
potential if these can be directly addressed.

Socio-demographic characteristics were the basis of the first differences

examined in marketing. Some characteristics like social class and
profession have lost their appeal to marketers over time because the
significance of the social norms attached to them has faded. Other
characteristics have remained important but have taken on a different
meaning. Gender and age, for instance, still account for important
consumption differences, but in new ways, partly because of the changed
economic conditions. We address research linking these two factors to
loyalty in Chapter 2.

Personality traits are obviously an important determinant of consumption

behaviour. For instance, Venkatraman and Price (1990) show that a
consumer’s openness to innovation correlates with his or her impulsivity
and inclination towards social risk taking. This type of finding is not easy
for marketers to apply because if managers want to target consumers with
certain personality traits they not only have to identify them, they also need
to find ways to communicate with a specific segment without too much
spillover into other segments. Identification ideally requires a personality
survey, difficult to apply in practice, unless indirect indicators are available
(like the types of activities a consumer engages in). However, a person’s
dispositions and behaviours are affected not only by his or her personality
make-up, but also by the norms and beliefs of the cultural environment
(Triandis, 1989). Culture shapes personal perceptions, dispositions and
behaviours. More concretely, the cultural environment stimulates individual
dispositions that fit that environment and restrains those dispositions that
are not a good fit (Triandis, 1989). Hofstede (2001) talks about ‘mental
programs’ developed in early childhood in the family and reinforced later
on in schools and other organizations. Most research in this area studies the
influence of national culture and reviewing this work is the main topic of
the chapter. This nationally defined cultural environment is much easier to
identify for marketers and there is substantial research that examines its
influence on consumption behaviour.
Section 2: National Cultural Differences
Hofstede’s Five Dimensions
Analyses of cultural differences among people have often been influenced by
the seminal work Culture’s Consequences by Hofstede (1980, 2001). Hofstede
reports an analysis of more than 116,000 questionnaires filled out by IBM
employees from 50 subsidiaries around the world. This research on the
influence of national culture has become a classic for both academics and
practitioners. In an era of increasing globalization, Hofstede’s work drew
attention to the importance of cultural differences on attitudes to work in an
international organization.

In his analyses, Hofstede aimed to capture cultural differences between nations

in four dimensions: power distance, uncertainty avoidance, individualism–
collectivism, and masculinity-femininity. Each dimension was measured by a
number of items. For instance, the first dimension – power distance – was
based on three items, of which the central one was ‘How frequently, in your
experience, does the following problem occur: employees expressing
disagreement with their managers?’ An additional dimension, long- versus
short-term orientation, was added in a new book, a decade later. Some of these
dimensions had been around in the social science literature for some time but
Hofstede provided quantitative data on the differences in responses across
nations (see Table 6.1).
Exercise 6.1 Where would you place these
countries?
Hofstede presents the aggregate scores for individuals from each country in two-dimensional maps.
For this exercise, focus on individualism versus collectivism, and weak and strong uncertainty
avoidance, and assign two countries to each of the quadrants of Hofstede’s map below (eight
countries in total). In alphabetical order,these countries are: Canada, France, Great Britain, Hong
Kong, Italy, Mexico, Singapore and South Korea. (Answers in note 1 at the end of the chapter.)
A Critical Analysis
Hofstede produced a monumental study that was, and is, referenced by
virtually all those dealing with cultural differences. His work clearly has
implications for individuals working in another culture than their own and
for companies operating across different cultures. His findings have also
often formed the basis for consumer research on cultural differences, as we
will see later on in this chapter.

Although Hofstede’s work was very influential, it also met with criticism. A
fundamental concern was that he carried out an eclectic analysis of data,
based on theoretical reasoning and correlation analysis. In other words,
although his framework is based on hard data, he uses some creativity in the
analysis. McSweeney (2002) summarized this verdict in the title of his
article ‘A Triumph of Faith – A Failure of Analysis’. He also attacks the
assumption that one can extrapolate the results of IBM employees to an
entire nation. IBM employees may not be representative of their nation.
McSweeney thinks they may even be atypical. Moreover, the number of
relevant questions in the IBM questionnaire was limited, and they may not
be adequate to test Hofstede’s propositions fully. The questionnaire was not
designed to identify national cultures, and questionnaires designed for this
specific purpose by other researchers have produced different results (see
below). A third critique is that what Hofstede uncovered in the workplace
may be situationally specific and not transferable to other situations (e.g.
home, retail behaviour, recreational consumption). Fourth, again on the
method side, McSweeney objects to the way in which Hofstede sought to
validate his results by an analysis of historical and contemporary events,
and he claims that these illustrative stories ignore counter-evidence. The
2001 edition of the book also presents cross-validation with other studies
and claims that the findings are consistent with those from 140 other
studies. A further criticism comes from Ailon (2008), who argues that,
rather than capturing and mapping differences in societal values, Hofstede’s
work actually constructs reality and is therefore a product of a specific
cultural milieu and knowledge-producing tradition. It therefore reaffirms a
scheme in which western values are always the idealized reference point.
Hofstede anticipated this criticism; he realized that his personal value
system might influence the results and therefore outlined in the book his
personal position on various questionnaire items. Revealing this potential
bias does not of course eliminate it.

In this debate, there is a danger of stereotyping. On average, one population

may differ from another but plenty of people in the two populations could
still share the same values and react in the same way to innovations. Work
continues in this area; one recent paper by deMooij and Hofstede (2010)
reviews the studies and applies them to work on global branding and
advertising strategy.
The Universal Content and Structure of Values
It was Hofstede’s intention to analyse the values held by his respondents.
Rokeach (1973: 159–60), the key reference on this topic, defines values as
enduring beliefs ‘that a specific mode of conduct or end-state of existence is
personally and socially preferable to alternative modes of conduct or end-
states of existence’. These values are fairly general and Rokeach suggests
they are activated in a large variety of situations. As already implied in this
definition, Rokeach distinguishes between terminal values, referring to
desirable end-states that a person would like to achieve during his or her
lifetime, and instrumental values that are modes of behaviour to achieve the
terminal values. When taking the Rokeach Value Survey (see Table 6.2),
respondents have to order lists of values according to the importance each
has as a guiding principle in their lives. Using this instrument, those in
different cultures and sub-cultures can be compared.
 Exercise 6.2 The Rokeach Value Survey
Take the Rokeach Value Survey (Table 6.2). Arrange the 18 terminal values, followed by the 18
instrumental values, into an order of importance to YOU, as guiding principles in YOUR life
(Rokeach, 1973: 27). To what extent does your ranking relate to your consumption behaviour? How
do you explain divergences?

Source: Wikipedia (https://en.wikipedia.org/wiki/Rokeach_Value_Survey). Available under

Creative Commons Attribution-ShareAlike License: https://creativecommons.org/licenses/by-
sa/3.0/

It is interesting that the work on cultural values discussed here was done by
psychologists. Until the 1980s, culture was still thought of as ‘out there’ rather
than as an individual disposition (Triandis, 2004). Most psychologists therefore
held the view that cultural differences were a topic for anthropologists to work
on. A major turning point came with the publication of a review by Markus and
Kitayama (1991) that showed major cultural differences in cognition, emotion
and motivation. The authors proposed that the self-construal of interdependence
with other individuals guides people in many Asian cultures, whereas it is
independence that is the guiding construal in western cultures. These differences
manifest in people’s consumption behaviour and could influence brand-buying
behaviour.

The programmatic work of Schwartz and Bilsky (1987, 1990) provides a

comprehensive academic report on cross-cultural research into values. They
developed a theory of a universal psychological structure of human values and
tested it with an analysis of data from the Rokeach Value Survey. Their theory
received empirical support from 97 studies in 44 countries with 25,000
respondents. The analysis is an interpretation into groupings and mapping of all
the values, as shown in Figure 6.1. It turns out that the groupings are stable
across samples and they correspond to ten value types. The value types also
always have the same neighbours. In fact, the five value types that primarily
serve individual interests (power, achievement, hedonism, stimulation, self-
direction) form a contiguous region opposite another contiguous region formed
by the three value types that serve primarily collective interests (benevolence,
tradition, conformity). Universalism and security serve both types of interests
and are located on the boundaries between these regions. The relationship among
value types can also be summarized in terms of four higher-order value types
that form the bipolar dimensions of Figure 6.1:

Openness to change (following your own intellectual and emotional

interests) versus conservation (preserving the status quo and the certainty it
provides).
Self-enhancement (advancing personal interests, even at the expense of
others) versus self-transcendence (promotion of the welfare of others and
nature).

Figure 6.1 Relations among motivational types of values (Schwartz, 1992)

 Note: The dotted lines around the hedonism wedge signal that this value is
linked both to openness to change and to self-enhancement

Different societies share different values and value types, but not all
combinations are possible. The simultaneous pursuit of some value types is
possible while other combinations are incompatible.
Section 3: Cultural Differences in Consumer
Research
The research in cultural psychology discussed in the previous section has
influenced consumer research and marketing in two ways. First, it has
helped us to segment markets and to understand their differences so that
segments can be targeted with different strategies and marketing plans.
Second, it has aided researchers in testing their theories on how opposite
mindsets such as individualistic versus collectivistic affect consumer
behaviour. Cultural psychology has indicated that nations have different
cultural mindsets and that consumer researchers need to take this into
account when developing and advertising products, especially those
designed for export to other national markets. Below we give two examples
of these influences.
Cross-National Comparisons of Consumer
Psychology
Steenkamp and Geyskens (2006) examine how country characteristics
influence the perceived value that consumers derive from visiting websites,
using data from almost 9,000 consumers in 23 countries. In general, the
experience of visiting a website has a utilitarian (acquiring useful
information) component and an emotional component (the authors focus on
pleasure and arousal). According to the authors, consumer evaluations of
websites (in this case for consumer packaged goods) are driven by
perceived privacy and security, customization to the individual’s needs, and
congruity with the local culture. The central proposition of this research is
that the perceived value of the experience and the website characteristics
mentioned above are influenced by the characteristics of the consumer’s
country. The authors examine the influence of individualism/collectivism
and propose that when national-cultural individualism is high, the effect on
perceived value of the emotional experience, of the perceived privacy and
security protection, and of the customization, is great.

Even within a seemingly homogeneous cultural context, such as the Asian

culture, differences exist in how consumers perceive value in goods and
services. Shukla, Singh and Banerjee (2015) identify the underlying reasons
for the heterogeneity of Asian markets in luxury value perceptions, by
examining the value perceptions of consumers of luxury goods in India,
China and Indonesia. The authors test a structural model based on Karl
Popper’s three world conceptualization (1979), later developed by Berthon
et al. (2009). The results of their quantitative survey, conducted with real
luxury consumers in these three countries, identify variations in perceptions
of the symbolic, experiential and functional value of luxury brands. The
findings show that whilst both India and Indonesia are conventionally
categorized as collectivistic cultures, other-directed symbolism is a
significant influence on luxury value perceptions for Indian consumers,
reflecting the hierarchical nature of this society (vertical collectivism). Self-
directed symbolism, on the other hand, emerges as highly influential for
Indonesian consumers, demonstrating an equality orientation (horizontal
collectivism). Such evidence is crucial for marketers planning expansion
into Asian markets.

Consumer differences in innovativeness are explained by individual-level

and national-level differences in cultural values by Steenkamp, Ter
Hofstede and Wedel (1999). They used the Schwartz Value Survey to
measure individual values (Schwartz and Sagiv, 1995) and categorized the
11 European countries of the sample in terms of individualism, uncertainty
avoidance and masculinity by using an update of Hofstede’s earlier work.
Using Schwartz’s bipolar value dimensions, they hypothesize that a
consumer’s tendency to seek out new products and brands is impacted
negatively by conservation and positively by self-enhancement (see Figure
6.1). In terms of Hofstede’s cultural dimensions, individualism would be
expected to have a positive effect and uncertainty avoidance a negative
effect. Masculinity stands for assertiveness (versus nurturance) and should
have a positive effect on innovativeness. In this large-scale consumer
survey, all hypothesized effects were supported but the finding for self-
enhancement was not statistically significant.
Cultural Differences as Contrasts in Consumer
Research
Aaker and Williams (1998) examine the effectiveness of emotional appeals
in advertising. More specifically, they compare ego-focused (e.g. pride) and
other-focused (e.g. empathy) advertising appeals. The contrast between
individualist and collectivistic cultures in this work represents the effect of
self-perception (referred to as self-construal) that is either independent (a
focus on qualities of uniqueness) or interdependent (the self as inseparable
from others and from the social context). According to the authors, these
cultural differences lead to variations in the accessibility of certain types of
emotions. Ego-focused emotions should be more accessible in
individualistic cultures and other-focused emotions in collectivistic cultures.
This in turn should have a positive influence on the ability to process the
respective emotions when they are presented in persuasive messages. As a
test of their hypotheses, the authors presented a print ad for a fictitious new
beer with one of the two following appeals (Aaker and Williams, 1998:
245):

‘Acing the last exam. Winning the big race. Receiving deserved
recognition. Ohio Flag Beer. Celebrating life’s accomplishments.’
‘Reminiscing with old friends. Enjoying time together with family
during the holidays. Relaxing near the fire with best friends. Ohio Flag
Beer. Celebrating the relationships that matter most.’

They compared the effect of each appeal on two samples: Chinese students
(born and raised in China), as members of a collectivistic culture, and North
American students, as members of an individualistic culture. The results
showed a pattern opposite to the expectations: Chinese students showed
more positive attitudes towards the advertisement and brand when exposed
to the ego-focused emotions, as compared to the other-focused emotions,
while for American students the scores were reversed. As explanation, the
authors propose the novelty of the appeal as the driving factor. For members
of a collectivistic culture, ego-focused appeals are more unusual and
therefore trigger more attention. Similarly, the North American students
may have found the collectivist messages unusual and were therefore more
responsive to them. This idea was tested in a new study with ads for colour
film that showed either (individual) happiness or peacefulness. The results
confirmed that the novel thoughts provoked by an unusual appeal drove the
subsequent attitudes to the advertisement and to the brand.

Members of collectivist cultures are generally considered to be caring and

understanding. Does this also mean that they are easier to satisfy as
consumers and that they are more tolerant of service failures? A recent
review of the literature on cross-cultural services seems to indicate so
(Zhang, Beatty and Walsh, 2008). But Chan, Wan and Sin (2009) addressed
a different facet of collectivism: the need for attention and care. They
proposed that, in some situations, consumers from more collectivistic
cultures may be less tolerant of failure than those from individualistic
cultures. The criterion is whether the service failure is social (e.g. status and
esteem) or non-social (e.g. money and time). They developed a scenario for
each and exposed it to American or Chinese students (a between-subjects
design where each student sees one scenario). The students had to imagine
going to a restaurant where either the waiter does not smile, messes up the
order and does not apologize, or the restaurant has run out of the food they
selected and does not have the second choice available. The Asians were
indeed more dissatisfied with a social failure than the westerners and less
dissatisfied with a non-social failure. This research also examines the
factors that drive this effect. The Asians’ belief in fate explains their higher
tolerance of non-social failure. Their concern for face (i.e. a positive image
of self that is affirmed by interaction with others) makes them more
sensitive to failures in a social context.

Cultural background seems to affect customer perceptions when service

recovery is attempted following a failure. For example, in a cross-cultural
study, Hui and Au (2001) found Canadian customers to be concerned about
monetary compensation following a service failure, whereas Chinese
customers laid more emphasis on recovery procedures. Similar results were
reported by Mattila and Patterson (2004) in a study involving North
American and East Asian customers. In a scenario-based experimental
study on recovery fairness perceptions, Kalafatis et al. (2014) found that the
Thai hotel customers gave importance to employee behaviour and their
ability to empathize with the customer. These findings help service
managers to create culturally-sensitive and effective recovery strategies for
training their staff.
Box 6.1 Back-translation and other
problems
A particular challenge for cross-cultural research, be it academic or commercial, is the
translation of the measurement instrument. When questionnaire items are translated into
another language, there may be no exact way of rendering the same meaning. This
problem has been addressed by back-translating the translation into the original language
to see whether the original and the back-translation have the same meaning. Brislin
(1970) claimed that a functionally equivalent translation can be demonstrated when
responses to the original and target versions are compared and found to be near-identical.
However, it is pointed out that some bad translations that merely substitute terms (as in a
machine translation) are easy to back-translate, but may be poor in conveying the same
meaning.

It seems reasonable to argue that meaning differences that are introduced in translation
may be a basis for apparent cross-cultural differences. Such concerns are alleviated when
studies using different methods and measures converge on the same outcome – if they do.
Note that this problem is not necessarily removed by conducting all studies in English
since this restricts sampling to those familiar with English, and even in such sub-groups
there may be differences of usage.

Other problems in comparing across cultures include differential response rates, differing
responses to incentives and variation in the tendency to please interviewers.
Market-Relevant Facts
Broad cultural differences between nations are of great interest but
sometimes quite specific facts about difference in practice and thinking are
important. For example, when Levi 501s were relaunched in Europe by
BBH and McCann, they argued that jeans were perceived differently in the
USA and Europe and, because of this, the existing US advertising was
inappropriate for Europe (see Box 6.2).

There are other cases where cultural differences have particular forms that
marketers should know about. For example, it is common practice on the
Indian Subcontinent to stay with relatives. Indians and Bangladeshis have
been heard to complain about the discomfort that occurs when relatives
squeeze up to make room for a visitor. Sometimes the visitor would prefer
to stay in a hotel but this would cause offence. Hotel firms in the West that
seek to expand in other countries need to know about such practices.
Globalization
Up until now we have stressed the cultural differences among consumers
living in different regions of the world. However, with the globalization of
markets, these differences may be diminishing and the debate of
globalization versus localization has concerned international marketing
practitioners in the past few decades. Advocates of globalization stress a
convergence of consumer attitudes and behaviours across geographical
boundaries and try to identify and reach consumer segments across
boundaries with common products and marketing programmes. In this
context, three Dutch researchers segmented European consumers from 11
different countries using a means–ends chain survey (Ter Hofstede,
Steenkamp and Wedel, 1999). A means–ends chain examines how
consumers connect to a product by asking questions about the link between
attributes (e.g. organically produced, low fat) and benefits (e.g. spending
less money, good taste) and the link between benefits and values (e.g. self-
respect, warm relationships). They identified four cross-national segments
that were then linked to descriptive data on socio-demographics, product
consumption and media usage, and information on personality and attitudes.
 Box 6.2 A small difference in perception
When Bartle Bogle Hegarty and the McCann agencies handled the relaunch of Levi 501
jeans in Europe in 1985, they refused to use the US advertising (Feldwick, 1990). Their
analysis was that, in the USA, jeans were workaday clothes and increasingly old-
fashioned, but in Europe wearing jeans could still be a fashion statement. Moreover, 501s
had genuine provenance and were worn by a small number of opinion leaders. Thus, the
agencies sought to re-establish the slightly baggy 501s as ‘the right look’ for young
people, the definitive classic jean. To do this, they needed different advertising.

A set of commercials was developed of which the most famous was ‘launderette’,
featuring Nick Kamen sitting in a launderette in his boxer shorts as he waited for his 501s
to wash; the music backing was Marvin Gaye’s ‘I Heard It Through the Grapevine’. It is
still to be found on YouTube if you search using ‘Levi’ and ‘launderette’. These ads
multiplied Levi’s sales twenty-fold in the next three years at a higher price, and it was
claimed that even sales of boxer shorts increased.

The point of this example is that the successful strategy related to a difference in
perception about jeans – hardly a matter of cultural difference but immensely important to
the success of the advertising. Such differences of perception are more a matter of market
research than cross-cultural investigation, but they cannot be ignored by marketers.

Marketing in China
China, as well as being the home of a large part of the world’s population,
has shown a spectacular rate of GDP growth. In 2010 it was 10.4 per cent
(World Bank). Despite concern about the decline in China’s growth in
recent years, 6.9 per cent was reported for 2015. Meanwhile, western
markets have recorded low or negative growth.2 Because of its importance,
we have compared China with western markets. How different is China?
Although Chinese consumers buy different foodstuffs from the West, many
other aspects of purchase are very similar and the Chinese are purchasing a
large number of luxury brands from the West, from perfume to whisky.
Uncles (2010: 70) believes that the main theme is convergence: ‘Significant
aspects of the Chinese retail landscape now conform to what might be
described as an international norm and, superficially at least, consumer
attitudes and behaviours appear to be more alike.’ Uncles (2010: 70) notes
that ‘retail formats, institutions, infrastructures, and management practices
are becoming similar to those seen in international markets’. In grocery
retailing, hypermarkets and supermarkets are rapidly growing and took 62
per cent of grocery sales in 2008, a figure not far behind that of many
western countries. By comparison, traditional retailing such as wet markets
is stationary. Uncles notes that much of the modern retail expansion is
driven by international companies such as Carrefour and Tesco, who often
work in partnership with Chinese companies. This speeds up the transfer of
western business practice.

Chinese consumers accept the new pattern of retailing. One reason for this
is environmental determinism. In Chapter 1, we noted that behaviour is
moulded by the conditions under which people live; the environment
defines what actions are possible and rewards and punishes behaviour
according to how it fits the environmental conditions. It is hardly surprising
that attitudes and values develop that broadly endorse a retail system that is
well established, and it would be naïve to assume that values always come
first and systems are developed to be consistent with them. This control by
the environment is enhanced when urban renewal projects replace the
traditional market with a western format. Uncles notes that more traditional
retailing practices persist in China but points out that there are
corresponding traditional forms in the West.

Uncles, Wang and Kwok (2010b) studied brand performance metrics to see
whether patterns familiar in the West are also found in China. As elsewhere,
they found the double jeopardy effect with toothpaste and soy sauce in two
different cities, Shanghai and Xi’an, which was consistent over five years.
They also showed that multi-brand loyalty was the norm, just as it is in
Europe, Australasia and the USA. In two other studies, Uncles and Kwok
(2008, 2009) examined patterns of category purchase by store. Again, the
Dirichlet patterns were found: what people buy may differ but how they buy
fits patterns of behaviour found in the West.

In another study, Thøgersen and Zhou (2012) investigated the take-up of

organic food. They found that Schwartz’s universalism value lay behind the
adoption of organic food in China, as in the West. Furthermore, buying
organic food was strongly linked to beliefs about health, taste and care for
the environment in both the West and China. Thus, this research suggests
that globalization has proceeded fast in China, and that in many respects
there is convergence in the consumption patterns of China and the West.
Section 4: Age and Gender Differences in
Consumption
Age
The present scale of population ageing is ‘unprecedented, pervasive,
enduring and has profound implications for many facets of human life’
(United Nations, 2002). The post-retirement proportion of the population is
rising and many members of this group are becoming richer. Euromonitor
(2011) reports that the percentage of the world’s inhabitants over 65 has
moved from 5.9 in 1980 to 8.0 in 2011. In 2014, 18.5 per cent of the
population of the European Union (28 countries) was over 65 and in Japan
the figure was 26 per cent. Moreover, the current over-65s are closely
followed by an even larger group of those born between 1946 and 1964 (the
‘babyboomers’) who are now beginning to retire; these people also vary
widely in their spending, social commitments and lifestyle and are,
collectively, the wealthiest group of older people in history (Euromonitor,
2006). Marketers and policy-makers must take account of these changes and
understand them because they will change economic priorities. Central to
this understanding is how older consumers buy (Uncles and Lee, 2006).

Market research has focused on the goods and services that are typically
bought by consumer segments. However, to explain preferences among older
consumers, it is necessary to study how the process of decision-making
varies with age. So far, this aspect of decision-making has focused mainly on
variations in cognitive and affective processes as people age (Lambert-
Pandraud, Laurent and Lapersonne, 2005; Cole et al., 2008; Drolet, Schwarz
and Yoon, 2010; Lambert-Pandraud and Laurent, 2010).

Lambert-Pandraud et al. (2005) examined the impact of age on loyalty. They

analysed the answers of 30,000 respondents to a survey of recent new-car
buyers who had replaced a previous car. The survey covered the purchase
process of both the recently acquired car and the previous car. Because a car
purchase is one of the most involving consumer purchases, one would expect
consumers to engage in at least some information search and comparison. In
fact, 21 per cent of consumers between the ages of 60 and 74 considered
only one brand, and among consumers aged 75 or over, 27 per cent were in
this group. Many respondents purchased the same make of car and, as age
rose, so did the repeat-purchase rate, shown in Figure 6.2. The effect of age
remained strong in a multivariate model that includes contextual variables
and other socio-demographics. Older car-purchasers showed other
characteristics. They looked at fewer car models when they considered more
than one, they dealt with fewer dealers, and when they did change the make
of their car they were likely to buy a familiar national brand. They were also
more prone to defer purchase, a matter that should interest marketers (and
governments when economies are stagnating). Taking all these facts
together, we can see that the decision-making of the older consumer can be
described as conservative.

For age-based behavioural loyalty, however, several studies contradict the

fact that older buyers may differ in their brand purchase behaviour.
Anderson and Sharp (2010), for instance, highlight similarities in the brand
loyalty patterns of older and other age-based segments. These results are
consistent with findings in the USA (Uncles and Ehrenberg, 1990), the UK
(Ahmad, 2002), Australia (Uncles and Lee, 2006) and Japan (Singh,
Dall’Olmo Riley, et al., 2012), where the buying propensities of older
customers were found to be similar to those of other age groups. These
findings indicate that in certain fields, such as groceries, there may be fewer
than expected differences between age groups in terms of brand purchase
behaviour.

Figure 6.2 The repeat purchase of cars by age group (from Lambert-
Pandraud, Laurent and Lapersonne, 2005)
The conservatism of older buyers could relate to cognitive decline, which
increases the cognitive effort in decision-making and pushes the older person
towards decision heuristics such as choice repetition or purchase deferral
(Phillips and Sternthal, 1977; Lambert-Pandraud et al., 2005). However,
there is evidence that it is the speed rather than the ability to process
information that is measured in tests of cognitive process (Roedder-John and
Cole, 1986; Cole and Houston, 1987). This suggests that older people may
have the capacity to decide effectively but make less use of this capacity
because of their slower processing speed.

A second explanation is socio-emotional selectivity: that older persons tend

to have more activation of emotional centres leading to a focus on affective
information as age advances. This leads to more attention being paid to
established emotional contacts and less to gaining new information, which
results in older consumers having fewer social interactions and concentrating
on the people they know well (Castensen, Isaakowitz and Charles, 1999).
This could explain the smaller choice set of car brands/dealers among older
people found by Lambert-Pandraud et al. (2005).
A third mechanism producing conservative behaviour is change aversion
(Wallach and Kogan, 1961). Change aversion leads to purchase deferral and
draws decision-makers toward more familiar options where risks may be
lower. However, such change aversion may relate to a justifiable cynicism
about brand differences that grows with experience and therefore age. Risk
aversion was studied by Simcock, Sudbury and Wright (2006). They found
that older buyers of cars were more likely to express risk-related concerns.

These three explanations for age-related conservatism suggest a fairly

inexorable process of change as people age that is difficult to influence.
However, it is possible that this work has focused too much on individual
processes and not enough on the social element in consumer decision-
making. We know that many decisions occur because advice from others
draws attention to new alternatives and pushes the individual to think
through the possibilities. To explore this, East, Uncles and Lomax (2014)
examined a set of studies on word of mouth (WOM) to see whether there
were any age-related differences in volume of word of mouth received.
These studies covered bank accounts, cameras, computers, holidays, credit
cards, and two studies each of mobile phones, mobile phone airtime, coffee
shops, restaurants and supermarkets. The results were startling and are
shown in Figure 6.3.

From Figure 6.3, we can see that word of mouth about brands received by
respondents falls sharply with age. On average, respondents aged 65 and
over receive 1.5 instances of WOM compared with 3.9 by those aged 25.
Men tend to report receiving more WOM than women, except in the 65+
segment. A number of explanations could fit these findings. The older
respondents might recall fewer instances of advice because of cognitive
decline, but against this the effect is fairly continuous over the age range,
and seems to start before the age when cognitive decline is believed to begin.
Also, retirement, the death of friends and the departure of children from the
household – all leading to reduced social contact – are consistent with the
decline in advice observed. The same research also examined the period of
time that respondents had owned their current brand in seven categories and
this period jumped to nearly twice the level for post-65-year-olds. Using
regression analysis, the researchers examined whether the volume of word of
mouth received was related to the length of time that the current brand had
been owned (which was seen as evidence of purchase deferral). It was found
that the greater the volume of positive advice received, the less the purchase
deferral.

Figure 6.3 Average sum of received positive and negative WOM about
brands by age and sex (from East et al., 2014b)

This evidence therefore suggests another explanation for conservative

behaviour by the older consumer: that compared to younger consumers they
have lower levels of social influence. If this is so, the older consumer may be
induced to consume more if marketers can influence those who give advice,
such as their children, or if the deficit in advice can be covered by increased
advertising designed for this age group. There is clearly more to discover
about differences in age. We have not discussed how the importance of
different categories changes as people get older; older people may process
decisions like younger people when the issue is more important to them, for
example when it concerns finance, groceries or medical aid.
Gender
It is a common stereotype that men show lower levels of sexual loyalty than
women. How about consumer loyalty? Melnyk, van Osselaer and Bijmolt
(2009) make the distinction between loyalty to people and to companies.
They present a series of studies indicating that female customers are more
loyal to individuals, whereas male customers are more loyal to groups and
firms. They cite work (Cross and Madson, 1997) which shows that in
western cultures, women, more than men, want to connect with other
people and maintain existing relationships. Thus, women focus on
interdependence. One view is that men see themselves as independent and
focus on uniqueness and individuality, but Melnyk et al. also discuss the
idea that both women and men are interdependent, though in different ways.
In their account, women attach more importance to close relationships with
specific individuals, while men focus on relationships with larger and more
abstract groupings of people. Thus, men are interdependent too but their
interdependence is with groups rather than individuals.

Applying this thinking, Melnyk and her co-authors (2009) propose that
women may be more loyal to individual employees, while men are more
loyal to companies/brands. In a scenario study, the authors present
participants with a situation where instead of going to the closest bakery for
a birthday cake, there is the option of going to a bakery owned and run by
somebody they went to high school with. In an alternative condition, the
word somebody is replaced by a group of people. In subsequent studies, the
authors asked participants directly whether they were more loyal to
individual service providers or to companies in different categories for their
actual consumption behaviour. Across these studies, they show that women
do not always exhibit more consumer loyalty than men, but their loyalty
does indeed extend to individual employees, while men show more loyalty
to companies. They also found evidence that these types of loyalty are
driven by the quest for relational versus collective interdependence, for
women and men respectively.

The difference in loyalty between men and women found by Melnyk et al.
(2009) could be underpinned by genetic difference or adjustment to social
norms. The generalizability of the findings will depend on which
explanation applies; if the difference relates to role differences that are
widely found in western economies, we might find varying patterns of
loyalty in other societies where the role relationships are different. This
would mean that the loyalty difference is more cross-cultural than inherent
in genetics. From the findings of Melnyk et al., we might expect women to
talk more about categories where interpersonal relationships are common.
Some unpublished work at Kingston University indicates the product fields
that women talk about to others more than men. Men talk more about
mobile phones, airtime suppliers for mobile phones, computers, credit cards
and bank accounts, while women talk more about holiday destinations,
luxury brands, supermarkets, restaurants and coffee shops. These
differences fit gender roles. Women may give more advice about
supermarkets because they do so much more food shopping than men
(Chang-Hyeon, Arentze and Timmermans, 2006), yet women also use
phones more than men, according to Nielsen (2011b) in evidence from the
USA, and despite this men appear to talk more about them. Perhaps devices
like phones, being impersonal, attract more interest from men. Whatever the
bases for these differences, they are of importance to marketers and ad
agencies in making decisions about the targeting and content of
communications.
Summary
The cultural environment in which an individual is raised and lives shapes
his or her perceptions, dispositions and behaviour. National cultural
differences have been studied extensively, largely under the influence of
seminal work by Hofstede. Hofstede categorizes countries on the level of
power distance that is accepted by their members, their tolerance for
uncertainty, the degree to which economic achievement versus caring for
other people is valued, the degree to which the delay of gratification is
accepted, and the degree to which individuals are supposed to look after
themselves or remain integrated in groups. This last dimension, referred to
as individualism versus collectivism, has been much used in consumer
research. An excellent mapping of all relevant cultural differences is
provided by Schwartz and Bilsky (1990); based on the Rokeach Value
Survey, they identify ten value types.

As examples of how consumer researchers have used these classifications

of cultures, we described how the perceived value of visiting websites and
consumer innovativeness are influenced by differences in individualism–
collectivism. We also examined differences between American and Chinese
consumers in their sensitivity to ego-focused and other-focused appeals in
advertising and in their tolerance to service failures.

National differences between consumers continue to influence marketing

communications and service format, but at the same time we can observe
convergence. We illustrated this with an international segmentation study
and a study of consumption behaviour in China.

Understanding consumer group differences is useful for market

segmentation. Behavioural differences between individuals and groups are
often easy to observe, but understanding the fundamental drivers of this
behaviour has broader application potential. We examined the reasons for
increased brand loyalty by older consumers. We also compared the loyalties
of men and women and found that men are more loyal to companies while
women are more so to individual employees.
Further reading on cross-cultural studies
This chapter has examined the most relevant consumer differences that have
also received research attention. For the reader specifically interested in
cross-cultural issues from a marketing practitioner’s point of view, we
recommend the following handbooks:

de Mooij, M. (2010) Global Marketing and Advertising: Understanding

Cultural Paradoxes, 3rd edition, London: Sage.

de Mooij, M. (2011) Consumer Behavior and Culture: Consequences for

Global Marketing and Advertising, 2nd edition, London: Sage.

Usunier, J.C. and Lee, J. (2011) Marketing across Cultures, 5th edition,
Upper Saddle River, NJ: Prentice Hall.
Notes
1 Answer to Exercise 6.1:

2 Some of the spectacular growth in consumption in China is attributable to the

one-child policy in the country, now being eased. Demographers have drawn
attention to the resources that have to be spent on servicing an increasing
population. If the population increases by 2 per cent each year, so must the
schools, firms and other social institutions if the new generation is to have the
same opportunities as before. This is called the demographic investment and it
holds back development. The problem is made worse by the fact that a growing
population contains a larger proportion of children who do not assist the
economy and need the attention of adults, thus further restricting economic
activity. China has avoided the demographic investment. The one-child policy
has the eventual consequence that the ratio of old to young shifts and a large
older population has to be supported by a smaller workforce. In India,
population growth has continued at a rate of 1.6 per cent per year (Rosenberg,
2011) and this will hold back the expansion of individual wealth.
Part 3 Explaining Decision-Making
7 Predicting and Explaining Behaviour
Learning Objectives
When you have completed this chapter, you should be able to:

Define attitude, belief and intention, and explain how these concepts are measured.
Understand the expected-value theory of attitude applied to products.
Report on the research linking behaviour to attitude and intention.
Describe the theory of planned behaviour, its applications, strengths and weaknesses.
Understand the problems of predicting behaviour.
Overview
The cognitive approach to consumer behaviour relates consumers’ attitudes, beliefs and
intentions to their behaviour. Among the theories used, the theory of planned behaviour
(TPB) has been most successful, although a number of problems with this theory remain.
We first examine the nature and measurement of attitudes, beliefs and intentions and their
relationship with behaviour, and then we explain how the TPB can be used to predict and
explain behaviour. This has a clear relevance to marketing: we need to predict purchase
and to understand why people prefer one brand to another if we are to create products in
the right quantity and of the right quality.
Section 1: Definitions and Measurements
Attitudes are what we feel about a concept, which may be a brand, category,
person, theory or anything else we think about and attach feelings to. An
important class of concepts is actions, particularly commercially relevant
behaviour such as buying, renting, using and betting. We focus on attitudes
to such actions because it is these attitudes that best predict the actions.
Thus, it is the attitude to playing the National Lottery or buying a smart
phone that most interests us. Compared with the attitude to the action, the
attitude to the object (the National Lottery or the smart phone) is less
directly related to the action.

A person’s attitude may be inferred from his or her actions or measured

using a systematic questioning procedure. According to the mental
representations model, introduced in Chapter 3, a concept is a node linked to
other nodes. The linkages to other nodes can be seen as beliefs about the
central concept. When the concept is an action, the beliefs often concern the
outcomes of the action. If I play the National Lottery, there are outcomes
such as dreaming of untold wealth, being excited by the draw and, usually,
being disappointed by the result. Such beliefs have an outcome likelihood
(or belief strength) and an evaluation which can be measured by the scales
below:

If I play the National Lottery I will be excited by the draw:

Being excited by the draw is:

The first scale measures the likelihood of the outcome, while the second
measures the value of the outcome if it occurs. These seven-point measures,
called semantic differential scales, were first used in research by Osgood,
Suci and Tannenbaum (1957). We usually denote the likelihood measure as b
(for belief) and the evaluation measure as e. The full outcome measure is the
product of b and e, which we call the expected value of the outcome. An
expected value can be negative as well as positive because the evaluation
can be negative.
The Expected-Value Theory of Attitude
Most of the alternatives from which we choose are multi-attribute. To
assess the value of going to Wales for a holiday, I have to take account of
weather, cost, travelling effort, food, opportunities for recreation, etc. Using
the method above, we can measure an expected value for each outcome,
and my overall (or global) attitude to going to Wales for a holiday is given
by the sum of the expected values. So, if A is the global attitude:

A = b1e1 + b2e2 + b3e3 + ...

or A = Σ biei

Rosenberg (1956) pioneered this approach in attitude theory and Fishbein

(1963) tested the relationship by separately measuring the global attitude, A,
and the sum value, Σbiei. If the theory is correct and A is related to Σbiei,
then subjects with high scores on one measure will have high scores on the
other. Thus, by correlating respondents’ scores on the two measures, we can
find out how much A is related to Σbiei. Fishbein found a correlation
between the sum score and the global measure of 0.80, which gave strong
support to the idea that global attitudes are based on the sum of the
expected values of the attributes. Fishbein’s expected-value treatment of
attitude has been confirmed in a large number of subsequent studies, though
the correlations are generally lower than 0.8 (typically, in the range 0.4 to
0.6).

Fishbein’s treatment of attitude assumes a process of compensation: for

example, that the unspoilt beaches can offset wet weather in Wales. At best,
compensation is likely to be partial. Just taking account of the main
outcome of one alternative requires some thought, and when several
outcomes are involved the assessment is obviously more complicated. As
noted in Chapter 1, extended thought before choice is a rarity but we
probably consider more attributes when important decisions are taken.
People choosing between several options usually take the one with the
largest expected value. Edwards (1954) described this as the subjective
expected utility (SEU) model of decision. This way of thinking about
decisions treats any product as a bundle of expected gains and losses.
Modal Salient Beliefs
Fishbein’s theory of attitude is about what individuals think and feel, but it
has to be tested on groups of people and each member of the group may
have a somewhat different basis for their attitude. To take account of this,
some studies have asked each person separately about the attributes that he
or she thought were important – for example, Budd (1986) on cigarette use
and Elliott and Jobber (1990) on company use of market research. This
technique increases the observed association between global and sum
measures but is laborious. Fortunately, on many issues there is substantial
agreement between people on the factors that are important, and the same
questionnaire can be used on all respondents with only a modest loss of
precision.

To establish the commonly held beliefs about a concept, it is necessary to

perform an elicitation, which is described in Exercise 7.1. This is a series of
questions about the positive and negative associations of the concept which
are put to members of the target group. The beliefs that come easily to mind
are recorded and those that occur frequently in a group, called modal salient
beliefs, are used for the questionnaire. In an elicitation, the questioning
should be low pressure. Fishbein and Ajzen (1975) (‘Ajzen’ is pronounced
‘Eye-zen’) argue that beliefs which have to be dredged up from the recesses
of the mind are unlikely to have much effect on behaviour. Exercise 7.1
covers not only the gains and losses of a prospective action, but also factors
discussed later in the chapter – the influence of other people and the
personal and environmental factors that make the action easier or more
difficult to perform.
Exercise 7.1 Eliciting salient beliefs
Define the action clearly. For example, ‘buying Snickers’, ‘getting a new computer’,
‘giving blood when the blood transfusion service comes to the campus’.
Define clearly the target group. For example, you might be particularly interested in
children buying Snickers or in older computer buyers.

Elicit salient beliefs. In a sample of people from the target group, ask each person
questions about the advantages and disadvantages of the defined action. After each
response, prompt with ‘anything else?’ but do not press hard for ideas. Record the
responses for each person. A typical encounter might be:

Q. Can you tell me what you think are the advantages of getting a new computer?

A. You get a lot for your money now.

Q. Anything else?

A. Probably more reliable.

Q. Anything else?

A. Not really.

Q. Can you tell me what you think are the disadvantages of getting a new computer?

A. It will have to be set up with the right programs.

Q. Anything else?

A. I’ll have to get used to new versions of Word and PowerPoint.

Q. Is there anything else that you can think of about getting a new computer?

A. No.
Consider the negative action. Certain actions may have different salient beliefs
associated with not doing the action. For example, ‘not having children’ and ‘not taking
drugs’ may be seen as actions with their own rationale and are not just the opposites of
having children and taking drugs. For such negative actions, it is wise to also elicit
salient beliefs about the negative action.
Be aware of salient referents. Ask each respondent in the sample whether there are
people or groups who think that the respondent should do the defined action. Repeat
with ‘should not’. Ask if there are other people or organizations that come to mind when
they think of the action. Use the prompt ‘anyone else?’ but do not press for responses.
(Recent work has often included measures of what salient referents think the respondent
will do as well as should do.)
Think about control factors. Ask each respondent about conditions that make the
action easier or harder to perform. Again, prompt with ‘anything else?’
 Refine the list of beliefs. Combine similar beliefs. Compile a list of modal salient
beliefs using the ones most frequently mentioned. The decision to include a belief
depends on the frequency with which it is mentioned and the time and money available
to support the research. When the questionnaire is intended to be used both before and
after exposure to advertising or the product, it is important to include beliefs that may
become salient as a result of this exposure.

After similar responses have been grouped together, the list of modal salient
beliefs is usually quite short. Complex issues, such as getting married or
buying a house, may have about ten salient beliefs relating to attitude;
simpler issues, such as buying specific chocolate bars, will have fewer.
Do Attitudes Predict Action?
Our interest in attitudes is partly based on the belief that they predict behaviour.
Following Allport (1935), an attitude is usually seen as ‘a preparation or
readiness for response’ and thus should be a predictor of behaviour, except
when freedom of action is restricted. However, when Wicker (1969: 65)
reviewed 47 studies on attitudes and behaviours, he concluded that: ‘It is
considerably more likely that attitudes will be unrelated or only slightly related
to overt behaviors than that attitudes will be closely related to actions’.

Schuman and Johnson (1976) suggested that other unreported variables affect
behaviour in addition to attitude. This was supported in work by Fishbein and
Ajzen (1975), Ajzen and Fishbein (1980) and Ajzen (1985, 1991). These
researchers showed that, in addition to attitude, behaviour is controlled by
beliefs about the wishes of people and groups important to the respondent, and
by beliefs about the way personal ability and the environment can promote or
restrict behaviour.

In addition to the ‘other variables’ explanation for the poor prediction of

behaviour, Ajzen and Fishbein (1977) pointed out that researchers frequently
measure the wrong attitude. The correct attitude for predicting behaviour is the
attitude to that specific behaviour. Fishbein and Ajzen (1975: 360) concluded
that ‘many of the studies that have been viewed as testing the relation between
attitude and behaviour are actually of little relevance to that question’. Thus, if
you want to predict the quitting of smoking, it is the attitude to quitting
smoking and not the attitude to cigarettes, or even smoking in general, that
should be measured. This lack of compatibility between the attitude and
behaviour measures is neatly demonstrated by an unpublished study conducted
on 270 women by Jaccard, King and Pomazal (reported by Ajzen and Fishbein,
1977). In this work, three attitudes relating to birth control were measured and
correlated with the use of birth control. As the attitude comes closer to the
specification of the behaviour, the correlation rises (see Table 7.1). This effect
is readily explicable if we think of the motivations of different women. For
example, a woman who wanted to become pregnant would neither use the pill
nor be positive in her attitude to using it (giving a high correlation between her
behaviour and her attitude), but she might still be positive about the pill and
birth control in general (giving a low correlation with her non-usage of the
pill). Similar results were obtained in another study of the correlations between
attitudes to ‘religion’, ‘church’, ‘attending church this Sunday’ and actual
church attendance this Sunday.

Another example of using the wrong attitude takes the form of trying to predict
what people will do from measures of past satisfaction. As we saw in Chapter
2, satisfaction can be a poor predictor of future behaviour because people can
be positive about their past experience with a product without necessarily
wanting to use it in the future. Needs change and sometimes products change
so that what was satisfactory in the past may not be expected to be satisfactory
in the future. A somewhat better prediction of retention would be obtained by
using the attitude to ‘buying the product again’.
Specifying Measures
From this work, it is clear that the more compatible the measures of attitude
and behaviour, the more they will correlate. Compatibility is specified by
target, action, context and time (think TACT). In the case of oral contraceptive
use, the target is the oral contraceptive, the action is using it and the
context/time is implicit in its use. In other cases, the context and time could be
more important. For example, I might avoid shopping in my local supermarket
on a Saturday morning because the store is so busy at that time. In addition to
the TACT variables, it is important to ensure that respondents are talking about
their own attitudes and behaviour rather than some general idea. Ajzen and
Fishbein (1977) applied these compatibility criteria in a meta-analysis of 142
attitude–behaviour associations. They sorted the studies into those with low,
partial and high compatibility between the measures and sub-divided the last
group because some measures were not clearly specified. Table 7.2 shows their
findings. It is clear that low compatibility between attitude and behaviour
explains why many previous studies showed a weak connection between
attitude and behaviour.

Another measurement challenge occurs when attitudes embrace a set of

behaviours rather than one specific behaviour (Ajzen and Fishbein, 1977). For
example, a measure of a person’s attitude to protecting the environment might
give a rather low prediction of their bottle-recycling behaviour because specific
factors may affect the decision to recycle bottles. If a multiple-act measure of
environmental behaviour is constructed that also includes use of recycled
paper, use of low-energy bulbs, installing insulation, recycling of metals and
newsprint, donations to environmental groups, refusal to buy tropical
hardwoods, boycotting the products of environmentally suspect firms,
criticizing women in fur coats, etc., we would expect this measure to have a
stronger correlation with the attitude to protecting the environment. This is
because the specific factors affecting single actions have little effect in the
combined measure compared with the common theme of helping the
environment. Consistent with this, Weigel and Newman (1976) found that the
attitude to environmental preservation correlated better with a multiple-act
measure of environmentally concerned behaviour than with single measures.
In consumer research, the compatibility principle means that attitudes to the
purchase, hiring and so on of the product must be measured if it is those
actions that we want to predict. This simple lesson about using compatible
measures has not been well learned in marketing. Often, attitudes to the brand
are studied rather than attitudes to purchasing the brand. Often, there is
substantial overlap between these measures but, as Ajzen and Fishbein show
(1980: Ch. 13), this is not always so. Many of the early studies reviewed by
Wicker (1969) in this field used incompatible measures, thus explaining the
low association that he found between attitude and behaviour. The notion of
compatibility has much improved attitude research.
 Exercise 7.2 Compatibility test
Circle the numbers that are closest to your feelings.

Swimming is:

Swimming in my local swimming pool is:

For me, going swimming in my local swimming pool over the next month is:

I will go swimming in my local swimming pool over the next month:

In Exercise 7.2, we are not able to measure behaviour so a measure of

intention, the last scale, is used as a proxy. The scores of students in a class can
be entered on a data file and the correlations between the last intention
measure and each of the other measures calculated. You should find that the
more compatible a concept is with the intention measure, the higher the
correlation between the measures.
Purchase Intentions
Intentions may predict behaviour but do not tell us why the behaviour is
undertaken; for this, reasons are required. However, sometimes in
marketing, a prediction of behaviour may be all that is needed.

Purchase behaviour may be predicted either from stated intention or from a

person’s estimate of their purchase probability. Measures of intention have
been well tested in the field of consumer durable purchase. In research in
this area, Pickering and Isherwood (1974) found that 61 per cent of those
who said they were ‘100 per cent likely to purchase’ actually did so; this
compares with the 5 per cent of respondents who made a purchase, even
though they had expressed no intention to purchase the durable in the next
12 months. These findings were close to those obtained by Gabor and
Granger (1972) in a similar study.

Since intention-to-buy measurements can discriminate quite well between

prospective buyers and non-buyers, it is possible to compare prospective
purchasers and non-purchasers in the same way that users and non-users are
compared. In this way, it should be possible to find out what makes
products attractive before they are bought. Marketers can use this
knowledge to adjust the product and/or focus promotion on the factors that
buyers think are important.

There are two reasons for discrepancies between predicted and actual
purchase: first, the true probability of purchase may be inaccurately
measured by the scale point checked by the respondent, and second, people
may change their intention or be unable to fulfil it (Bemmaor, 1995). The
second inaccuracy is difficult to avoid but the first type of discrepancy is
reduced by improved scaling. Juster (1966) used an 11-point, verbally
referenced scale to measure the likelihood of purchase (see Box 7.1). In a
review of intention measurement, Day et al. (1991) argue that the best
results are obtained using a Juster scale, and Wright and MacRae (2007)
show that the predicted purchase proportions obtained using the Juster scale
were unbiased estimates of the actual purchase proportions found from
panel data or purchase recall.
 Box 7.1 The Juster scale
This is an 11-point scale with verbal descriptions and probabilities associated with each number:

In planned behaviour research, the seven-point semantic differential scale has

usually been used (rather than the 11-point Juster scale described above). The scale
is used to measure either the intention or the self-prediction of the respondent that
he or she will perform the behaviour. Often, there is little difference between the
two measures but self-prediction seems likely to be more accurate because it may
take more account of conditions that may prevent action. For example, people may
intend to give up cigarettes but be more realistic if asked to estimate the likelihood
that they actually will give up. However, Sheppard, Hartwick and Warshaw (1988)
reviewed a large number of attitude–behaviour studies and found only marginal
superiority for self-prediction over true intention measures.

Normally, with durable goods like cars, a large majority of people express no
intention of buying in the next year so that even a small percentage of this group
who do buy provides a large fraction of the total number of buyers. Pickering and
Isherwood (1974) found that 55 per cent of all buyers came from the group
expressing no intention to buy. Theil and Kosobud (1968) in the USA, and Gabor
and Granger (1972) in Britain, found that 70 and 65 per cent of purchasers
respectively were in the group stating a zero-purchase probability.

The extent to which people fulfil their intentions was reviewed by McQuarrie
(1988), who assembled data from 13 studies. McQuarrie found that those who
intended to purchase did so, on average, 42 per cent of the time, whereas those not
intending to purchase did not purchase 88 per cent of the time; this asymmetry is
probably related to the fact that it is easier not to do something than to do it (see
Box 7.2).
 Box 7.2 Reasons for inaction
Why don’t people do as they intend? East (1993) found that only two-thirds of those
intending to apply for shares in British government privatizations actually did so. When
the other third were asked why they had not followed through on their intention, they
were equally divided between changing their mind (e.g. because the investment looked
less advantageous) and inertia (e.g. forgetting, too much trouble). Pickering (1975)
investigated the failure to follow through on an intention to buy consumer durables. In
this case, respondents had usually changed their mind due to unforeseen circumstances,
such as lack of money, or because their current durable was lasting better than expected.

Discrepancies between attitude and behaviour may also increase with the
period that elapses between attitude measurement and behaviour
measurement. The longer the period, the more opportunity people have to
change their minds in response to new information or changed
circumstances. For example, the attitude to voting for a political party may
be affected by unfolding political events, and a measure of voting attitude
taken close to an election should naturally be expected to have more
predictive value than one taken a year before. However, although this effect
of time lapse seems common sense, a study by Randall and Wolff (1994)
found no evidence that the length of the interval was related to the
correlation between intention and behaviour.

One study has raised some concern about the prediction of behaviour from
intention. Chandon, Morwitz and Reinartz (2005) found that correlations in
surveys between intention and subsequent behaviour are increased by the
process of asking about the respondent’s intentions. It appears that, after
questioning, respondents are more likely to do what they said they were
going to do.
Section 2: The Theory of Planned Behaviour
(TPB)
Attitudes, intentions and behaviour have been combined in a comprehensive
model of consumer choice called the theory of planned behaviour (TPB).
Figure 7.1 illustrates this theory and Figure 7.2 shows it applied to playing
the UK National Lottery.

The TPB was developed over a long period, starting with Fishbein’s (1963)
expected-value theory of attitude. This theory was extended in a number of
studies to predict intention and behaviour (e.g. Ajzen and Fishbein, 1969;
Ajzen, 1971; Ajzen and Fishbein, 1972). In addition to attitude (AB), the
authors included subjective norm (SN) as a determinant of intention. SN
measures the person’s beliefs about what other people think they should do.
This extended model was renamed the theory of reasoned action by Ajzen
and Fishbein (1980) in a book in which they applied the theory to practical
concerns such as health, consumer behaviour and voting. In 1985, Ajzen
introduced the TPB by adding perceived control (PC) as a determinant of
intention. PC measures a person’s beliefs about the opportunities for an
action which are based on the environment and their own abilities. The
relative strengths of AB, SN and PC in determining an action are given by
the weights w1, w2 and w3. Since these weights vary from category to
category, they are established empirically, using regression analysis or
structural equation modelling.

Figure 7.1 The theory of planned behaviour (TPB)

Figure 7.2 The theory of planned behaviour applied to playing the National
Lottery

As can be seen from Figure 7.1, the three global variables determine
intention, which then determines behaviour. The weight w4 reflects the fact
that circumstances may stop people from realizing their intentions, or that
their intentions could change. Finally, the model includes a second direct
effect of perceived control on behaviour, with weight w5. This covers
behaviours such as giving up smoking and eating less where a lack of
personal control can undermine intention. This might apply to compulsive
gamblers who find the National Lottery difficult to resist.
This theory also covers altruistic behaviour, which can be driven by the
subjective norm, and it takes account of a person’s self-assessed abilities and
opportunities. As such, it is an advance on simple subjective expected utility
(SEU) models that do not allow for such influences. The subjective norm is
an internalized influence, exerting its effect through the agent’s memories
and values. The social agents that are recalled need not be present, or even
exist, for them to have an effect. In recent work, a measure called the
descriptive norm has been added. This is what others expect that the
respondent will do (as opposed to should do). For example, Forward (2009)
found that adding a descriptive norm variable to the model improved the
prediction of driving violations. Similarly, White et al. (2009) found that the
addition of the descriptive norm improved the prediction of recycling
behaviour. The TPB can be used to explain any behaviour where there are
reasons for action, but it is more appropriate for explaining category use than
brand use where there are few differences in beliefs and attitudes between
brands. Among the many behaviours to which the TPB has been applied are:
taking exercise, attaining college grades, condom use, health self-
examinations, escaping addiction, blood donation, mothers’ diet
management of their babies, food choice, recycling, buying environmentally
friendly products, Internet use, reducing risky driving, accident avoidance,
seeking funding, buying gifts, applying for shares in initial public offerings,
and complaining.

When behaviour has yet to occur, or is difficult to measure, studies predict

intention rather than behaviour, so it is important to know how strong the
link (w4) is between intention and behaviour. Conner et al. (2007) measured
intention and actual behaviour with regard to breaking speed limits using an
extended model of planned behaviour. The model predicted 82 per cent of
intention but only 17 per cent of actual behaviour was predicted. A review of
185 studies by Armitage and Conner (2001) found that, on average, 39 per
cent of intention was predicted and 21 per cent of objectively measured
behaviour. Another meta-analysis by Webb and Sheeran (2006) also gave a
modest association between intention and behaviour. Furthermore, the
associations found may have been inflated by the effect of prior questioning,
as indicated earlier (Chandon et al., 2005). In marketing, we are concerned
with behaviour since it is through behaviour that profits are made; for this
reason, evidence that there is usually only a moderate association between
intention and behaviour is a matter of concern.
Using the Theory of Planned Behaviour to
Explain Action
The TPB can be used in three ways to explain actions.

Level 1: Behaviour
The immediate precursors of behaviour in the TPB are intention and
perceived control, so one form of explanation concerns the relative impact
of these two factors. This has been examined by Madden, Ellen and Ajzen
(1992) who found that perceived control is usually the weaker factor.
Perceived control is likely to become important where there is substantial
difficulty in performing the predicted behaviour.

Level 2: Intention
Next, we need to know the relative importance of AB, SN and PC in
predicting intention. This varies from application to application. For
example, Jaccard and Davidson (1972) found that, among college women,
the use of the contraceptive pill was associated more with AB than with SN;
this probably reflected the importance of avoiding pregnancy for this group
at that time. Davidson and Jaccard (1975) found that among married
women with children, the intention to use the contraceptive pill was more
determined by SN.

Level 3: Specific Factor Explanations

The third type of explanation relates specific outcome, referent and control
beliefs to intention or behaviour. In the theory, specific beliefs relate to
intention through the appropriate global variable, but it is not always clear
which global variable a factor belongs to when a research study is being
conducted. For example, embarrassment about complaining might be seen
as an outcome or as a control factor. East (2000) found that embarrassment
acted primarily as a control factor, obstructing complaining, rather than as
an emotional cost if a complaint was made. At a practical level, this issue
can be resolved by correlating beliefs directly with intention.

Sometimes, the obvious reason for an activity is not its prime motivation.
Membership of sports clubs may be driven by the need to meet people, for
example. East (1993) found that application for shares in state privatizations
was driven more by access to finance, a PC factor, than by the expected
financial outcomes, an AB factor. This suggests that business people should
not become too focused on the good quality of their product or service since
many of the people who might buy it may be constrained by finance or
other control factors.
 Box 7.3 An application to giving up
cigarettes
In 1978, the UK Joint Committee on Research into Smoking recommended that a new
study should be undertaken on attitudes towards smoking. One stimulus for this work was
a 1977 report on attitudes and smoking behaviour prepared by Fishbein for the US
Federal Trade Commission. This recommended a shift of focus from ‘attitudes to
smoking’ to ‘attitudes to giving up cigarettes’. The new study was conducted for the
Office of Population Censuses and Surveys by Marsh and Matheson. The findings were
published by the government in 1983 but the main features appear in a paper by Sutton,
Marsh and Matheson (1990). The research measured AB and confidence in being able to
stop smoking, which was a form of PC measurement, but no measure of SN was made.

The researchers predicted behaviour on the basis of the difference between the expected
values of stopping and continuing to smoke. It is this difference that shows the personal
gain or loss of taking one option rather than the other. The research showed that the
majority of smokers accepted that smoking causes lung cancer (73 per cent) and heart
disease (59 per cent), but the study also showed that most of these people believed either
that they did not smoke enough to do any damage or that any damage was already done
and was irreversible. For such people, cessation held little promise of reduced risk. The
only smokers who saw a benefit in stopping were the minority who believed both that
they had an enhanced risk and that cessation would diminish this risk. The researchers
predicted that these people would make more attempts to stop and this was confirmed
when the smokers were followed up six months later. This evidence therefore supports a
causal process from attitude to action.

This research showed where to place the emphasis in health education in order to
encourage people to try to quit smoking; for example, by explaining that the risk from
cigarettes is related to the number smoked, that there is no threshold at which health
hazards begin, and that there are health benefits for nearly all people who stop smoking.
The study also showed that eventual success in quitting was strongly dependent on
confidence, and that health education should therefore emphasize ‘how to stop’ methods
which would build this confidence.

Level 3 explanations help us to choose intervention strategies. For example,

if a firm wants to encourage customers to let it know about service failures,
the key factors affecting complaining must be addressed. If many people
lack confidence in complaining, it is best to provide a clear procedure,
explain that it is supported by the company, and draw attention to this
procedure on receipts and on the company website.
Exercise 7.3 Research using planned
behaviour theory
Choose an action that is individually performed and voluntary, such as watching a popular TV
programme, carrying an organ donor card, installing solar water heating, going to the dentist
regularly or playing the National Lottery. Make sure that the action is appropriately specified
in terms of target, action, context and time:

Choose the target group.

From a sample of the target group, elicit the salient outcome, referent and control beliefs
about the action, using Exercise 7.1. Ideally, about 20 people should be used. Reduce the
salient beliefs by merging similar ones and dropping those that are rare.

Create a planned behaviour questionnaire, including:

the title of the questionnaire
the intention
the outcome beliefs
the referent beliefs
the control beliefs.

Set up different scales for each item and the form of the items from input. If using a program
package, mistakes and poor grammar in the questionnaire need to be eliminated by word
processing. In particular, the phraseology for control items can be clumsy and needs
adjustment. Despite this, using a program speeds up questionnaire development. Some items
are added automatically but these may be deleted if not required. The questionnaire will
usually cover two sides of paper when printed in a two-column landscape format (10 point).
The questionnaire looks better if scale referents such as ‘extremely’, ‘quite’ and ‘slightly’ are
italicized. The scaling is designed for a proportional font such as Times New Roman.

Gather data from 50+ respondents.

Analyse the data.

Use the COMPUTE function in SPSS to:

1. Create products between outcome probability and outcome evaluation (and the
corresponding products for the referent and control beliefs, but see later comment,
Multiplying Ordinal Measures).
2. Produce three sum measures by aggregating products for the outcome, referent and
control items.
3. Aggregate measures for each of the global variables and intention when more than one
scale is used. (A reliability test may be appropriate here to check that each scale is
measuring the same variable.)

Next:

(d) Test the correlations between sum and global measures.

(e) Perform a structural equation analysis or regression analysis to test the theory and
establish the relative weights of AB, SN and PC in the prediction of intention. (If you
use regression, ordinal regression is appropriate.)

Examine your analysis and answer the following questions:

Is intention most related to AB, SN or PC?
Do the sum measures correlate with the corresponding global measure better than
with the other two global measures?
Which belief factors correlate most with intention?
Which belief factors might be used to improve the product’s design or
positioning, if any?
What are the shortcomings of this study and analysis?
Applying Evidence from Planned Behaviour
Research
Planned behaviour research can suggest what may be important in decisions
on product development, positioning and advertising themes. When the
findings from planned behaviour research are used to influence others, the
influence attempt may not succeed for a number of reasons. First,
correlations do not mean that there is a causal relationship. Second, beliefs
may be strongly anchored and resist change, or may not change because of
a ceiling effect (no room for change). Third, the attempt at influence may be
interpreted in an unexpected way that does not bring about the intended
change. Fishbein and Ajzen (1981) stated that studies of the existing basis
for action give only an indication of where to place the emphasis in an
influence attempt; tests are needed to verify any effect.
Section 3: Problems with The Theory of Planned
Behaviour
Multiplying Ordinal Measures
In planned behaviour research, scales may be unipolar (1 to 7) or bipolar (–
3 to +3). Bagozzi (1984) and others have explained that the scale
measurement used in planned behaviour research is ordinal, but is treated as
ratio-scale when any two numbers from scales, such as outcome likelihood
and evaluation, are multiplied to form a product. This introduces error and
alters correlations with other variables. This alteration can be substantial
when a switch is made between bipolar and unipolar scales. In this
situation, Ajzen (1991) recommends optimal scaling, i.e. adding a constant
to each scale to produce the highest correlation between the sum and global
variables. This procedure gives some benefit to random effects and there is
no specific justification for taking the scaling that gives the best correlation.
An alternative method is to use the four combinations of 1 to 7 and –3 to +3
to see whether the results vary much. If similar results are obtained,
whatever the scaling, there is more assurance about the results.
The Principle of Sufficiency
The TPB is based on beliefs. Therefore, any change in global variables,
intention or behaviour must come about through the acquisition of new
beliefs or the modification of existing beliefs. In other words, belief changes
are a sufficient explanation for subsequent changes in global variables and
intention. Ajzen and Fishbein (1980) accept that variables external to the
theory, such as past experience, personality, age, sex and other social
classifications, will be associated with behaviour, but they argue that this
occurs only because these variables are related to the relevant beliefs and
hence to AB, SN or PC. They state:

Although we do not deny that ‘external’ variables of this kind may

sometimes be related to behaviour, from our point of view they can
affect behaviour only indirectly. That is, external variables will be
related to behaviour only if they are related to one or more of the
variables specified by our theory. (Ajzen and Fishbein, 1980: 82)

Thus, beliefs and the other components of the TPB should mediate the effect
of external variables, as shown in Figure 7.3. This argument has been tested
in a number of studies by including external variables in the regression
analysis to see whether these significantly improve the prediction of
intention compared with the global variables alone. Often, demographic
variables have little effect; for example, Marsh and Matheson (1983) found
no direct effects of age or sex on intention in their study on smoking
cessation, and Loken (1983) found no direct effect of external variables on
television watching.

Figure 7.3 Feedback from experience should affect beliefs

However, it is usually found that past experience has a direct effect on
intention and behaviour (Bagozzi and Kimmel, 1995). In the Marsh and
Matheson (1983) study, the previous experience of attempting to stop
smoking had a direct effect on intention and a small direct effect on attempts
to stop smoking. Similar direct effects of past behaviour on both intention
and subsequent behaviour have been found by Bentler and Speckart (1979,
1981), Fredricks and Dossett (1983) and Bagozzi (1981). One possible
explanation is that people are partly controlled by their environment as a
result of habits set up by experience but that this control is not fully captured
by the global measures (Bentler and Speckart, 1979; Fredricks and Dossett,
1983; Triandis, 1977). This may be because habits are set off by stimuli that
are not consciously recalled.
The Development of AB, SN and PC
As experience increases, people become more informed and the belief basis
for future action is changed. This is a situation that is particularly relevant
to consumer behaviour. Consumers are naïve when they enter markets that
are new to them and as they repeat purchase they become more
experienced. When the experience is positive, intention will be enhanced in
a positive feedback loop; if the experience is negative, intention will be
reduced and further trial curtailed. Thus, under the voluntary conditions that
attach to most consumer behaviour, we would expect to find that those who
are highly experienced have stronger intentions. Applying the TPB, there
should be changes in AB, SN and PC as intention develops, as illustrated in
Figure 7.3. This can occur because the beliefs underlying the global
variables become more numerous and more strongly linked to the
behaviour. If experience makes the belief basis of planned behaviour more
complicated, we may ask whether it has the same effect on AB, SN and PC.
East (1992) suggested that the progression from novice to expert consumer
involves a movement from actions based mainly on SN to actions based
more on AB and PC. This proposal is founded on the idea that, in the
absence of detailed knowledge, people have to make decisions on the basis
of naïve ideas, and that ‘what others think I should do’ is better known, or
more easily guessed, than the benefits and opportunities related to an
unfamiliar prospect. East identified three minor studies that supported the
idea that SN gave way to AB and PC as people became more experienced.
However, as further studies accumulated, no clear-cut pattern emerged and
the work was not published. This issue has yet to be resolved. The studies
that did not support the theory were mainly financial decisions, such as the
purchase of endowments and pensions, and these may not have permitted
much development of knowledge from experience.

Despite the lack of support for the theory, the issue of how experience
affects the different beliefs underlying global variables is important and
deserves further study. In many TPB studies, SN is a relatively weak
determinant of behaviour and there has been some tendency to downgrade it
(e.g. Sheppard, Hartwick and Warshaw, 1988). Armitage and Connor
(2001) found that part of the weakness could be attributed to poor
measurement.
Deliberate and Spontaneous Action
Fazio (1990) divides the explanation of behaviour into two fields. Where
decision-making is deliberate, Fazio supports the TPB, but when the
decision is spontaneous, he proposes a different model in which attitude
comes first. This second model is shown in Figure 7.4.

Figure 7.4 Fazio’s (1986) theory of the attitude–behaviour process

In Fazio’s (1986) account, attitudes are automatically activated by

observation of the attitude object. The attitude then guides perception and
the individual becomes aware of aspects of his or her environment related to
the attitude. The definition of the event then occurs as these perceptions are
associated with a normative understanding of the situation, and out of this
definition of the event behaviour may follow. The main point here is that the
environment is driving cognitive processes in an automatic way and attitude
rather than belief is the foundation of the internal process.

Fazio argues that attitude activation occurs only when the object and its
evaluation have been well established in memory, usually through direct
behavioural experience. Thus,the spontaneous production of behaviour is
restricted to familiar contexts, leaving planned behaviour to explain the more
unusual situations. However, this rather cosy division of the field has been
disrupted by evidence from Bargh et al. (1992) that a wide range of objects
can elicit attitude-driven effects. This casts doubt on the separation between
the automatic and deliberate control of behaviour. It seems possible that
these two types of explanation may often work together. Fazio (1990)
showed that responses in measurements of deliberate behaviour are affected
if people are asked to focus on emotion-evoking objects before
measurement. Baldwin and Holmes (1987) found that systematically
different measurements were obtained when different social referents were
visualized before response. These findings reflect the effect of automatic
processes even when people are making considered responses.
Adjustments and Alternatives to the Theory of
Planned Behaviour
A number of researchers have suggested modifications of the theory.
Bagozzi and Kimmel (1995) and Bagozzi (1992) suggested distinctions
between intention, desire and self-prediction, and Armitage and Conner
(2001) found some support for these distinctions. There have also been
proposals to include a moral norm to cover the respondent’s personal
normative control, but this has been found to have an additional predictive
function only for some actions.

Ajzen and Driver (1992) distinguished between the short-term and

expressive consequences of action and the longer-term and more cognitive
consequences of action. Often, a short-term pain has to be balanced against
a long-term gain. On this basis, AB should be seen as having two
components.

In its present form, the theory includes only one type of subjective norm –
what respondents think that others think that they should do. As noted
earlier, what others actually do (descriptive norm) exerts an additional
influence on behaviour (Deutsch and Gerard, 1955). When descriptive
norms are included, predictions are usually better supported (Rivis and
Sheeran, 2003). Other developments are possible; Armitage and Conner
(2001) separate measures for the control element in the theory into the
control that a person has by virtue of his or her abilities and the control that
is made possible by the environment. However, all these adjustments raise
the complexity of the theory and make it harder to use.

In the consumer behaviour context, the technology acceptance model

(TAM) has been offered as an alternative method of predicting the intention
to use new systems (Davis, 1989). This theory bases behavioural intention
on two factors: perceived usefulness and perceived ease of use. This model
has the advantage of simplicity and avoids elicitation but it applies to some
situations only and may not give as much explanation as a belief-based
model. In TAM, usefulness has some correspondence with AB, and ease of
use with PC, and while the theory originally did not include SN, Venkatesh
and Davis (2000) have modified the TAM model to incorporate SN.
Reviewing Planned Behaviour Theory
The development of the TPB has been a success story. Social psychologists
have emerged from the dark days of 1969 when Wicker claimed that there
was little or no connection between attitude and behaviour. We now have a
predictive and explanative model which works effectively, though there
remains a nagging doubt about the scale of behaviour prediction.

However, there is a concern about the methods that have been used to test
the theory. If the theory is causal, it should be possible to supply
information and then observe how the effects of that information change the
components of the theory. This requires research designs in which
comparisons are made between experimental conditions that induce
changes of belief. The fact that there is little published evidence of this sort
may indicate that little has been done in this area or that the results of such
research have not been conclusive.

Finally, there is a need to apply the theory more effectively in marketing.

The theory is of limited use in explaining the preference for one brand over
another, unless the brands are markedly different (e.g. one restaurant versus
another or one holiday destination rather than another). The main strength
of planned behaviour research is at the category level. It can explain why
categories are liked and, therefore, it can assist managers to position brands,
target more responsive population segments and select themes for
advertising.
Summary
Attitudes are an evaluative response to a concept. The concept is a cluster
of attribute beliefs about a particular topic, and each belief has values
attaching to it. Thus, the attitude to the concept should relate to the
aggregate value of the attribute beliefs. Generally, this view of attitude has
been upheld by research and it fits the idea that the purchase of a product
can be seen as the acquisition of a bundle of expected costs and rewards.

In many studies, the correlation between attitude and behaviour measures is

found to be weak. There are two reasons for such weak relationships. The
first is that ‘other variables’ may swamp the association between attitude
and behaviour. The second reason is that the measures of attitude and
behaviour may not be compatible, i.e. these measures may not refer to the
same action, target, context and time. A mismatch here means that the
wrong attitude is being used to predict behaviour.

The closest prediction of behaviour is provided by measures of intention –

those who state that they will buy a product are, unsurprisingly, found to be
much more likely to buy it than those who state that they will not buy it.

The TPB (Ajzen, 1985, 1991) formalizes this link between beliefs,
attitudes, intentions and behaviour. In this theory, there are three global
variables – attitude to a behaviour, subjective norm and perceived control.
These three variables have a combined effect on intention. Behaviour is
predicted by intention and perceived control. The global variables rest on
beliefs about the outcomes of behaviour, the referents who think that a
person should or should not engage in the behaviour, and the ability and
opportunity to engage in the behaviour. The TPB provides different levels
of explanation which can be used in product positioning, development and
advertising themes.
Additional Resources
Issues on the nature of attitudes and their relationship with behaviour are
well discussed in Chapter 4 of Eagly and Chaiken (1993) and in Eagly and
Chaiken (2005). A review of the TPB is provided by Ajzen (2002) and by
Armitage and Conner (2001). Ajzen’s website is
www.people.umass.edu/aizen/faq.html (note: ‘aizen’ is an alternative
spelling).
8 Information Processing and Decision-
Making
Learning Objectives
When you have completed this chapter, you should be able to:

1. Understand the way schemas and response competition are involved in thinking.
2. Explain what is meant by a heuristic mechanism and describe how these mechanisms
may bias decision-making.
3. Describe how human beings respond to objective probability and value.
4. Understand the ideas of framing, mental accounting and editing.
5. Describe how this work can be used to influence others and improve consumer choice.
Overview
In the previous chapter, the theory of planned behaviour was described. Though it assists
prediction, this theory does not really describe how people think. People do not assign
likelihoods and evaluations, multiply them and sum the products to form their attitudes.
Planned behaviour theory works as if people figure out their interests in this way, but no
claim is made that they actually do so.

An alternative approach, described here, focuses on automatic mechanisms that guide

information-seeking and choice. This approach seems to get closer to the thought
processes (often unconscious) that govern behaviour. We deal first with the way
structures called schemas can guide thought and recognition. Then we consider the
situation where there is no clear schema available to interpret a stimulus, which leads to
response competition and sustained attention as alternative schemas are tried. When such
ambiguous stimuli are repeatedly exposed, they may be liked more as response
competition is reduced. After this, the chapter focuses on the work of Kahneman, Tversky
and their many colleagues who have transformed thinking about the way human beings
process information and make choices. Kahneman and Tversky show that human
decision-making is affected by associative mechanisms that may be invoked by the
situation in which people find themselves. Their methods have mainly involved simple
choices, presented to participants, with a count of the preferences expressed. Using such
methods, they have documented many instances where judgement is biased away from a
rational model, and decision-making departs from economic assumptions.
Section 1: Schemas and Attention
Schemas
According to Crocker, Fiske and Taylor (1984: 197), a schema is:

an abstract or generic knowledge structure, stored in memory, that

specifies the defining features and relevant attributes of some stimulus
domain, and the interrelationships among those attributes ... Schemas
help us to structure, organize and interpret new information; they
facilitate encoding, storage and retrieval of relevant information; they
can affect the time it takes to process information ... Schemas also
serve interpretive or inferential functions. For example, they may fill
in data that are missing or unavailable in a stimulus configuration.

The notion of the schema was implicit in the early work by Bartlett (1932)
on remembering. Bartlett wrote of the ‘effort after meaning’ and showed
how unusual structures that fell short of representing any object were
interpreted by reference to more familiar ideas. One of Bartlett’s stimuli
was a diagram of an ambiguous object that was variously recognized as a
battleaxe, turf cutter, anchor or key, although it was not quite like any of
them. People were using the schema for a more familiar object to make
sense of the ambiguous object.

More frequently encountered stimuli are more easily retrieved. In addition,

stimuli with particular characteristics are recognized or recalled more
easily: ‘Colorful, dynamic, or other distinctive stimuli disproportionately
engage attention and, accordingly, disproportionately affect judgements’
(Taylor, 1982: 192). This leads to the idea that thinking is based on
cognitive accessibility. The more rapidly an idea can be brought to mind,
the more likely it is to figure in cognition and subsequent processing. This
retrieval bias has parallels with accessibility in the physical environment.
For example, people buy more from those supermarkets that are nearby.
Internet search engines use a related process when they take account of the
frequency of relevant outcomes in previous searches. Retrieval bias is quite
efficient because it ensures that the more likely candidates are considered
first. However, concepts that occur infrequently or are hard to visualize for
other reasons will tend to be left out of cognitive processing. In general,
experience-based concepts are more easily retrieved than communication-
based concepts, events are favoured over states, recent occurrences over
long-past occurrences, and the clearly defined concept over those that are
obscure. Bartlett’s example involves the simplest of schemas – those used
to classify objects. More elaborate schemas include classifications of
persons or groups, grammatical forms and social roles. In their most
abstract form, schemas may cover relationships like logical validity,
causality and symmetry.

When more than one schema fits a situation, people experience response
competition as they struggle to give meaning to the stimulus (see Box 8.1).
This leads to extended attention as people grapple with the problem.
Box 8.1 Response competition
Response competition is demonstrated by the Stroop test. In this test, you are asked to call
out the colour that is written the moment that it is displayed. In one condition, the word is
RED written using a red colour; in the other condition, RED is written using a blue colour
and, in this second case, respondents take longer to respond than in the first condition. In
the second case, two competing responses are aroused – to say ‘red’ (reading the word)
and to say ‘blue’ (the colour of the text) – and the competition delays the production of
the correct response.
Managing Schemas
Since thought is guided by schemas, it follows that negotiation and other
forms of influence may succeed by manipulating the selection of schemas
that people use to interpret their experience. One standard ploy in
negotiation is to try to anchor the discussion around a particular range of
outcomes. Early in the discussion, a negotiator might say, ‘The normal rate
for this type of work is £1,000 a day’. This can constrain offers to those
fairly close to the rate mentioned. Turning to advertising, schemas may be
used to develop the product concept. For example, ‘No FT, no comment’
implies that those who read the Financial Times are likely to be better
informed – it conjures up a notion of the business expert dispensing
wisdom. Often, public relations exercises can be seen as attempts to
manipulate the schemas used in judgement. For example, firms may play up
their green credentials by drawing attention to their energy-saving actions.
Some interesting examples of schema management are shown in Box 8.2.
Do People Like Response Competition?
Jokes are often based on response competition. People may anticipate one
outcome and have their expectations confounded at the punchline. But other
forms of response competition may be disliked. Even jokes may be disliked
when a person is under stress. The explanation for this variable reaction is
that it depends on the degree of arousal of the person involved. Boredom
(low stimulation) and high stress (high stimulation) are both conditions of
high arousal. People prefer an intermediate level of stimulation because, in
this region, arousal is lower. Conceptual conflicts may be welcome when
people are inactive or bored (e.g. when watching television) because the
extra stimulation will reduce arousal; at other times, unusual stimuli may
raise arousal (Berlyne, 1965; Berlyne and McDonnell, 1965) and, when this
occurs, the stimuli may be disliked. This explains why Harrison (1968) and
Saegert and Jellison (1970) found that stimulating objects were often
disliked.
Box 8.2 New schemas for old
An activist group took on the tobacco companies in Australia. The group was called the
Billboard Utilising Graffitists Against Unhealthy Promotions (BUGA UP) and it
specialized in ‘refacing’ tobacco posters. When a cigarette company offered a car as a
prize, the poster was given the caption: ‘From the people who put the “car” in
carcinogen’. The adjustments to the posters, and the speeches in court when members of
the group were prosecuted, gave entertainment to many Australians, who much
appreciated the sight of multinational tobacco companies being humbled in this way.

When a tobacco company sponsored work at the Sydney Opera House, well-dressed
members of BUGA UP distributed leaflets expressing regret at this unsavoury association
between tobacco and the arts. Another BUGA UP enterprise sabotaged a Marlboro ‘Man
of the Year’ competition in Australia. BUGA UP proposed its own candidate – a man
disabled by smoking, confined to a wheelchair, and smoking through the hole in his throat
provided by a tracheotomy operation. The man himself was a willing accomplice and
starred in a poster which was printed and sold in large numbers. The schema of the strong
heroic figure that Marlboro had tried to cultivate was ridiculed. In its place were put the
schemas of disease and disability which are more accurately related to smoking
cigarettes. A further ‘anti-promotion’ counteracted the distribution of free cigarettes in
shopping malls. To most people, a gift is a kindness and the giver is regarded as well-
meaning. To oppose such promotions, BUGA UP arranged for children to parade around
the mall with banners saying ‘DANGER – DRUG PUSHERS AT WORK’. This changed
the perception of the tobacco companies’ motives from kindness to self-interest. Tobacco
companies have a squalid record of refusing to admit to the hazards of their products;
they are licensed drug sellers and an important part of their public relations has been to
counter such facts by sponsoring orchestras, sport and research. BUGA UP’s achievement
was to reassert the drug-seller schema as the one by which the tobacco companies’
actions could be judged.
Mere Exposure
Response competition has been used to explain an interesting phenomenon
first reported by Zajonc (pronounced Zi-onse, 1968). Zajonc observed that
repeated exposure to a new stimulus often made people like it more. This
effect of mere exposure was so called because the change in the observer’s
evaluation occurs without the use of reinforcement (discussed in Chapter 1).
Zajonc observed this effect in both laboratory and field experiments, using
nonsense words, obscure characters and photographs of unknown faces as
the unfamiliar stimuli. For example, in Zajonc and Rajecki’s (1969) field
experiment, nonsense words such as NANSOMA were printed like
advertisements in campus newspapers. Later, the researchers got large
numbers of students to rate the words on evaluative scales and there was
clear evidence that the frequency of exposure correlated positively with the
evaluative rating. Zajonc’s explanation for this was that the nonsense words
created response competition in the minds of readers and that this was
reduced as people developed a familiarity with the nonsense word after
repeated exposure. If the response competition created by nonsense words
is generally disagreeable, a reduction in competition (e.g. through
familiarity with the nonsense word) should produce a more positive
evaluation. Harrison (1968) measured response competition as the time
delay before any response to the stimulus and found that the delay was
reduced as the number of exposures increased. Lee (1994) offered another
ingenious explanation which was based on the availability heuristic
discussed later in this chapter: repeated exposure speeds up recognition; the
stimuli that we recognize more easily tend to be those that we like; this
association guides judgement and leads us to give such stimuli higher
evaluations.

However, not all stimuli become more liked on repeated exposure. This
suggests that prospective brand names (which are often like nonsense
words) should be pre-tested to see whether they are well liked after they
have become familiar.
The Response to Thought and Feeling Stimuli
We tend to think that recognition is a necessary precursor to any evaluation
of a stimulus. If you do not know what the concept is, how can you have
any affective response to it? Strangely, it seems that we can have an
evaluative response without recognition. Zajonc (1980) showed that thought
and feeling are initially processed independently, and Kunst-Wilson and
Zajonc (1980) found that evaluative responses occurred slightly ahead of
recognition. Zajonc points to the survival value attaching to a fast response
to dangerous stimuli: it is better to jump without thought than to recognize
that it is a car that is hitting you!

In a further experiment, Marcel (1976) used an instrument called a

tachistoscope to present stimuli at short intervals and low levels of
illumination so that they were hard to recognize. The stimuli were either
words or blank spaces, with equal likelihood. Words were either short or
long and either ‘good’ or ‘bad’. A good word might be ‘food’ while a bad
word might be ‘evil’. If the participants in the experiment thought they saw
a word, they were asked to judge its length against comparison words and
also to give an evaluative judgement, to say whether the word was good or
bad. The duration of exposure was reduced until the subjects were guessing
the presence of words at chance level and could not therefore have been
recognizing anything. At this duration, Marcel found that word-length
judgements (also cognitive) were at chance level too. However, at this point
the subjects were still scoring at above chance on their evaluative
judgements of words when these were present, indicating that the evaluative
response was generated faster than the recognition response. Vanhuele
(1994) reviews this work.

Fast recognition judgements of the sort studied by Zajonc are quite different
from the choices typically faced by consumers, but these studies show how
unconscious mechanisms can underlie consciously experienced thought and
feeling.
Attention and Value
In decision-making, the observable action is often restricted to the overt choice.
However, when the alternatives are physically present, it is possible to observe
the direction of gaze and to infer from this which alternative a person is
thinking about. Gerard (1967) used this method of investigation; he employed
two projectors to show two alternatives (Impressionist paintings), while light
reflecting off a mirror attached to the back of the participant’s head showed
which alternative was receiving attention. A multi-channel recorder logged the
data. Gerard reported that the participants looked most at the alternative that
they did not choose and suggested that they were trying to come to terms with
not having this alternative.

To test Gerard’s result, East (1973) conducted two experiments using a battery
of slide viewers connected to a hidden time recorder (illustrated). As in the
earlier study by Gerard (1967), the choice was made between French
Impressionist paintings. The participants were led to believe that they would
get a poster of the picture that they chose. Control of the viewing equipment
was left entirely in the hands of the participant so that the time spent on the
different alternatives was unconstrained.
East’s first experiment presented subjects with two alternatives, while a second
experiment presented three alternatives. Both experiments had two levels of
choice difficulty: high, between alternatives that had previously been rated
equally by the subject, and low, between alternatives that had been rated
unequally.

The results showed that the subjects spent more time looking at the alternatives
that they liked, so that the ratio of attention times was an approximate function
of the ratio of the evaluations (see Table 8.1). East’s result was the opposite of
Gerard’s reported findings and it is possible that, with Gerard’s rather
complicated method for recording attention, the records were inadvertently
linked to the wrong alternative. A later study by Russo and Leclerc (1994)
supported East’s finding. Russo and Leclerc used video equipment in a
simulation supermarket situation and measured the number of eye fixations on
alternatives (rather than duration of time spent) at different phases in the
decision sequence. They found that in the main phase of the decision the
number of fixations clearly favoured the alternative later chosen. In another
study, Pieters and Warlop (1999) also found that more attention was paid to the
alternative that was eventually chosen.

Thus, there is a simple mechanism that directs attention to the more valued
alternative. This mechanism is unconscious; when asked about their potential
behaviour, people do not know which alternative they would look at most. This
evidence suggests that the more valued features of a person’s environment
generally get more attention than those of lesser value. We can see that this
mechanism helps people to benefit from their environment. To avoid harm, it is
likely that people attend more to unpleasant stimuli than to neutral ones but we
do not have data on this.

If evaluation guides attention, it means that second and third preferences will
get proportionately less attention and their worthwhile attributes are less likely
to be discovered. Only when investigation of the first preference leads to it
being down-rated will more time be allocated to lesser alternatives. This
mechanism therefore carries a bias in favour of existing preferences but it is an
efficient way of using time since it ensures that little time is wasted on low-
rated prospects. Given this evidence, people may be encouraged to buy an
initially lower-rated brand if information about this brand is attached to a
message on the initially preferred alternative, since it is then more likely to
receive attention. This may be done in comparative advertising, and there is
evidence that small brands benefit from this (Grewal et al., 1997).
Section 2: Heuristics
 Exercise 8.1 Availability effects
In the UK, approximately 600,000 people die each year from all causes. How many people die
prematurely each year from the following two causes? Enter the figures that you think apply:

Smoking:
Road accidents:

We now consider biases that may affect everyday decision-making. If you

want to find the river, go downhill! This is a heuristic rule which often helps
but may mislead when there is no river in the valley. The term ‘heuristic’
was used by Kahneman, Slovic and Tversky (1982) to cover inexact or rule-
of-thumb processes which may be used consciously or unconsciously to
make judgements. Kahneman et al. argue that people do not appear to
follow the statistical theory of prediction when making such judgements.
Instead, they rely on a limited number of heuristic processes which often
yield reasonable judgements but sometimes lead to error. In particular,
people seem to attach higher probability to ideas that are easily retrieved;
this is called the availability heuristic. This arises because frequently
experienced concepts become easier to retrieve; as a result, ease of retrieval
becomes associated with higher probability.

Markus and Zajonc (1985) provide an example of the way in which

availability may quite unjustifiably support the prestige of the medical
profession. People often recover naturally from illness, but when treatment
has been given there is a tendency to assume that this treatment was the
cause of the recovery. The treatment is more cognitively available as a
cause of recovery than the unseen natural processes counteracting illness.
As a result, people may judge therapy to be more effective than it is.

The judgement of risk is notoriously erratic. Some of the reasons for this
may lie in poor information about actual risks in the media, but judgement
may also be distorted by heuristic thinking. Lichtenstein et al. (1978)
suggested that the risk of occurrences that are referred to often in the media
becomes exaggerated because of the availability heuristic. Those who
completed Exercise 8.1 are likely to have overestimated the risk of death
from road accidents because these events are more salient in media reports.
Smoking deaths are less well reported and are likely to be underestimated.
The approximate annual numbers of deaths in the UK in 2013 were:

All causes 576,000
Smoking (approx.) 100,000
Road accidents 1,713

Misjudgement of risk has also affected the use of oral contraceptives.

Exaggerated fears of potentially harmful side-effects have caused women to
abandon the pill, even when the identified risk was very small in absolute
terms. In 1996, a 10 per cent increase in legal abortion in Britain was
attributed to earlier announcements that a number of contraceptive pills
should be phased out because of small associated risks.

People seem to have difficulty in taking account of background risk and

may focus instead on large percentage increases. For example, women may
be shocked to hear that those who are over 35 and who smoke and take the
contraceptive pill have 18 times the risk of pulmonary embolism compared
with those who do neither of these actions. The ‘18 times’ fact seems to be
more available and to dominate in judgement, but embolisms are very rare
in the 35–45-year-old age range so this is 18 times a very small number and
other hazards present far more risk, particularly smoking, since smokers
lose, on average, about six years of life. A more responsible way of
handling the data would be to report the personal increment in the risk for
smokers using the pill, for instance that a smoker who takes the pill has an
extra risk of dying of one in a million.
 Exercise 8.2 Who was to blame? (Abridged
from Tversky and Kahneman, 1980: 62)
Solve the following problem:

A cab was involved in a hit-and-run accident at night. Two cab companies, the Green
and the Blue, operate in the city. You are given the following data:
85 per cent of the cabs in the city are Green and 15 per cent are Blue.
A witness identified the cab as a Blue cab. The court tested his ability to identify
cabs under appropriate visibility conditions. When presented with a sample of
cabs (half of which were Blue and half of which were Green), the witness made
correct identifications in 80 per cent of the cases.

Question: What is the probability that the cab involved in the accident was Blue rather than
Green?

Decide on your answer before reading on.

Tversky and Kahneman (1980) put this problem to several hundred participants; the
median response was an 80 per cent likelihood that the cab was Blue.

Thus, participants tended to take note of the witness’s skill in recognizing cabs and ignored the
market shares of the two cab companies. Clearly, if there had been no Blue cabs, the witness
could not have been right so the proportion of Blue cabs is relevant. The probability that the
witness was right is the ratio of correct identification of cab colour as blue to total
identification as blue (both correct and incorrect). The chance that the cab was Blue (0.15) and
was recognized correctly (0.8) is 0.15 × 0.8, and the chance that the cab was Green (0.85) and
was recognized wrongly as Blue (0.2) is 0.85 × 0.2. The required ratio of correct identification
to total identifications is therefore:
( 0 . 15 × 0 . 8 ) ( 0 . 15 × 0 . 8 ) + ( 0 . 85 × 0 . 2 ) = 0 . 41

So there is a 41 per cent chance the cab was Blue.

The tendency to ignore base rates, such as the market share of the cab
company, is the basis of the representativeness heuristic. Likelihood is
judged by reference to visible similarities rather than background
probabilities. For example, a person may be seen as a barrister because of
features of dress and delivery of speech. In this case, the judgement draws
on the stereotype of a barrister, but such a judgement takes no account of
the low number of barristers in society which makes it unlikely that a
person belongs to this group.

The bias towards the witness test in Exercise 8.2 may also relate to the fact
that this test is an event. As previously reported, events are more available
than continuing states such as market share. We seem to be tuned to change
and direct our thinking to the more active aspects of a problem. In contrast,
data dealing with unchanging features of our environment do not attract as
much attention. This mechanism serves a useful purpose by drawing
attention to happenings that may require a response but it can cause
mistakes in some cases. For example, it may lead to consumer financial
decisions that are related to active features of the environment rather than to
their impact on wealth. Wealth is a state rather than an event and does not
usually figure in individual judgements even though, normatively, it should.

People are also prone to give more weight to causal data, which is related to
change. For example, respondents are more likely to agree with the
proposition that a girl has blue eyes if her mother has blue eyes than a
mother has blue eyes if her daughter has blue eyes, though the two events
are equally probable (Tversky and Kahneman, 1980). The mother-to-
daughter inheritance is causal, unlike the daughter-to-mother relationship.

The focus on events rather than states and the different heuristic rules were
described by Kahneman (2002) as intuitive thinking in his presentation
following his award of the Nobel Prize for economics. Kahneman likens
such thinking to the way perception seems to be governed by mechanisms
over which we have little conscious control. What we perceive is a function
of the context from which reference points are drawn. Small changes in
problems can affect the reference points and change the judgement. There
are criticisms of this work; see, for example, Gigerenzer (1991), who has
raised questions about the interpretation of effects (response by Kahneman
and Tversky, 1996).
Relevance to Marketing and Management
The greater cognitive availability of events and causal data has a relevance
to marketing. For example, we may exaggerate the scope for market
interventions, such as brand extension. As we saw in Chapter 4, conditions
such as market share control the likely outcome of such interventions, but
these conditions may get less attention than they deserve because of their
constancy.

In addition, retrieval bias will move decisions towards the option that is
easiest to bring to mind. This may lead us away from the prevention of
undesirable occurrences before they happen (proaction), and as a result we
may have to respond to undesirable occurrences after they have happened
(reaction). The choice between proaction and reaction depends on costs.
Sometimes, it is best to let things happen and then to focus resources on the
problem – management by exception – but in other cases, for example
avoiding accidents, prevention is usually best. Our point is that there is a
bias against proactive intervention because successful prevention produces
no visible outcome and this choice is therefore less cognitively available. In
sum, retrieval bias will operate in favour of the visible, well-defined events
and against those that are hard to bring to mind. This suggests that people
may:

give too much support to the status quo: what is happening is

available, but what could happen is harder to bring to mind
make a poor assessment of the opportunity cost, which is the
alternative use of resources when a course of action is selected
find it easier to sell products that have a physical form that is easily
seen.
Section 3: Processing Value and Probability
Objective value, expressed in money or other units, and objective
probability, the likelihood of something occurring, are processed by human
beings to produce subjective evaluations (or utility) and subjective
estimates of probability. These subjective representations do not exactly
correspond with the objective forms and this affects decision-making.
Value
The relationship between objective value and utility has a long history
going back to Bernoulli (1738), who described how the curve of utility
against wealth flattens as wealth increases, and marginal utility therefore
diminishes with each increment in wealth. Bernoulli gave the example of a
pauper who, finding a lottery ticket offering an equal chance of winning
20,000 ducats or getting nothing, might quite reasonably ensure a gain by
exchanging his ticket for a guaranteed 9,000 ducats. The value function is
curved, so that half the utility of 20,000 ducats is less than the utility of
9,000 ducats.

Figure 8.1 The value function (Kahneman and Tversky, 2000)

This curvature assists exchanges. To a person who has enough of a good,
the marginal utility of additional supplies is low and this will encourage
exchanges with others who possess different goods that the person needs.
Both parties in such an exchange can gain in utility. However, Bernoulli
was mistaken to think that such exchanges were reckoned against total
wealth. Normally, people assess gains or losses against a more salient
criterion – and this is often the zero point. This relationship between gains
and losses relative to zero, and utility, is shown by the curve in Figure 8.1.
You can see that the curve is concave to the x-axis for both gains and losses
and that the response to losses is larger than the response to gains. This
relationship has been established by observing the preferences expressed by
individuals about different choices. Comparisons between positive and
negative choices have been particularly interesting (see Exercise 8.3).
 Exercise 8.3 Positive and negative choices
1. Which do you prefer?
A: £9,000 for certain, or
B: £10,000 with a probability of 0.9; otherwise nothing.

2. Which do you prefer?

C: Losing £9,000 for certain, or
D: Losing £10,000 with a probability of 0.9; otherwise nothing.

In Exercise 8.3, people generally prefer A to B. The shape of the value–

utility relationship for gains explains why people prefer £9,000 for certain
rather than have 0.9 × £10,000. They also prefer D to C. The shape of the
value function in the negative region means that losing £10,000 with a
probability of 0.9 is less painful than losing £9,000 for certain and people
prefer the smaller disutility. These outcomes mean that people are generally
risk averse on gains but risk prone on losses. This pattern of preference
reversal is regularly found and is called the reflection effect. However, in
1992, Tversky and Kahneman suggested that the data were more consistent
with the idea that people are risk averse on gains and risk prone on losses
when outcomes have medium or high probability; for low probability
outcomes they suggest the reverse risk profile occurs. This modification
arises because of the way people weight low probabilities, as we explain
later.

Figure 8.1 shows another interesting effect. This is the steepness of the
negative part of the value function in comparison to the positive part. This
effect is captured by the aphorism losses loom larger than gains and, more
formally, as loss aversion. Loss aversion is behind the endowment effect
which is that people often demand much more to give up an object than
they would be willing to pay to acquire it. Generally, people will be
reluctant to trade what they own, except at a high price. The endowment
effect has been demonstrated in a number of studies reviewed by
Kahneman, Knetsch and Thaler (1991a). One simple example of the effect
is the reluctance of people to engage in a 50:50 win/lose bet. On average,
people will only wager a dollar on a coin toss if they can win more than two
dollars. Not surprisingly, the endowment effect has been tested by critics;
see, for example, Shogren et al. (1994).

As we stated, gains and losses relate to some reference point. In some cases,
it will be a prior cost. For example, a person may see the $20,000 cost of
building work on a newly acquired house as an addition to the $1million
price paid for the house. Viewed like this, the $20,000 seems a modest
increment to the purchase price, but ten years later, when the purchase of
the house has faded into the past, further building work costing $20,000
will be evaluated on its own and will be psychologically more painful.
Thaler (1999) points out that the extent to which a cost is psychologically
linked to a benefit can vary. When people pay a fixed cost for a service,
irrespective of their amount of use, usage is decoupled from the payment
since any extra use is free. Another decoupling occurs when a credit card is
used. This postpones payment and also aggregates costs into one bill, which
reduces the total psychological cost compared with the sum of several
smaller separate costs (because of the curve in the utility function).
Probability
What are your answers to Exercise 8.4?
 Exercise 8.4 The Allais paradox (Allais, 1953)
Allais asked one group of subjects to choose between two options:

A: $4,000 with a probability of 0.8; otherwise nothing.

B: $3,000 for certain.

Which do you prefer?

Another group was asked to choose between:

C: $4,000 with a probability of 0.2; otherwise nothing.

D: $3,000 with a probability of 0.25; otherwise nothing.

Which do you prefer?

Faced with the choices in Exercise 8.4, Allais found that 80 per cent of
respondents preferred option B to A but 65 per cent preferred option C to D.
This seems paradoxical because the ratio of the probabilities is the same in
each choice pair. One explanation for this pattern is that probability is
weighted as it is converted to subjective probability. Figure 8.2 shows how
the weighting of objective probability reduces the subjective impact of high
probabilities and increases the impact of low probabilities. The x-axis is
objective probability and the y-axis is the weighted outcome. Applied to the
data in Exercise 8.4, the weighting reduces the appeal of A and increases
the appeal of C.

No mathematical expression has been given for the probability weighting

function; it is determined empirically. One partial explanation for the effect
is that a rule of diminishing sensitivity with distance from a reference point
applies. The probability function has two natural reference points, 0 and 1;
the weights, or relative differences between subjective and objective
probability, initially increase with distance from these anchors. However, as
these relative differences are in opposite directions, they must come
together again somewhere in the central area.

The weighting of small objective probabilities fits the evidence that people
are positive about insurance and like to place long-odds bets. Above an
objective probability of 0.40, the weighting depresses subjective probability
so that risks are subjectively discounted; for example, a 50 per cent
probability is nearer to 40 per cent, subjectively. At the extremes, the
weighting is unstable; a one in a thousand chance may be dismissed as no
chance or taken seriously (bettors on national lotteries accept very long
odds).

Figure 8.2 The probability function (Kahneman and Tversky, 2000)

The sensitivity near the reference point (0, 1) is illustrated by Exercise 8.5.
 Exercise 8.5 Risk preferences
Suppose that you have a 99 per cent chance of getting $1,000. How much would you
pay to move that probability to certainty?
Suppose that you have a 50 per cent chance of getting $1,000. How much would you
pay to move that probability to 51 per cent?
Suppose that you have a 0 per cent chance of getting $1,000. How much would you pay
to move that probability to 1 per cent?

Most people will pay more in the 0 per cent and 99 per cent conditions for a
1 per cent increment in probability, but the long-run benefit of a 1-per-cent
gain in probability is the same at any point on the probability range.
Prospect Theory
Kahneman and Tversky (1979) and Tversky and Kahneman (1992)
incorporated the subjective conversion of value and probability into a
theory of choice called prospect theory, and they proposed that the choice of
an alternative (prospect) is established in two stages. In the first stage, the
choices that are framed in a communication are restructured or edited by the
receiver. Then, in the second stage, the receiver chooses the best option
based on the values assessed in the first stage.

Framing refers to the manner in which the choice is presented to the

decision-maker, and editing refers to the processes used by the decision-
maker to rethink the choice. In framing, an alternative or prospect may be
presented either as a loss or as a gain, as in Exercise 8.6.
 Exercise 8.6 Life and death (Tversky and
Kahneman, 1981: 453)
An unusual disease is expected to kill 600 people. Two interventions are proposed. Which
intervention do you prefer on the basis of the following information?

If programme A is adopted, 200 people will be saved.

If programme B is adopted, there is 1/3 probability that 600 people will be saved and 2/3
probability that no people will be saved.

When you have decided, consider how you would react to these alternatives:

If programme C is adopted, 400 people will die.

If programme D is adopted, there is 1/3 probability that no one will die and 2/3
probability that 600 people will die.

The second pair of alternatives in Exercise 8.6 is the same as the first (A =
C, B = D). Yet Tversky and Kahneman found that 72 per cent preferred
programme A to B and 78 per cent preferred programme D to C. By
framing the problem in terms of the gains (lives saved), it is possible to
steer preference to the risk-averse option, A, rather than B. In the second-
choice pair, the framing in terms of lives lost makes people risk prone and
steers them to option D. Using the appropriate frame is clearly a ‘must’ for
anyone in the field of persuasive communication.

Thaler (1985) suggests some interesting implications of framing for those

presenting gains and losses to others. Losses are best presented in
aggregate, to minimize their impact, and gains are best presented singly to
maximize their effect. It may also be better to offset some losses against
gains because, separately, losses have more impact than gains. Framing
effects seem to apply to the presentation of discounts. A saving of $10 on a
$50 item may be presented as such or as a 20 per cent discount. When the
percentage discount is small, it may be best to present it as an absolute cost
reduction.

In the editing phase of decision-making, complex choices may be

simplified. Thus, a person may see the price of a car not as $20,000, but as
$2,000 less than he or she expected to pay; another edit by the prospective
buyer may be to aggregate the cost of extra features with the basic price so
that there is a single purchase price.
 Box 8.3 Overpaid?
The way bond traders are paid may help them to get very high remuneration. Bond
dealers typically work on commissions that are very low percentages. A commission of
0.1 per cent does not seem much. However, if it is based on a principal of $100 million, it
is $100,000. In general, people assess costs in proportional terms and will take more
trouble to save $5 off a $20 item than $5 off a $100 item.

In negotiation, our editing processes may lead us into a poor deal and we
should beware of the way the other side frames choices. It is wise to focus
on the total cost of a deal. Give way on small items but resist concessions
on the larger ones. It could even be worthwhile to include small items in a
proposal so that you can concede them later.

Thaler (1999) introduced the term mental accounting, which covers some of
these editing functions. This uses the metaphor of accounting to explain the
way in which people organize, evaluate and keep track of activities. As in
accounting proper, people maintain separate accounts for different
activities; for example, they may accept the idea of flying business class to
Australia and back at an extra cost of $6,000 but find it difficult to justify
spending $6,000 to reduce discomfort over two days in another area of their
life. Also, people may close an account after a defined period; gamblers on
the race track tend to think of gains or losses over a day and investors in the
stock market may operate with a one-year horizon.
Box 8.4 The Nudge Unit
Thaler and Sunstein (2008) have written a popular book called Nudge in which they argue
that national administration can be much more effective if it incorporates the known
findings from mental accounting and other psychological research. They focus on
activities such as paying taxes, investing in pensions and securing organ donation. These
ideas have caught the interest of governments. In the UK, the Cabinet Office is advised
by a ‘Nudge Unit’, which claims to have saved substantial sums of money by simple
changes such as the redrafting of letters to tax payers.
Section 4: Financial Applications of Prospect
Theory
Prospect theory provides explanations for some puzzling behaviour. We
now provide several examples of its explanatory power, drawing on a
review by Camerer (2000).
Investment
One of the effects of loss aversion is that investors may tend to hang on to
losers and sell winners in share markets. Investors may have some
equilibrium concept in mind, believing that the swing of the pendulum will
take a losing stock back towards its purchase price and that a rising stock
could similarly fall. They may also be reluctant to sell a losing stock
because this turns a potential loss into an actual loss, so that it is more
painful. From a rational standpoint, the buying price is sunk cost and should
not figure in the decision to sell shares, except for calculating any capital
gains tax. The selection of an alternative to sell should be based on the
prospective return, based on the current valuation of the share.

A bias in favour of selling winners and keeping losers has been

demonstrated experimentally by Weber and Camerer (1998) and in real data
by Odean (1998). Such behaviour appears to be a mistake since Odean
estimated that investors lose an average of 3.4 per cent in the subsequent
year by selling winners and keeping losers. However, this behaviour by
investors may contribute to the effective operation of share markets. These
markets are potentially unstable since a loss of confidence can precipitate
selling, which may further aggravate the loss of confidence and lead to a
collapse of the market. In practice, panic selling on a large scale is rare and
markets show corrections but do not usually collapse. If the market has
fallen, many investors will bide their time and refuse to sell. There seems to
be evidence that market professionals are much more willing to realize a
loss (see Box 8.5). They may operate with a momentum rather than an
equilibrium concept of the market.
Box 8.5 ‘Mutual-fund pros panicked in
peso crisis while small investors stood their
ground’ (Headline in the Wall Street
Journal, 13 January 1995)
Among share-trading professionals, the folklore is to stay in a rising market and sell when
it turns. Loss aversion suggests that small investors might find this difficult. In particular,
once they see that the market has fallen, and that they have made a loss by reference to
recent prices, they may resist selling because now they are risk prone. Also, by selling,
they close the account so that the potential loss becomes a real loss. The headline above
suggests that ordinary investors follow loss aversion but that the professionals have
learned to reverse it.
The Equity Premium
Equities have traditionally provided a greater return on investments than
fixed interest investments (bonds). Sometimes bonds give a return that is
little better than inflation or even below it, while equities have given an
average return in the USA of about 10 per cent per year over the period
1926–2003 and 9.6 per cent per year in the UK over the period 1900–2003
(Dimson, Marsh and Staunton, 2004). Traditionally, the equity premium has
been explained as a compensation for the greater risk associated with
equities but an alternative explanation has been offered by Benartzi and
Thaler (1995). A feature of equity markets is the much greater fluctuation in
value compared with bond markets. Investors who look at their equity
investment a month after purchase are about equally likely to see a gain or a
loss compared with the purchase price, but after ten years there is nearly
always a gain. Since losses loom larger than gains (the ratio of the dis-
attraction of losses to the attraction of gains is about 2.2), an investor with a
short-term horizon will suffer more pain from losses and may be put off
from investing in equities. Benartzi and Thaler, therefore, explain the equity
premium as the additional return required to compensate for the loss
aversion effect. This explanation would be more convincing if there were
evidence that those investors with longer-term horizons held a greater
proportion of their portfolios in equities rather than bonds.
Long-Shot Bias and the Value Premium
As noted, when the probabilities are low, the psychological weighting of
probability makes long odds more attractive. This seems to operate in
betting where there is a tendency to back outsiders or long shots – horses
with high odds that are unlikely to win. This may reflect the overweighting
of small probabilities. Bookmakers tend to offset risk so that
disproportionate betting on outsiders will tend to reduce the odds on these
horses and raise the odds on favourites. The result is that in the long run
long shots do worse than favourites. This effect becomes more pronounced
over the course of the day as bettors lose money. Their mental accounting
period is usually a day, and in order to end up ‘in the money’, they
increasingly need to win on a long shot, so still more money goes on such
horses. Under these circumstances, an each-way bet on the favourite for the
last race will make money, on average, even allowing for a 15 per cent tax
on winnings (Ali, 1977). If you want to test this out for yourselves, when
visiting the racetrack it is best to place the bet on the Tote which
compensates for weight of betting more effectively.

Long-shot bias, or something rather similar, may lie behind the persistent
value premium effect in share investment. Value or income stocks typically
give fairly high dividends and sell at prices that are rather lower than is
justified by their fundamentals. Value stocks are contrasted with growth
stocks that pay low dividends. With growth stocks, investors forgo current
income and hope for more rapid growth. By analogy, the growth stock is
rather like a long shot, and some companies, like Google and Apple,
emerge to give spectacular results. However, the fundamentals are just that
and Dimson, Marsh and Staunton (2004) find that, in the long run, value
stocks give a distinctly better performance, and show a premium of about 3
per cent per year over growth stocks. This effect is transnational. In a
comparison of 14 countries, only one country (Italy) showed a superior
return from growth stocks.
Impact on Economics
Economics is founded on rational assumptions. These include an
assumption that the accumulation of wealth drives individual behaviour and
that alternatives should be assessed against this wealth criterion. However,
it is apparent that individuals do not think about their wealth but instead
about gains and losses relative to norms defined by the context.
Furthermore, they operate with a number of accounts so that a gain in one is
not necessarily offset (in their minds) by a loss in another. This pattern of
behaviour violates the principle of fungibility, that money in one account is
as good as money in another account. Also, it is possible to construct
choices where preferences violate the principle of dominance. People may
prefer A to B and B to C, but C to A.

Thus, people do not think according to the canons of economic logic, and
traditional economics has been severely challenged by prospect theory.
Previously, economists have argued that wealth will increase most rapidly if
people act rationally and money is fungible. This is a normative theory.
They claim that those who do not apply normative economic principles will
lose out, so that self-interest or reinforcement will direct behaviour towards
a rational pattern. Thus, actual economic behaviour will be driven towards
the rational pattern. But if people do not act rationally, normative
economics needs to take this into account. The most rational policy is to
anticipate the irrationality of others and adapt to it. Therefore, a study of
everyday decision-making is needed so that the rational person (or policy-
maker) can exploit the systematic biases in the choices of others.
Problems with Prospect Theory
The possibility of using prospect theory to explain behaviour, choose
profitably and negotiate successfully has seized the imagination of
researchers but the potential may be reduced when the theory is more
widely evaluated. Van der Plight and van Schie (1990) gathered evidence
on risk aversion and risk proneness among European populations. Their
work confirmed Kahneman and Tversky’s findings, but the effects were less
strong. In addition, Leclerc, Schmitt and Dubé (1995) found that people
making decisions under risk are often risk averse about time loss when,
according to prospect theory, they should be risk prone. The value of time
has now been examined in a series of studies by Festjens and Janiszewski
(2015).

Many of the studies on prospect theory have used hypothetical and rather
artificial examples. In these studies, the judgement of the majority has been
used and it would be interesting to know more about those people who do
not fit the model. However, there is plenty of field evidence which supports
the theory – though often only after ad hoc assumptions have been made
about the mental accounting period and the separation of accounts.

The more fundamental problem in this field is the lack of an underlying

rationale for the finding that most, but not all, losses loom larger than gains.
Why is time different from money with regard to loss aversion? We have an
accumulation of important findings but no fundamental explanation. One
possibility is that the effects we see are related to the relative frequency of
different types of occurrence. We have seen that frequency affects
availability but it may have a wider relevance. If there are generally more
positive than negative events, reference points will tend to be based on the
positives. As a result, negative outcomes are more at variance with
assumptions and are therefore more disturbing. This is the explanation
offered by Fiske (1980) for the phenomenon of negativity bias (considered
in Chapter 11). However, more research is required before this provides a
coherent explanation.
Summary
This chapter is about the automatic mechanisms involved in recognizing,
evaluating, judging, investigating and deciding. One process that seems to
underlie a person’s thought processes is the use of schemas – cognitive
structures that are fitted to information to make sense of it. When the fit is
poor, people may give more attention to the stimulus until a fit is achieved.
Repeated exposure of a stimulus that is initially obscure often leads to
increases in the evaluation of that stimulus.

The evaluative response to stimuli is often faster than the cognitive

response, and these evaluative responses, in the initial stages at least, seem
to involve relatively independent processing. When we make choices,
attention is related to the evaluation of alternatives.

When making judgements, people use simplifying processes called

heuristics. They make more use of information that is more available – that
is discrete, eventful, recent or established through personal experience –
and they tend to neglect information dealing with persisting conditions.
More available information is also given higher probability.

We convert objective value and probability into subjective forms. The effect
of this is that people are risk averse on gains and risk prone on losses at
medium and high probabilities. The value function is steeper in the negative
region. This gives rise to loss aversion and makes people more reluctant to
part with things that they own.

The way choices are presented (framed) and the information processing of
receivers (editing) affect the way in which decisions are made. If the
framing is manipulated, people can be pushed towards particular
alternatives. People tend to think in terms of different accounts which may
be closed after different periods. Thus, mental accounting, loss aversion and
risk tolerance will affect the evaluation of prospects.

These processes produce effects that are contrary to axioms in economics

but they help to explain a number of puzzles. These include the preference
for selling winning rather than losing shares, a reluctance to fully invest in
shares as opposed to bonds, and a tendency to buy growth shares rather than
income shares.
Additional Resources
In 2002, Kahneman was awarded the Nobel Prize for economics – no mean
achievement for a psychologist. The lecture that he gave at the time is
available at:
http://nobelprize.org/nobel_prizes/economics/laureates/2002/kahneman-
lecture.html.

Much of the work referenced here has come from two volumes edited by
Kahneman, Slovic and Tversky (1982) and Kahneman and Tversky (2000).
In particular, read Thaler (1999) ‘Mental accounting matters’ (reprinted in
Kahneman and Tversky, 2000: 241–60). Kahneman (2012) has incorporated
many of the findings from this research stream in his book Thinking Fast
and Thinking Slow. This book brings out the practical implications for
decision-making and national policy in a thoughtful and accessible way.
9 Consumer Satisfaction and Quality
Learning Objectives
When you have completed this chapter, you should be able to:

Describe theories of consumer satisfaction/dissatisfaction (CSD).

Measure customer satisfaction and service quality.
Know the evidence on the relationship between satisfaction and company profit.
Understand research on consumer complaining behaviour.
Report on research on how consumers respond to service delay.
Overview
In the USA, Hunt and Day set up the first conference on consumer satisfaction in 1976
and work in this field grew rapidly. In 1993, Perkins noted over 3,000 academic
references relating to this area. In Europe, consumer satisfaction and dissatisfaction
(CSD) has received rather less emphasis than in the USA, and starting with the work of
Grönroos (1984), the focus has been more on the perception of quality, particularly with
regard to services. Customer satisfaction often depends on the quality of goods and
services, and therefore CSD research is closely associated with quality research. The
measurement of quality has been developed in the USA by Parasuraman and his
colleagues (Parasuraman, Berry and Zeithaml, 1991; Parasuraman, Zeithaml and Berry,
1985, 1988, 1994), but this work has been widely criticized. This chapter reviews the
field, describes the measurement of satisfaction and quality, examines theories of CSD,
reports research linking satisfaction and company profit, and sets out evidence on
complaining behaviour and service delays. This chapter relates to work on the effects of
CSD on customer retention/defection described in Chapter 2.
Section 1: Introduction
Most of the work on consumer satisfaction and quality is motivated by its
effect on loyalty. To become loyal when they have a choice, customers have
to be satisfied with the products and services they purchase and appreciate
their quality. Reichheld (1996b) and others have argued that it costs less to
retain existing customers than to gain new ones, and that quality helps to
retain customers (Chapter 2). Brand extensions are likely to be more
successful when the brand has higher perceived quality (Chapter 3). Buzzell
and Gale (1987) showed that when a firm’s quality was high, profit margins
were larger and firms could grow more easily. In addition, Hirschman
(1970) suggested that unsatisfactory delivery of goods and services led to
two types of consumer response, which he described succinctly as exit and
voice. Exit is switching to other products or suppliers. Voice has a number
of forms: complaining to suppliers and seeking redress, negative word of
mouth to other consumers (e.g. on blogs and in customer reviews on the
Internet), and occasionally, formal complaints through legal or trade
authorities. Interestingly, Ganesh et al. (2000) show that customers who
switch to a firm because of dissatisfaction with a previous service provider
have a propensity towards higher satisfaction with and loyalty to the new
firm, compared with this firm’s existing customers.

Because of the impact on the bottom line, especially in the long run,
companies want to diagnose satisfaction and quality problems as quickly as
possible by intercepting unhappy customers (see Box 9.1). Many now scan
the Internet to identify incidents. They also want to make an overall
assessment of their performance, benchmark it against industry standards
and competition, and track their performance over time to check if changes
in their way of operating have the desired effect. However, quantifying
dissatisfaction and comparing levels of satisfaction reliably across different
product or service categories or in different countries is not straightforward.
Standardized measures have been developed (mainly SERVQUAL), but
these have been criticized.

Some consulting organizations publish yearly barometers of customer

satisfaction on large representative samples of customers (e.g. the American
Customer Satisfaction Index; the Kundenmonitor for Germany, Austria and
Switzerland). These studies show that customer care still poses a lot of
problems and that customer satisfaction has not fundamentally increased
even after decades of company attention and investment. The 2013
Customer Rage Study (CCMC, 2013) is a replication of a national US
survey that was initiated by the White House in 1976 and is carried out now
in collaboration with Arizona State University (around 1,000 telephone
interviews on a nationally representative sample of households). The sixth
follow-up in 2013 observes a customer problem rate of 50%, an increase of
5% compared to 2011 and of 18% compared to the original White House
study. Complainant satisfaction has not increased since the start of the study
and 56% of complainants consider that they obtained nothing as a result of
their complaining. Customer rage, measured as the percentage of
respondents who were very or extremely upset (when thinking about a
serious problem over the past 12 months), is still at the same level as for the
studies run in 2003 and 2007.

Service quality and customer satisfaction have not been defined in a

consistent way in the literature. In many articles, the terms are used
interchangeably, although historically the concept of customer satisfaction
has been developed in studies of product marketing, while judgement of
quality is more prominent in services marketing. Both quality and
satisfaction are obviously forms of evaluation but they have a slightly
different focus (see Iacobucci, Grayson and Omstrom, 1994). Quality (what
is good?) is a logical predecessor of satisfaction (did you like it?). In that
sense, quality is more cognitive and satisfaction more affective. Satisfaction
requires some direct experience, while quality can also be judged on the
basis of external criteria or the opinion of other people. Quality is therefore
more an external judgement and satisfaction more internal and personal.

Much of the work on consumer satisfaction has focused on services.

Maintaining the quality of a service is difficult because of the nature of
services. A service is consumed as it is produced and any mistake by the
provider becomes part of the service delivered. In contrast, mistakes in the
production of goods can often be corrected before sale. Quality is usually
raised by uniformity in goods but attempts to make services uniform can be
counter-productive. Especially in high-end services, service providers need
to adapt to the needs of the customer and what suits one person may not suit
another; for this reason, service standardization can produce a rather
formulaic interchange with customers, which may not meet their
requirements. Getting services right is important because they constitute an
increasingly large fraction of modern economies. In most OECD countries,
services account for more than 70 per cent of GDP (OECD, 2000), a
number that has remained stable in most countries (World Bank, 2015).
Services also count for most employment growth in OECD countries
(OECD, 2005).
Box 9.1 How dissatisfying one customer
can become a PR disaster
In spring 2008, the guitar of David Caroll was witnessed being thrown by baggage
handlers of United Airlines in Chicago. He later discovered that this guitar that had cost
him $3,500 was severely damaged. United did not deny the incident but for nine months
the various people that Caroll interacted with put the responsibility on other employees
and did nothing in order to compensate for the damage. Caroll promised the last person
he interacted with that he would write three songs with accompanying videos. At the time
of writing this box, the first video had received more than 15 million views on YouTube
(‘United breaks guitars’). Caroll also wrote a book and is available as a speaker. United’s
website says: ‘In the air and on the ground, online and on the telephone, our customers
have the right to expect – to demand – respect, courtesy, fairness and honesty from the
airline they have selected for travel.’
Section 2: Theories of Consumer Satisfaction
The Confirmation Model
Early thinking about satisfaction treated it as meeting consumer
expectations. This is the confirmation model of consumer satisfaction,
which is illustrated in Figure 9.1. Oliver (1989) described the outcome as
contentment; for example, we are contented when a refrigerator continues to
keep food cold. This low arousal state is matched by low-arousal discontent
when negative expectations are met. Such discontent applies to the routine
use of inadequate services, such as congested roads, late buses and long
queues at airport security, and to unsatisfactory products such as dripping
taps, lumpy mattresses and toasters that eject the toast prematurely. In these
situations, the discontent is often subdued because of habituation. People
get used to a problem and no longer notice it. As a result, it does not occur
to them to do anything about it and any effect on behaviour is weak (shown
as a dotted line in Figure 9.1). Consumer contentment and discontent may
not be expressed but are revealed when people are questioned, or when
other factors raise the salience of a product’s performance; for example,
others may comment on the dripping tap or an ad for beds may make people
think of their own bedtime discomfort.

Satisfaction research conducted in this tradition gives consumers a list of

product or service features that are considered to influence satisfaction.
Consumers then have to rate the importance of these features and to rate
retrospectively the extent to which each of the features was effectively
delivered.

Figure 9.1 The confirmation model of consumer satisfaction: meeting

expectations
 Exercise 9.1 Recognizing dissatisfactions
Think of everyday products that you use, for example refrigerators, shoes and clothing. Are
these satisfactory? If you look at these products with a more critical eye, are there weaknesses
that could be corrected? Does your refrigerator ice up or fail to drain condensation? Are your
shoes too worn? Are you using clothes sizes (bras, collars, etc.) that are no longer appropriate?
Such frequently unrecognized problems often occur because what was once satisfactory has
slowly become less so – too slowly to be noticed. What should advertisers do to draw your
attention to needs that you have not recognized yourself?

Some problems persist for a long time before a solution is found. Many of these solutions
could have been invented earlier; what is often missing is the idea, rather than new technology.
Consider the wheels on luggage which are now so common. These could have been produced
years before they became common. Often, people may not have had the idea for the innovation
because habituation stopped them from recognizing a problem. Can you think of products that
we take for granted but which could be made better?

If consumers have little awareness of the shortcomings of everyday

products, they will feel little pressure to change their behaviour and this can
be a matter of concern. Our tendency to adjust to our environment may be
to our disadvantage. When poor products are frequently experienced, any
improvement would be frequently experienced too. Although people may
not notice deficiencies in currently used goods and services, they may well
notice and appreciate the change when the product is improved; thus, it is a
pity if habituation leads to an absence of complaint or a lack of effort to
find a better product. However, many of the discontents that we experience
relate to public services (e.g. transport and parking constraints), and here,
because of limited choice and influence, it may be difficult to achieve
change. Research has moved away from the confirmation model but we can
see this model relates to much consumer behaviour, particularly the
toleration of delay, which is considered later.
The Disconfirmation Model
The disconfirmation model is based on the general notion that people usually
make performance judgements by making a comparison to a standard. In the
disconfirmation model of CSD, this standard is provided by consumers’
expectations. The consumer is surprised by product features that are better or
worse than expected. The magnitude of surprise is related to the size of the
discrepancy between expectation and experience. Most research has focused
on the high arousal condition where experience of goods or services
disconfirms expectation, either by exceeding it and giving satisfaction, or by
falling short of expectation and causing dissatisfaction (see Figure 9.2).

Two additional determinants are shown in Figure 9.2 – we show these with
dotted lines because they are not part of the core disconfirmation model. The
perception of the performance of the goods or services affects satisfaction
directly; the better it is, the more we like it. Expectation also has a direct
effect. There is variation in the emphasis placed on the different factors
affecting satisfaction. Some researchers (e.g. Oliver, 1980, 1981; Swan and
Trawick, 1981) have emphasized expectations, while others (e.g. Churchill
and Surprenant, 1982; LaTour and Peat, 1979; Tse and Wilton, 1988) have
given attention to perceptions. Several studies (e.g. Oliver, 1980; Swan and
Trawick, 1980) found that satisfaction is influenced mainly by
disconfirmation. At odds with these results, Churchill and Surprenant (1982)
found that satisfaction with a video disc player was determined by perceived
product performance, and any disconfirmation had no additional impact on
satisfaction. It is therefore quite likely that the relative importance of the
different components of the model depends on the category.

Figure 9.2 The disconfirmation model: exceeding or falling short of

expectations
The disconfirmation explanation of satisfaction has gradually evolved.
Cardozo’s (1965) laboratory work is often cited as the first empirical
treatment of disconfirmed expectation and Howard and Sheth (1969: 145)
were among the first to suggest that people use standards of assessment in
judging products when they wrote that satisfaction was ‘the buyer’s
cognitive state of being adequately or inadequately rewarded for the sacrifice
he has undergone’. Oliver suggests that the emotion felt from
disconfirmation of expectations decays into the attitude to the product:

the summary psychological state resulting when the emotion

surrounding disconfirmed expectations is coupled with the consumer’s
prior feelings about the consumption experience. Moreover, the surprise
or excitement of this evaluation is thought to be of finite duration, so
that satisfaction soon decays into (but nevertheless greatly affects)
one’s overall attitude toward purchasing products. (Oliver, 1981: 34)
Attribution
Although, as we have seen, repurchase intentions are much reduced by
dissatisfaction, Oliver (1980), Oliver and Swan (1989), Feinberg et al.
(1990) and Fornell (1992) find that many consumers are reluctant to
respond effectively. For example, Feinberg et al. (1990) found that, after an
unsatisfactory warranty repair, repurchase intentions for different goods
were still at 47–84 per cent compared with more than 90 per cent
repurchase intention when the repair was satisfactory. Understanding the
mental process that leads to satisfaction or dissatisfaction may help
understand whether a failure will lead to exit or acceptance. A crucial factor
here is the attribution process used to interpret product or service failures.

Disconfirmations may be interpreted in different ways by consumers. The

model in Figure 9.2 therefore has an attribution stage where a causal
explanation is developed in reaction to the product experience that
generated the dis/satisfaction. This attribution will affect later behaviour by
the consumer. When consumers explain a positive experience as a chance
effect, they are unlikely to recommend the product, but if a negative
experience is attributed to neglect by the service provider, negative word of
mouth and complaint to the provider may ensue. Burns and Perkins (1996)
cover a wide range of possible responses in such situations. The attribution
that consumers make may be affected by other conditions, and in particular
the availability of explanations and types of causal inferences have been
studied. Exercise 9.2 relates to attribution.

Sometimes subtle cues can change the attribution that is made and therefore
influence the satisfaction with a consumption experience. Pham et al.
(2010) show that the presence of mirrors increases the likelihood that
consumers attribute service outcomes to themselves, rather than to the
service provider. As a result, unfavourable outcomes create less expressed
dissatisfaction. The downside of the phenomenon is that favourable
outcomes also lead to less expressed satisfaction. Self-awareness is the key
driving factor here: cues that increase self-awareness (e.g. the small talk of
a sales person centred on the customer) make self-attributions more likely.
The effect is shown both in experimental scenarios in a lab setting and in
real-life shopping situations. Interestingly, it also operates for the evaluation
of past experiences. A limiting condition is that consumers have to bear
some responsibility for the negative outcome (e.g. returning or exchanging
previously purchased items).
Exercise 9.2 Your experience
The apparently fresh Brie is acrid, the new vacuum cleaner blocks or the waiter in an
expensive restaurant is unhelpful – here the dissatisfactions arise because we expected a better
experience. Conversely, we may be pleasantly surprised and satisfied when expectations are
surpassed, such as when the roads are unusually clear, the fruit in supermarkets is ripe or the
plane arrives early.

Think back to the last time you were surprised by your experience as a consumer. Did the
surprise make you satisfied, dissatisfied or neither? How would you explain what happened?
What is the first reason that comes to mind and who was responsible? Are there other possible
reasons? Why did you come up with the first one first?
Availability
According to the availability heuristic, vivid events are more easily brought
to mind than routine occurrences and, in addition, these vivid events are
judged more probable than they are in reality (see Chapter 8). Folkes (1988)
gives an interesting example of how this can work. She asked people who
were approaching the escalators to their apartment in a six-storey building
how often the escalators broke down. The escalators only went to the fourth
floor so that those who lived on the fifth and sixth floors always had to
climb the stairs for the last part of their ascent. These people estimated that
the escalators broke down less often than the people who used the first four
floors, for whom an escalator failure (leading to the unusual event of
climbing the stairs) was more vivid. The distinctiveness of a product failure
raises its vividness, and the availability heuristic raises the perceived
likelihood that it will occur again (Chapter 8). This, in turn, raises
dissatisfaction. The supplier should, therefore, try to make failures less
distinctive. For example, if customers are occupied in some way when
service quality is reduced, they may form a less distinct memory of the poor
performance.
Causal Inferences
Weiner (1980, 1990) has examined the explanations given for success and
failure and has suggested three causal dimensions that are relevant to
consumer response: stability, locus of causality and controllability. Stability
occurs when the cause can be consistently attributed to a particular person
or feature of the environment; locus of causality relates to whether the
purchaser, the supplier or some other party is seen to be at fault; and
controllability reflects the ability of an agent to intervene and change
outcomes. An unstable negative event may not trigger negative attributions
by consumers; for example, an out-of-stock item may be seen as
exceptional (unstable) and unlikely to be repeated. It is, therefore, better for
a seller if the cause of their failure is seen as unstable by the customer. By
contrast, it is better to have stability in product success since this
encourages continued usage. Folkes (1984) suggests that, when failure is
perceived to be stable, a consumer will prefer to have a refund for a product
failure since a replacement carries the same risk as the original; if the
failure is seen as unstable, consumers will be more willing to accept a
replacement. The perception of stability may vary across customer
segments. Bolton (1998) examined defection from the service supplier as a
consequence of the failure of a mobile phone network. If the customer was
long-term, the failure was offset against past good performance and seen as
unstable, so the customer was disinclined to defect. Recent customers who
lacked this experience were more likely to see the failure as stable, blame
the supplier, and defect.

Weiner’s other dimensions also affect consumer response after product

failure. For example, with respect to locus of causality, customers may
blame themselves when they purchase a poor product, and therefore expect
no redress; but if they see the failure as the responsibility of the
manufacturer or retailer, they may then expect a replacement or refund. If
people feel that they have no control over their outcomes, they may feel
anger towards those who they think do have control, for example when
public transport services are inadequate. In another application of the effect
of controllability, Mohr and Bitner (1995) show that the perceived effort of
the service provider has a strong impact on satisfaction with a service
transaction because that effort is perceived as a controllable factor of the
provider.
Exercise 9.3 Managing satisfaction
Understanding the implications of research findings on CSD can help us create better goods
and services and manage the process better when things go wrong. Create a scenario in which a
major service failure is occurring and you are managing the service.

What are you going to say to customers? Consider the attributions they will make and the
response they will want. What are the customer expectations and how can you manage these so
that the failure is more acceptable? What compensation should you give, if any?
Section 3: Measuring Satisfaction and Service
Quality
Satisfaction and service quality are usually measured through surveys.
Surveys have the advantage of being easy to administer, direct, clear in their
purpose and high in face validity. Many measurement instruments are based
on the disconfirmation paradigm.
Satisfaction
Satisfaction with a product is experienced by a customer or an ex-customer.
In this respect, it is more restricted than the attitude to the product and
assessments of quality, which may be given by anyone, customer or not.
However, satisfaction with the product is an attitude; people are satisfied for
reasons that can be measured, as reported in Chapter 7, and they will give
an overall satisfaction like the global attitude described in the theory of
planned behaviour. This overall satisfaction can be measured by a question
such as:

Considering your main supermarket, would you say you are:

Very satisfied [7]

Quite satisfied [6]
Slightly satisfied [5]
Neither satisfied nor dissatisfied [4]
Slightly dissatisfied [3]
Quite dissatisfied [2]
Very dissatisfied [1].

However, as we saw in Chapter 2, if we are trying to predict what people

will do, a relative measure is better because it indicates the advantage of
one product over another. Once relativity is admitted, we have a choice of
comparators. In the American Customer Satisfaction Index (ACSI,
www.theacsi.org), Fornell et al. (1996) use three questions about recently
purchased brands that relate to overall satisfaction, expectancy
disconfirmation and perceived performance compared with ideal
performance (see Box 9.2).

Here, the latter two items are relative to two different comparators:
expectations and the ideal. For practical purposes, the ACSI is the average
of responses to the three items, though Fornell and his associates use
different weights for specific predictions. When a firm markets several
brands, a composite ACSI score for the firm can be derived from the scores
for each brand. The ACSI items are part of a larger set of questions that
cover complaints, loyalty, expectations, value and quality.
 Box 9.2 The American Consumer
Satisfaction Items
1. Please consider all your experiences to date with your main supermarket. Using a
ten-point scale on which ‘1’ means ‘very dissatisfied’ and ‘10’ means ‘very
satisfied’, how satisfied or dissatisfied are you with your supermarket overall?

Write in number (1 to 10) ……..

2. To what extent has your main supermarket fallen short of your expectations or
exceeded your expectations? Using a ten-point scale on which ‘1’ now means ‘falls
short of your expectations’ and ‘10’ means ‘exceeds your expectations’, to what
extent has your supermarket fallen short of or exceeded your expectations?

Write in number (1 to 10) ……..

3. Forget your main supermarket for the moment. Now imagine an ideal supermarket.
How well do you think your supermarket compares with that ideal supermarket?
Please use a ten-point scale on which ‘1’ means ‘not very close to the ideal’ and
‘10’ means ‘very close to the ideal’.

Write in number (1 to 10) ……..

Note: the ACSI methodology was developed in Sweden as the Swedish Consumer
Satisfaction Barometer. Now the method is used in many countries. Other countries that
have adopted ACSI methodology include Barbados, Brazil, China, Colombia, India,
Kuwait, Malaysia, Portugal, Saudi Arabia, Serbia, South Africa, South Korea, Turkey,
and Singapore and the UK.

There has been less research emphasis on the components of satisfaction –

the specific reasons why a product is liked or disliked. We can measure this
using the methods of the theory of planned behaviour discussed in Chapter
7, treating satisfaction as a reaction to a bundle of costs and benefits.
Westbrook (1980) tested this approach. He examined the way in which
customers combine the dis/satisfactions relating to a store and found that a
global measure of retail satisfaction correlated well with a simple addition
of the satisfactions and dissatisfactions customers felt about different
aspects of store service.
Quality
Whereas satisfaction measures focus on the global attitude, quality
measures tend to give more attention to the reasons why a service is
perceived as having high quality. The field of service quality measurement
has been dominated by an instrument called SERVQUAL, designed to
assess any service using one standard set of questions. It was developed by
Parasuraman, Zeithaml and Berry (1985, 1988). SERVQUAL measures
customers’ expectations of what firms should provide in the industry being
studied and their perceptions of how a given service provider performs
against these criteria. The 1988 version of the instrument contained 22
expectation questions covering such specific service facilities as up-to-date
equipment, visually appealing premises and polite employees. In 1991,
Parasuraman, Berry and Zeithaml modified the instrument slightly. They
changed two items and altered the wordings of some others; the negative
scoring on some items was removed and the wording of the expectation
measures was changed so that respondents were asked what an ‘excellent
service would provide’, rather than what ‘firms in the industry should
provide’. Box 9.3 shows the expectation items in the 1991 scale, applied to
telephone companies. A second set of questions (not shown) would deal
with perceptions about a specific telephone company.

Parasuraman, Zeithaml and Berry (1988) showed that these 22 items could
be allocated to five dimensions – tangibles, reliability, responsiveness,
assurance (knowledge and courtesy of employees and their ability to inspire
trust and confidence) and empathy (caring and individual attention to
customers). However, the five-factor structure has not been well supported;
Cronin and Taylor (1992) found only one general factor.

Some authors have fundamental objections to the approach of taking the

difference between perceptions and expectations and argue that a
perceptions-only approach is more appropriate. Cronin and Taylor (1992),
for instance, propose a performance-only measure that includes the
perception items of SERVQUAL but not the expectations. One paradoxical
implication of the SERVQUAL design is that lowering expectations should
increase service quality.
The idea that SERVQUAL will apply to a variety of services without much
modification has also been contested. Carman (1990) found that some
functions require additional measures for adequate explanation.
Koelemeijer (1992) and Finn and Lamb (1991) found that the SERVQUAL
instrument performed poorly in retail contexts. Dabholkar, Thorpe and
Rentz (1996) developed a scale for retail quality measurement which
customized the measure to the retail context. It is easy to argue that some of
the items are inappropriate for certain services – for example, religious
services having modern-looking equipment or a neat appearance in
academic settings. Thus, customization of SERVQUAL seems to be
required for application to different services and, although Parasuraman,
Zeithaml and Berry (1994) accepted this, the changes in questions may
need to be substantial.
Box 9.3 SERVQUAL expectation
components and classification (from
Parasuraman, Berry and Zeithaml, 1991:
446–7)
Tangibles
1. Excellent telephone companies will have modern-looking equipment.
2. The physical facilities at excellent telephone companies will be visually appealing.
3. Employees of excellent telephone companies will be neat-appearing.
4. Materials associated with the service (such as pamphlets or statements) will be
visually appealing in an excellent telephone company.
Reliability
1. When excellent telephone companies promise to do something by a certain time,
they will do so.
2. When customers have a problem, excellent telephone companies will show a
sincere interest in solving it.
3. Excellent telephone companies will perform the service right first time.
4. Excellent telephone companies will provide their services at the time they promise
to do so.
5. Excellent telephone companies will keep error-free records.
Responsiveness
1. Employees of excellent telephone companies will tell customers exactly when
services will be performed.
2. Employees of excellent telephone companies give prompt service to customers.
3. Employees of excellent telephone companies will always be willing to help
customers.
4. Employees of excellent telephone companies will never be too busy to respond to
customer requests.
Assurance
1. The behaviour of employees of excellent telephone companies will instil
confidence in customers.
2. Customers of excellent telephone companies will feel safe in their transactions.
3. Employees of excellent telephone companies will be consistently courteous with
customers.
4. Employees of excellent telephone companies will have the knowledge to answer
customer questions.
 Empathy
1. Excellent telephone companies will give customers individual attention.
2. Excellent telephone companies will have operating hours convenient to all their
customers.
3. Excellent telephone companies will have employees who give customers personal
attention.
4. Excellent telephone companies will have their customers’ best interests at heart.
5. The employees of excellent telephone companies will understand the specific
needs of their customers.

The computation of quality judgement from the questionnaire responses has

also raised problems. SERVQUAL uses the difference scores between
expectation and perception but expectations tend to be uniformly high and
show little variance. Although the difference measures correlate reasonably
well with an overall measure of quality (Babacus and Boller, 1992; Cronin
and Taylor, 1992; Parasuraman, Berry and Zeithaml, 1991), the low
variance in the expectation measures makes these irrelevant to the score so
that perception-only scores are equally predictive. For this reason, Cronin
and Taylor (1992) recommended that the measure be restricted to
performance perceptions (which they call SERVPERF), giving a
questionnaire of half the length. An added advantage of this approach is that
it avoids respondent boredom. Another alternative that offers the same
advantage is to measure the gap between expectation and perceived
performance with a single question that combines the two elements.
Babacus and Boller (1992) suggested a way of phrasing this and Box 9.4
shows the binary form and the combined measure. In a review of
SERVQUAL’s history, Smith (1995) concludes that few of the original
claims remain undisputed. The aim of generic measurement of service
quality is attractive, but this goal may be unobtainable because of variation
in the nature of services and because of a lack of agreement about concepts
and measures.
 Box 9.4 Alternative scales
1. SERVQUAL expectation and perception scales:

Firms in XYZ’s field should have modern-looking equipment:

Firm XYZ has modern-looking equipment:

2. Combined item:

XYZ’s modern-looking equipment:

Independent of how satisfaction and perceived quality are measured, a

majority of respondents usually indicate satisfaction, and the distribution of
responses is negatively skewed (a long tail to the left and a mode far to the
right) (see Peterson and Wilson, 1992, for an analysis of this phenomenon).
Section 4: Outcomes of Satisfaction and
Dissatisfaction
Does Increased Satisfaction Lead to Increased
Profit?
If managers can increase product and service quality and hence customer
satisfaction, a number of beneficial effects may follow for the supplier, as
detailed in Figure 9.3. These are more customers (either by retention or
acquisition), additional purchases per customer and a higher profit margin,
either through an increase in price or a reduction in costs per sale. However,
there may also be additional costs associated with innovation and quality
management. Thus, evidence is needed on whether an increase in
satisfaction is normally associated with increased profit and, if so, how this
comes about.

Figure 9.3 Routes to increased value

To investigate the consequences of increased satisfaction, the measure of

satisfaction developed for the American Customer Satisfaction Index (ACSI)
is often used. This is because the ACSI, and similar indexes in other
countries, provide comprehensive measures across a range of industries, and
in the case of ACSI for more than 200 specific companies. In a review of
studies using such measures, Zeithaml (2000) found a generally positive
association between quality and profit and suggested that this association
was mediated by improved customer retention. Using the Swedish Customer
Satisfaction Barometer, Anderson, Fornell and Lehmann (1994) were able to
demonstrate small increments in return on investment over five years as a
result of increases in satisfaction. These authors suggest that the connection
between satisfaction and profit could be mediated by a range of effects. They
cite greater retention, reduced price elasticity, a lack of interest in competitor
offerings, reduced costs for future transactions, reduced costs from failure,
lower costs of customer acquisition and advertising, new product advantage
through increased reputation of the firm, more customer recommendation,
greater willingness to try products and stronger relationships with suppliers.
These more specific explanations are largely covered by the aggregate
effects shown in Figure 9.3.

Anderson, Fornell and Mazvancheryl (2004) have shown that increases in

customer satisfaction are associated with increases in shareholder value, as
measured by Tobin’s q.1 Anderson et al. argue that higher customer
satisfaction may raise the bargaining power of a firm with its suppliers and
other partners because those partners value relationships with companies
who ‘own’ a valuable customer base. Increased bargaining power may assist
margin and sales and might thus contribute to shareholder value. Gruca and
Rego (2005) found satisfaction influenced shareholder value by both
increasing cash flows and reducing their variability. Fornell et al. (2006)
examined the growth obtained from investing in high satisfaction companies.
They found impressively high returns compared with standard investment
indices and argued that this extra gain was achieved without incurring
greater risk. An interesting additional finding is that news about ACSI
results does not move stock prices, meaning that investors do not seem to
factor in the relationship between satisfaction and firm performance. These
results were confirmed in a follow-up study by Fornell, Mithas and
Morgeson (2009) on more recent data. Aksoy et al. (2008) also found that
changes in the satisfaction score predict stock value changes, but only when
the economy is expanding. Anderson and Mansi (2009) made the connection
between satisfaction and the corporate bond market. They showed that firms
with higher customer satisfaction ratings (as measured by ACSI) benefit
from better credit ratings and lower debt costs and therefore obtain a
financing advantage. Luo, Homburg and Wieseke (2010) also checked the
impact of satisfaction on financial markets and analysed the relationship
between ACSI satisfaction scores, recommendations of financial analysts
and company value for about 100 firms over a 12-year period. Customer
satisfaction is informative for analysts because it predicts the growth and
volatility of future cash flows. Luo et al. (2010) found that firms with higher
levels of customer satisfaction receive more positive stock recommendations
from analysts. This effect is greater in more competitive markets and also
when the markets are more risky. The authors also showed that satisfaction
impacts financial returns, both directly and indirectly through the analysts’
recommendations.

Although there is little evidence favouring any specific pathway from

quality/satisfaction to increased value, there has been some tendency to
emphasize the retention route (Anderson and Mittal, 2000; Zeithaml, 2000).
This emphasis on retention may be misplaced, since a review of the
connection between satisfaction and retention generally found weak
associations (Chapter 2). In addition, Wangenheim and Bayón (2007)
showed that increased customer satisfaction does lead to new customer
acquisition via increased word of mouth. We suspect that this is the main
route to increased profit but definitive research is needed here. Word of
mouth is considered further in Chapter 12.
Complaining

Factors Affecting Complaining Behaviour

People are often reluctant to complain when they are dissatisfied with goods or
services. Andreasen (1988) and Stø and Glefjell (1990) both found that 60 per
cent of dissatisfied consumers did nothing, while Benterud and Stø (1993)
found that 95 per cent of those dissatisfied with their TV shopping did not
complain. In the Customer Rage Study, 50 per cent of the customers who
experienced a problem with a product or service complained. The degree of
dissatisfaction has a modest relationship with the likelihood of complaint (Day,
1984; Malafi et al., 1993; Oliver, 1981, 1987; Singh and Howell, 1985). Oliver
(1981) reported a correlation of about 0.4 between dissatisfaction and
complaining. It is possible that the availability of customer-comment websites
has increased complaining behaviour because it makes complaining easy and
impactful. The Customer Rage Survey shows, for instance, that posting on
social networks almost doubled between 2011 and 2013 (see Table 9.1).

Day (1984) and Singh and Howell (1985) have noted that complaining is
affected by how people explain product failure, their expectation of redress and
the likely time-cost and effort involved. Research on complaining needs to take
account of all the possible reasons that people might have for this behaviour.
We can cover these reasons by using Ajzen’s (1985, 1991) theory of planned
behaviour, as discussed in Chapter 7. This deals with three types of influence:

1. Expected outcomes: Hirschman (1970) suggested that complaining by

customers was related to the expected returns and costs. Positive
outcomes may include replacement, apology and better goods or service
 in the future, while negative outcomes may include lost opportunities,
wasted time and embarrassment. The perceived likelihood of success in
obtaining redress has been found to be associated with complaining in a
number of studies (Day and Landon, 1976; Granbois, Summers and
Frazier, 1977; Richins, 1983, 1987; Singh, 1990). Of the respondents in
the 2013 Customer Rage Study who complained, in the end only 50%
considered that the time spent complaining to the offending company was
worthwhile.
2. Normative influences:consumers may also be influenced by what they
believe others think they should do, even when these other people are not
present. Normative influences on complaining have not been studied
systematically, although Richins (1981) has noted instances where
consumers felt that they ought to complain.
3. Control factors:these are knowledge, skills, time and other factors that
can make complaining easier or harder. Examples are the ease of access to
key personnel, an understanding of the workings of the organization
causing dissatisfaction, and confidence about complaining. Control factors
help us to distinguish between those who complain and those who do not.
Two studies (Caplovitz, 1967; Warland, Herrmann and Willits, 1975)
found that non-complainers seemed powerless and had less knowledge of
the means of redress. Grønhaug (1977) found that there were more
complaints to a Norwegian consumer protection agency from citizens who
lived close to it. Grønhaug and Zaltman (1981) also recognized the
importance of resources such as time, money and confidence, and this
study is cited by Yi (1990) as important in showing differences between
complainers and non-complainers. A matter of concern is that vulnerable
consumers (the old, ill and disadvantaged) complain less than others
(Andreasen and Manning, 1990) and this is likely to be related to the
reduced control that such consumers have.

East (2000) used the theory of planned behaviour to investigate complaining in

the UK. He found that complaining was focused mainly on getting a refund or
replacement, standing up for one’s rights, doing what friends expected and
confidence about being able to complain. Such evidence helps us to see what
factors may encourage complaining. For example, in order to increase
confidence, suppliers should make it clear that they welcome complaints and
also explain a clear procedure for handling them.
The Benefits of Receiving Complaints
Fornell and Wernerfelt (1987) argue that, within cost limits, it is profitable to
gather and evaluate complaints from dissatisfied customers. Companies put
toll-free telephone numbers and e-mail addresses on goods packaging and
websites for this purpose. One reason for the profitability of receiving
complaints is that these supply information about product deficiencies, which
can then be corrected. A second reason is to gather further sales when the
complainer gets in touch. For example, the complainer may be advised that
there is a newer version of software or an improved design of outdoor clothing
and this information may lead to a further purchase. A third reason is that an
effective response to the aggrieved customer by the company may reduce
negative word of mouth to other potential customers, which could damage
sales. Finally, if the complaint is well handled, the company may be able to
stop defection or recover customers who have already left.

There is evidence that customers who are retained when they complain after a
service failure may be more satisfied than they would be without the failure.
This is called the service recovery paradox (SRP). One explanation is that, if
customers appreciate the efforts of the company to satisfy their concerns, they
may recommend it more and be more inclined to repurchase. Examples of SRP
were found by TARP (1979) and Gilly and Gelb (1982), but Solnick and
Hemenway (1992) found that those who complained about health provision
were over four times more likely to defect than those who did not. More
patronage after complaint may occur because the complaint handling resolved
the customer’s dissatisfaction and exceeded expectation, but it is also possible
that those who intend to remain with the supplier complain because they want
to benefit from any changes resulting from their complaining.

Magnini et al. (2007) used role-play studies to investigate the SRP. They found
that the effect occurred when the failure was not severe, the firm had not failed
the customer before, the cause of the failure was seen as unstable and the
customer believed that the company had little control over the failure. In a
meta-analysis, De Matos et al. (2007) found a significant SRP effect for
satisfaction but not for repurchase intention or word of mouth.

While receiving complaints may indeed present opportunities, those complaints

must also be addressed in a satisfactory way. This may not happen, which can
cause further damage in the form of negative word of mouth. The Customer
Rage Study found that word of mouth from dissatisfied complainants is almost
three times higher than the word of mouth communicated by satisfied
complainants.
Response to Delay
Delay in the delivery of service is a perennial feature of retailing and other
services. Indeed, it is an inherent liability of something that is produced and
consumed during an interval of time. Consumers wait for counter service in
post offices, for train tickets in booking offices and at the checkout in
supermarkets; they also wait for public transport and get held up in traffic
jams; they may have to wait to talk to someone on a telephone service line.
Buying items and making reservations by computer or mobile phone may
avoid waiting but that makes the occasions where we still have to wait stand
out more. For the individual, delay is frustrating and for the economy it is
wasteful because people waiting in line are neither producing nor consuming.

It appears that dissatisfaction with waiting can affect the total service
experience: the longer the wait, the lower the evaluation of the whole service
(Feinberg, Widdows and Steidle, 1996; Katz, Larson and Larson, 1991; Taylor,
1994; Tom and Lucey, 1995). Organizations may not see delay in the same way
as their customers. Feinberg and Smith (1989) found that the customer at the
checkout thought the average delay was 5.6 minutes when the staff thought it
was 3.2; the actual time was 4.7 minutes.

It is likely that people are less bothered by delay when it is expected (Maister,
1985). Supporting this, an experiment on bank queues by Clemmer and
Schneider (1989) showed that the more unexpected the delay, the more it was
disliked. This suggests that we can identify two types of dissatisfaction with
delay: a low involvement discontent when the delay is predicted and a high
involvement disconfirmation effect when it is unanticipated. As delays get
more common, they also become more predictable and, paradoxically,
consumers may put up with them more easily. This may be why Taylor (1994)
did not find that common delays were associated with significantly more
irritation.

People also tolerate a wait better if they know how long it is. The provision of
delay information is standard for airlines, and many bus and train service
providers display boards reporting the wait before the next train or bus arrives.
In some countries, mobile phone apps can be used to find out when the next
bus will arrive. Predictability seems much appreciated by consumers; for
instance, phone queues are tolerated better when delay information is provided.
People are less irritated by delay if they are occupied in some way. One
example here is the way in which Disneyland entertains its queues with
costumed characters; another is the provision of mirrors in places where people
have to wait, for example at lifts. Taylor (1995) used two groups in an
experiment to test the effect of occupying people when they had to wait. Both
groups were delayed by 10 minutes but only one was allowed activities to
offset the delay. In this study, the evaluation of the service was reduced by
delay but the reduction in evaluation was less when people were occupied. The
principle of occupying waiting customers also applies for download delays on
the Internet, where waiting times are not counted in minutes but seconds.
Providing additional visual content during a search and download makes the
wait feel shorter and reduces negative affect towards the wait. However, for
short waits, providing this additional content actually makes the wait feel
longer (Hong, Hess and Hardin, 2013). Taylor’s work also suggested that, if the
service provider was thought to have control, customers assessed the service
more negatively. This is consistent with Weiner’s (1980, 1990) suggestion that
people become more irritated when they believe that the service provider has
control of the situation. Cheng and Tsai (2014) confirm that mentioning that a
delay is due to ‘uncontrollable circumstances’ reduces perceived waiting time
during train delays.

Studies by Dubé-Rioux, Schmitt and Leclerc (1989) show that when delay
occurs at the beginning or end of a process, in their example a restaurant meal,
it was evaluated more negatively than mid-process delays. It was better for
waiters to take the order, and then impose the delay, than to wait before taking
the order. However, it is not always clear when a service starts and finishes. For
example, delays to air travel occur when going to the airport, at check-in,
baggage check, passport control, the departure lounge, in the plane before take-
off, before landing, at baggage reclaim, and again at passport control and
customs. People may think of this sequence as a number of services or just one,
and the way they regard it may affect how they tolerate delay at different
points.

Perceived Responsibility for Delay

Seawright and Sampson (2007) ran a video simulation of a visit to a bakery
with different scenarios and demonstrated that when the reason for a wait was
obvious the perceived waiting time was slightly smaller than when no
explanation was available. It did not matter whether the cause was the server
(not enough change in the cash register) or another customer (taking a lot of
time to find the necessary cash). However, other research shows the importance
of perceived responsibility for the wait. If customers see the service provider as
responsible, they are likely to be more irritated by the delay and complain more
(Taylor, 1995; Weiner, 1980). Folkes, Koletsky and Graham (1987) found that
customers became angrier and expressed more resistance to repurchase when
they felt that a delay was avoidable by management. Taylor (1994) also found
that those who blamed management were angrier about the wait.

In a study by East, Lomax and Willson (1991), respondents were asked whom
they held responsible when they were delayed at the checkout of a
supermarket. Table 9.2 shows that more of the people who blamed
management disliked waiting. The investigations of delay in post offices,
building societies and banks gave very similar results. We see from the
bracketed numbers in Table 9.2 that few people are self-blaming; they are
much more likely to blame other customers than themselves. This is consistent
with the actor–observer bias (Jones and Nisbett, 1972) – that an actor tends to
see others as choosing and therefore responsible for their behaviour, while the
actor sees the same behaviour in himself as caused by the situation (e.g.
circumstances or other people).

When serious delays occur, managers may receive much of the blame. Often,
the cause is accidental or brought about by other service providers on whom
the firm had relied. This suggests that management needs to distinguish
between the cause of the delay and responsibility for its consequences. Staff
should clearly accept responsibility for remedying a problem, but it will help if,
when appropriate, they can explain that they did not create the problem. For
example, the cause of an aircraft take-off delay may be that a service company
has failed to deliver airline food on time, a mechanical fault has been revealed
by standard checks before take-off, bad weather has delayed the plane’s arrival,
or airport congestion is worse than normal. In all these cases, timely
announcements may help to deflect blame.
Queues can be organized as either multiple-line or single-line. Sometimes it is
possible to use a queuing ticket system, which allows people to conduct other
business while they wait. The relative attraction of these three systems was
tested by East et al. (1992) in their investigation of bank and building society
use (see Table 9.3). The single-line queuing system is much preferred to
queuing at each counter, despite the fact that this introduces a slight delay as
customers make their way to a vacant counter. The advantages of single-line
systems are twofold: delays are approximately equal and therefore fair and they
are more predictable since excessive delays at one counter will not affect
waiting time much when there are several counters in use. A single-line system
seems particularly useful when the start of a service is imminent, for example
in train stations where there is a danger of missing the train. Work by Pruyn
and Smidts (1993) indicates that, although people may prefer the single line,
the queuing system is of less consequence than the duration of delay and the
quality of the waiting environment.
 Box 9.5 Delay management
Research on the response to delay helps us to understand how to manage this service problem
better. There are three approaches:

1. Operations management: where feasible, managers should try to avoid delays by

increasing/reducing service supply in line with demand. For example, supermarket
managers can count shoppers entering the store and open checkouts in advance of the
calculated demand; they can also try and use their fastest checkout operators at peak times.
2. Influencing demand: the second approach uses regulation and incentives to draw demand
away from busy periods and towards quiet periods. A reservation system is used for many
services and differential pricing may also shift demand, e.g. cinemas may offer reduced
ticket prices on a day when demand is low.
3. Perception management: the third approach is to try to ensure that the customer sees the
delay in a way that does least damage. Tom and Lucey (1995) suggest putting literature at
the checkout and suggest that this is a good location for free samples. Maister (1985) also
recommends entertaining the queue but the form of entertainment should be chosen
carefully; it is easy to irritate customers with mindless music on the telephone and irrelevant
advertising on screens set up for the queue. Such arrangements imply that the delay is
normal (stable) and management is interested in alleviating, rather than eliminating, the
discomfort of waiting.

Waiting as a Cue for Quality

Although our coverage of waiting has so far considered it as a negative experience,
we should not forget that consumers can also be attracted to service providers with
long waiting times. A long line outside a restaurant or nightclub can be perceived
as the promise of a great experience. Consumers often rely on the behaviour of
others for guidance, and other people waiting in line is an easily observable cue. In
a series of four experiments, with either scenarios or actual waiting experiences,
Giebelhausen, Robinson and Cronin (2011) show that a required wait can increase
purchase intentions and experienced satisfaction. They also examine different
moderators of this effect. It appears more pronounced when the product or service
is difficult to evaluate because of a lack of familiarity or objective criteria and
when quality, as opposed to convenience, is a decision criterion. While waiting in
line, consumers also infer that products are more valuable when others queue up
behind them (Koo and Fishbach, 2010).
Summary
Consumer satisfaction, dissatisfaction, quality perception and complaining
behaviour are important because they relate to profit via word of mouth and
repeat purchase. A particular emphasis in this field has been on the quality
of services because these are more prone to failure than goods and
constitute a large and growing proportion of modern economies.

Consumers are contented when products meet positive expectations and are
discontented with goods and services that show expected weaknesses, but
this fulfilment of expectations normally produces little arousal and,
therefore, little ‘exit’ or ‘voice’ by the customer. Much research has focused
on those occasions where product performance surprises the consumer (the
disconfirmation model). Surprise is arousing and causes more behavioural
response and, therefore, more potential impact on profit. There is some
variation in the way researchers describe the detail of the disconfirmation
model but there is agreement that the discrepancy between expectation and
perception is the main focus; expectations and perceptions may have
additional direct effects on dis/satisfaction. In addition, researchers now
recognize an interpretative phase in which consumers’ explanations of their
experience may modify their responses to it.

There are a number of consequences of satisfaction, some of which were

covered in Chapter 2. An important concern is the association between
satisfaction and profit. Companies that increase satisfaction have been
found to show subsequent increases in profit. This gain in profit does not
appear to be fully anticipated in the share price so that buying shares in
companies that show satisfaction gains appears to be a sensible investment
strategy.

When customers are dissatisfied, they often do not complain. Companies

can benefit from complaints being expressed and research has been
conducted to explain the motivation to complain. Delay is a general liability
of service provision that causes dissatisfaction. Unexpected delay (causing
disconfirmation) is particularly disliked. When it occurs, delay is tolerated
better if people understand why they are delayed and how long it will last.
Also, if consumers can occupy their time they are less dissatisfied. It is
better to start a service and then impose a delay than to begin or end with
the delay. In some cases, delays can be a signal of the success of the product
and therefore a cue for quality.
Additional Resources
For reviews of SERVQUAL, see Smith (1995) and Buttle (1996). For
evidence of the impact of satisfaction on shareholder value, see Anderson et
al. (2004) and Gruca and Rego (2005).
Note
1 A firm’s q value is the ratio of its market value to the current replacement
cost of its assets (Tobin, 1969). Thus, it is higher for those firms that are
perceived to be using their assets more effectively.
Part 4 Market Response
10 Consumer Response to Price and Sales
Promotions
Learning Objectives
When you have completed this chapter, you should be able to:

1. Describe how consumers process price information.

2. Explain how marketers assess consumers’ price sensitivity.
3. Evaluate the different tactics you observe in the marketplace to present prices and price
changes and evaluate price differentiation tactics.
4. Assess the effectiveness of promotions by distinguishing between possible sources of
extra sales.
5. Discuss the long-term effectiveness of promotions.
6. Assess the effectiveness of the customization of promotions for both Internet and
‘bricks-and-mortar’ stores.
Overview
Consumers verify the prices of products in order to make sure they buy at the right price
and, for expensive items, stay within budget. Buying at the right price means getting good
value for money, compared with alternative products, and this may involve buying at the
right time and place. Benchmark prices may be externally available but often consumers
have to rely on their memory of previously encountered prices to make comparisons.
Price memory, unfortunately, appears to be rather unreliable.

Marketers have developed different methodologies to estimate how consumers react to

price changes. Price increases, though often necessary and justified, are unpopular and are
often considered unfair. Consumer researchers have examined the conditions under which
perceptions of fairness or unfairness occur. Prices also influence the perceived quality and
even the experience of using the product.

Price decreases are often temporary, in the form of sales promotions. Access to scanner
data has given researchers the opportunity to develop sophisticated quantitative models to
evaluate promotion effectiveness. Combining a price cut with additional promotional
activities, such as in-store displays and advertising features, may have a synergistic effect.
Promotions usually have large short-term effects on sales, but there is doubt about their
long-term effectiveness and contribution to profit. Customization of promotions, to adapt
them to the characteristics and purchase habits of individual consumers or consumer
segments, seems to hold promise – for example, Waitrose’s ‘Pick Your Own Offers’.
However, initial results show that customization is not necessarily more profitable
compared to mass promotions, especially in ‘bricks-and-mortar’ retailing.
Section 1: Consumer Response to Price
Reference Prices
A price is rarely treated in isolation: prices become informative by relating
them to other prices, a phenomenon studied under the heading of reference
price. A price in combination with a reference price, whether accurate or
not, allows a consumer to decide if it is better to buy here and now or to
wait and buy elsewhere. When a product is cheaper than expected, it is
more likely to be purchased and vice versa.

The concept of reference price was introduced to marketing by Monroe

(1973). It was inspired by Helson’s (1964) adaptation-level theory which
states that stimuli are judged with respect to internal norms. These internal
norms represent the aggregate effect of past and present stimulation.
Reference prices can be formed during the shopping occasion itself, based
on prices observed, and are in this case known in the literature as external
reference prices (ERP). Consumers can also call on their memory of past
prices paid or encountered. In that case, they hold what is referred to as
internal reference prices (IRP).

Interestingly, most research on reference price does not ask consumers for
their reference prices directly but instead derives them from choice models
estimated on panel data (Winer, 1986). These models include as explanatory
variables not only the elements of the marketing mix, including price, but
also a reference price term. This term captures the difference between the
current price of the product and some mathematical function of past prices
(in the case of IRP) or currently observable other prices (in the case of
ERP). The fact that this difference variable helps explain choice better,
when compared to a model with just the price variable, is taken as evidence
that consumers make comparisons with reference prices. The wide
availability of individual-level scanner data has permitted an extremely
productive stream of research in this area (see the review in Mazumdar, Raj
and Sinha, 2005).

Researchers have examined a broad range of reference price models.

Hardie, Johnson and Fader (1993) compared a model with ERP and one
with IRP and found that, at least for orange juice, ERP was a better
representation of reference price. However, Briesch et al. (1997) showed
that IRP gave a better model fit in other categories. It is of course possible
that the use of IRP or ERP varies across consumers and situations.
Mazumdar and Papatla (2000) pursue this idea and show that there are
indeed segments of consumers that differ in the extent to which they use
IRP or ERP. They also show that IRP is more frequently used for more
expensive product categories, while ERP is more often used in categories
with a longer inter-purchase interval and greater frequency of promotions.

Although most research examines ERP and IRP as defined above, a number
of alternative definitions of reference price have been proposed. People may
compare with the price they usually pay, would like to pay or the price they
expect to pay, given the expected evolution of prices, or the price they
regard as fair. Winer (1988) even proposed eight different possible
definitions of price reference. Another qualification is that it is possible that
instead of holding precise reference price points, consumers actually have
price zones in mind. Kalyanaram and Little (1994) therefore estimate a
‘latitude of acceptance’ around the reference price. This is a zone in which
the consumer is indifferent to deviations between the observed price and the
reference price. As with most reference price models, they also find the
asymmetric price effect predicted by prospect theory (see Chapter 8), with
stronger negative responses to price increases than positive responses to
price decreases.
Exercise 10.1 Your price knowledge
Keep the till receipt for one of your trips to the supermarket. When you get home unpack your
products, and without looking at your receipt, write down the prices you paid. Indicate each
time whether you think you remember the price accurately or whether you are just making an
estimate. Also indicate how accurate you think you are (very accurate, rather accurate, not
accurate). Score your responses according to whether they are exactly right, within 10 per cent
or outside this range. Compare your results with others.

1. How easy or difficult was it to determine whether your knowledge is directly drawn
from memory? In other words, how good is your introspective access to the cognitive
processes you use to answer the questions?
2. Is there any evidence that frequently bought items are better remembered?
3. Do you have other hypotheses about why you remembered some prices better than other
ones?
4. How can these hypotheses be tested in a study?
5. Are students more or less likely to remember prices compared to other consumers? Why
is that?
Price Memory
If consumers use IRP to assess the attractiveness of an offer, it is interesting
to verify how accurate their knowledge of prices is. Dickson and Sawyer
(1990) found that less than half of US shoppers could give the correct price
for the item that they had just put into their shopping trolley. They mention
in their article that marketing executives and academics who attended
presentations of this study were surprised by how imperfect shoppers’
attention to, and retention of, price information was at the point of purchase.
Nevertheless, the finding was corroborated by Le Boutillier, Le Boutillier
and Neslin (1994) for coffee, but not for soda, where 71 per cent could
recall the exact price they paid. Still, using the same interviewing
procedure, Wakefield and Inman (1993) reported the percentages of
corrected responses ranging from 52 to 78 per cent for four product
categories.

These results from in-the-aisle price knowledge surveys have received two
types of reactions from the research community. Some researchers consider
that they bring reference price findings into question, because price
knowledge is much lower than most researchers intuitively expected
(Kalyanaram and Winer, 1995). A second type of reaction concerns the
interpretation of results. The Dickson and Sawyer results are often
interpreted as indications that price memory is poor. Vanhuele and Drèze
(2002) point out that what may be measured in these in-the-aisle surveys is
short-term memory, not long-term price memory.

Vanhuele and Drèze (2002) therefore examine how price information is

stored in long-term memory, and based on theories from numerical
cognition hypothesize that three types of coding are used: verbal (similar to
recording your voice), visual (photographic) and magnitude coding (you
remember that the price was, for instance, somewhere between 35 and 40).
They show in an in-store survey that recall questions measure only part of
price memory. Recognition memory is clearly better and some knowledge is
also only present in an approximate form in memory. The Dickson and
Sawyer results are therefore not necessarily in conflict with reference price
research because they only reveal part of the price knowledge available.
Overall, price knowledge surveys may suggest that the current price is of
little or no consequence to many purchasers but this simplifies the issue too
much. Although consumers may take prices on trust on many occasions,
they might, at other times, check on how much they are paying and react
against suppliers who are seen to overcharge. In addition, the price of a
limited number of goods may be used as a key to the overall value for
money offered by a store. Even if there is only a small segment of
consumers who examine prices carefully most of the time, this segment of
so-called price vigilantes (Wakefield and Inman, 1993) may be sufficiently
important for retailers to keep prices low.

Researchers have also examined how prices are stored in memory.

Although some price learning may be incidental and unconscious – and
therefore less available to recall (Monroe and Lee, 1999; Vanhuele and
Drèze, 2002) – other price learning may be intentional and conscious,
especially for motivated price-sensitive consumers. What if we are
motivated to learn prices? How good can we become at price recall? How
well is our cognitive system adapted to memorizing prices?

With a series of experiments, Vanhuele, Laurent and Drèze (2006)

examined how immediate memory for prices functions. Participants had to
keep prices for two or three products in a given category (DVDs, cameras
and confectionary) in memory for five seconds before feeding them back.
Analyses of the responses confirmed that the three types of numerical
coding discussed by Vanhuele and Drèze (2002) were used: verbal, visual
and magnitude coding. Most intriguing is the effect of verbal coding. If
people want to hold a price (or whatever other information) in their short-
term memory, they cycle it through a sub-system of their working memory.
Because the capacity of this sub-system is restricted to 1.5 to 2 seconds of
length of speech, prices that take longer to pronounce are less well
remembered. Also, consumers who speak slower have poorer immediate
memory, because of this capacity restriction. And consumers who do not
respect the official pronunciation of prices and use shortcuts instead (e.g.
twelve ninety instead of one thousand, two hundred and ninety) have a
memory advantage. An implication of this finding, not examined in the
article, is that, across regions, language differences may handicap some
consumers in their price memory. For the French, for instance, a price of 90
euros is ‘quatre-vingt-dix euro’, but French-speaking Belgians and Swiss
know it in a shorter form as ‘nonante euro’.
The Price–Quality Relationship
When comparing different products in a given category, consumers may use
price as an indicator of quality. In a review of over 40 studies, Rao and
Monroe (1989) found robust but moderate evidence that consumers indeed
use price as a proxy for product quality. When such inferences occur, they
may actually be mistaken. An analysis of US studies by Tellis and
Wernerfelt (1987) found a mean correlation between price and objective
product quality of only 0.27. Zeithaml (1988), in addition, only found mixed
support for a relationship between price and subjectively assessed quality. It
seems that the inference of quality from price is therefore rather weaker than
has been supposed. One important reason is probably that quality
perceptions obviously depend on a number of factors in addition to price,
such as the appearance of the product, the reports of others, the brand and
the store that sells the product. Consumers who believe that price and quality
are related may persist in this belief because they selectively focus on cases
that confirm their belief (i.e. they attend most to high-price/high-quality and
low-price/low-quality products). Nevertheless, Kardes et al. (2004) show
that, when consumers are motivated and have the opportunity to do so, they
may process belief-inconsistent information for price–quality inferences.

Inferences about product quality from price are not always easy to make
even when there is clear covariation. De Langhe et al. (2014) examine the
case where the evidence is asymmetric in strength. For some product
categories, low prices consistently lead to low quality but higher prices are
associated with varying quality. Consumers in this case tend to overestimate
the quality of high-end products. The opposite occurs when there is a strong
correlation at the high end but not the low end; consumers then
underestimate the quality of the low-end products.

Combining this review with that on reference price, we see that price can
influence acceptability through a comparison with reference prices that
highlight the economic sacrifice or through quality inferences. This gives us
the model shown in Figure 10.1. Notice that, as price rises, the acceptability
of the product is raised by the price–quality relationship but reduced by the
comparison against reference prices. Normally, we expect the reference price
effect to be stronger since otherwise sales would rise with increase in price,
which is not usually observed. In a series of experiments, Bornemann and
Homburg (2011) showed that when consumers evaluate a product for
consumption in the distant future, the price–quality relationship receives
more weight, while for consumption in the near future price is instead
interpreted as a sacrifice (for example, participants in the experiments had to
imagine the launch of an electronic book reader that would be available in
the university book store in two days or six months).

Figure 10.1 Price, acceptability and purchase

In a variation on the price–quality inference theme, researchers have shown

that price discounts can unconsciously reduce the perceived quality of a
product. Shiv, Carmon and Ariely (2005) demonstrate that consumers who
pay a discounted price for an existing energy drink that is thought to enhance
mental acuity are, after consumption of the drink, able to solve fewer word-
jumble puzzles than consumers who paid the regular price. Plassmann et al.
(2008) show that when we believe that a wine is more expensive, we
experience more pleasure while drinking it. They presented the same wines
at different prices. A 90-dollar wine was in fact presented twice, once at its
real price and once marked at 10 dollars. Similarly, a 5-dollar wine was also
presented at 45 dollars. Subjects in this study rated the pleasantness of each
wine but the researchers also used fMRI scans to measure activity in the part
of the brain involved in the experience of pleasantness. The scans showed
increased activity in that area for higher-priced wines, in addition to higher
ratings. The fMRI scans indicate that these ratings cannot be dismissed as
rationalizations and are the result of genuinely experienced pleasantness.
Section 2: Estimating Price Sensitivity
The sensitivity of sales to changes in price is usually quantified as price
elasticity. Price elasticity is the ratio of sales change to price change,
expressed in relative terms. Thus, if unit sales go up 20 per cent when the
price is cut by 10 per cent (a negative change), the price elasticity would be
–2. For some products, for instance prescription drugs, price changes
produce very little change in demand. Price elasticity is close to zero and
demand is called ‘inelastic’. Conversely, the price elasticity for any supplier
in a commodity market is highly ‘elastic’ and the absolute value of price
elasticity is very high, indicating that sales will drop steeply if price is
increased and will increase considerably when price is dropped. It is
important to keep in mind that price elasticities are not the same at all price
levels. A price elasticity estimation is therefore only valid for the region of
prices used for its estimation. Price elasticities can, of course, also change
with general economic conditions or changes in specific market factors (for
instance, the development of new distribution channels such as the
Internet).

In consumer markets, three methods are used to estimate price sensitivity.

For new products, customer surveys are often used. For new and established
products, price experiments can be run. When historical information on the
market is available, econometric modelling can be applied. The wide use of
checkout scanners and other forms of information technology has permitted
detailed recording of transaction data. The result has been an enormous
increase in the ability of researchers to evaluate price effects.
Customer Surveys
There are two approaches to surveying consumers to assess their price
sensitivity. In the direct-survey approach, consumers receive a product
description or are exposed to the product itself, and then have to react to
potential prices. This method, developed in the 1960s, is simple to apply
and easily understood by respondents (Gabor and Granger, 1961, 1966;
Wedel and Leeflang, 1998). There are a number of biases that typically lead
consumers to overstate their willingness to pay, but corrections are possible.
An important drawback is, however, that price is treated in isolation in these
surveys while, in real-life choices, product characteristics have to be
weighed against price.

Conjoint analysis (Green and Krieger, 2002) is a widely used customer

survey method that presents consumers with simplified product descriptions
and asks them for their preference among these products while considering
all information conjointly. Price sensitivity and the sensitivity to changes in
other attributes are then inferred using estimation methodology. Conjoint
analysis is widely applied by commercial market researchers for new
product designs in general, not just for price analysis. Running conjoint
surveys by computer and on the Internet has now become standard practice.
A large number of applications have been published (see reviews in Green,
Krieger and Wind, 2001; Wittink, Vriens and Burhenne, 1994).
Price Experiments
The objective of an experiment is to control factors that influence the
outcome of interest (usually sales) but which are not the object of the study.
Sales are not only influenced by price changes but also by all the other
marketing actions of the firm, those of its competitors, and changes in the
context. Running a price experiment in one store and comparing sales at the
end of the experiment with sales before is therefore rarely conclusive; other
factors may have changed as well and have contributed to the observed
change in sales.

The art of running a good price experiment is in selecting a good

comparison basis, usually called the control group, which cancels out the
influence of other factors. Price experiments can be carried out in a
simulated shopping environment, but large market research companies,
such as Kantar, Nielsen and GfK in Europe, have test cities in different
countries in which they run experiments with consumer goods on a regular
basis. All the variables of the marketing mix can be controlled during these
experiments (for an example, see Fader, Hardie and Zeithammer, 2003).
Price experiments can also be conducted on the Internet (see Box 10.1).

Ehrenberg and England (1990) investigated price elasticity using a field

experimental method. Staff made fortnightly home visits to housewives and
offered a limited selection of cereal, confectionery, soup, tea and biscuit
brands for sale at prices that were a little below those in local supermarkets.
After two visits, the prices of some brands were raised or lowered. The
order of price changes was altered for different sub-groups so that any
effects based on price sequence could be detected. The authors found that
the response to price changes was immediate and was unaffected by the
order of earlier changes, i.e. it was zero order. Price increases had slightly
less percentage impact on sales than decreases. The mean elasticity
obtained by Ehrenberg and England (weighted by brand size) was –2.6.
Box 10.1 Price experiments on the Internet
Technically, the Internet allows price experimenting on a constant basis. In practice,
consumer reactions may complicate things (but see Exercise 10.2). In early September
2000, Computerworld reported that consumers logging on to Amazon at about the same
time could be charged very different prices for the same DVD. For example, at 2.40pm a
search for the Planet of the Apes DVD on the Amazon site using a Netscape Web browser
turned up a quoted price of $64.99 – 35 per cent off the original price of $99.98 – but
several seconds later, a similar search performed with Microsoft Corp.’s Internet Explorer
browser resulted in a price of $74.99 for the same product. A company spokesperson
acknowledged that Amazon was running price experiments: ‘Some customers will pay
the same for a certain item as customers paid last week, some will pay more and some
will pay less’, she said. In a statement several weeks later, Amazon formally denied
published rumours that the price differences in the test were based on customer
demographic information, and said that the price reductions, which were in the 20–40 per
cent range, aimed at determining how much sales were affected by lower prices. Amazon
also promised that, if they ever were to run such a test again, they would automatically
refund customers who purchased at a price higher than the lowest test price for the same
item.
Econometric Estimation
Econometric estimation of price elasticities has a long history and research
has sought to combine the knowledge accumulated over decades of pricing
research. A meta-analysis by Tellis (1988b) brought together 367
elasticities, drawn from 42 prior studies using a variety of estimation
methods. Tellis showed that a number of market factors were related to
elasticity and that the mean elasticity across all the studies was –1.8.

A more recent meta-analysis by Bijmolt, van Heerde and Pieters (2005)

found a substantially larger average of –2.6, estimated on 1,851 price
elasticities drawn from 81 studies. They examined the evolution over time
between 1961 and 2004 and observed that the (absolute) sales elasticity
became larger each year. They also compared different market and category
characteristics and found that price elasticities have greater magnitude in
the introduction and growth stages of the product life cycle compared with
the maturity and decline stages. Consumers are more price sensitive for
durables than for other products. Inflation also increases the magnitude of
price elasticity. The authors found no effects of country, income, data
source (firms, retail panels, household panels) and brand ownership (private
labels versus manufacturer brands).
Section 3: Psychological Reactions to Prices and
Price Changes
Perceived (Un)fairness of Prices and Price
Changes
What is considered as fair or unfair? Fairness is a judgement about the
justness, reasonableness or acceptability of an outcome (the price) or the
process to reach the outcome. It is a judgement that is induced by a
comparison with another outcome, such as ‘I paid more than another
customer did’ or ‘I paid more than I usually do’ (Xia, Monroe and Cox,
2004). Jin, He and Zhang (2014) show that the degree to which one feels
powerful plays a role in the type of comparison made. Low-power
individuals are more influenced by self-comparisons with purchases they
made in the past, while high-power individuals focus more on price
disadvantages relative to other consumers.

Charging different consumers different prices for the same or similar

products is a common practice. Economists refer to this practice as price
discrimination. Happy hour, senior and family discounts, low-season deals
are among the many examples of price discrimination. As we will see
below, many consumer promotions act as price discriminators, because in
practice it is mostly price-sensitive consumers who use them. There is
surprisingly little research on how consumers react to price discrimination
(see Exercise 10.2), although it is clear that some practices are considered
as unfairly discriminatory (see Box 10.2).

Most research on price fairness examines responses to price changes.

Consumers can react to these changes by purchasing now – and even
stockpiling – if the current price is becoming more attractive, or by
cancelling or postponing a planned purchase if the current price becomes
less attractive. They may also develop perceptions of fairness or, more
commonly, unfairness about the seller. These perceptions can lead to
negative attitudes about a company or an entire industry (e.g.
‘pharmaceutical prices should be controlled more strongly because of the
excess profits of the industry’), to boycotts and negative word of mouth.
Kahneman, Knetsch and Thaler (1991a, 1991b) identified, in a series of
surveys and experiments, the conditions of price change that consumers
consider as fair or unfair. They developed the principle of dual entitlement
as an explanation, arguing that fairness perceptions are determined by a
belief that firms are entitled to a certain profit and customers are entitled to
a certain price. Increasing prices to increase profits when there is heavy
demand is considered unfair, but it is fair to protect profit from rising costs.
The classic example is that a retailer is entitled to raise the price of snow
shovels in reaction to an increased wholesale price, but not in response to
extra demand brought about by a snowstorm.
 Box 10.2 The woman tax
On January 29, 2016, the French newspaper Libération ran a cover article on ‘The cost of
being a woman’. Even for everyday products and services, women pay more than men for
equivalent offers. The examples given are haircuts, razors (€15.95 vs. €11.95 for a pack
of eight of the same brand), deodorant and dry cleaning. An article on the website of
Fortune in 2015 (http://fortune.com/2015/09/23/what-is-the-woman-tax/) refers to a study
in Florida that surveyed 100 unisex hair salons and found that women paid on average
$35.02 for a basic haircut and men $22.78. The salons defended their price differences,
saying that women’s cuts ‘take longer’, ‘are more fussy’, ‘are more difficult’, and that
women ‘expect more’. In 2009, the American feminist Janet Floyd ran the following test:
she and her husband took the same shirt to the cleaner. The price for her was $8.75, while
for him it was $7.

Campbell (1999) expands on the dual entitlement principle and identifies

two important factors that affect the perceived fairness of a price change:
consumers’ inferences about the firm’s motive for a price change, and
inferred profit, relative to the past. The core idea is that consumers make
inferences about a marketer’s motive or intention for a particular price
increase. This idea is based on evidence from attribution research that
people search for causal explanations for negative and unexpected events.
When consumers infer that the firm has a negative motive (e.g. exploiting a
sudden increase in demand), the increase is considered unfair. When the
firm has a good reputation, consumers give it the benefit of the doubt. Even
when the scenario suggests a negative motive, if the firm does not appear to
make extra profit, consumers infer a more positive motive.
 Exercise 10.2 Price discrimination
Yield management is a pricing practice in which prices are adapted over time to optimize
profits and capacity usage. It is used in industries where capacity is fixed, such as the hotel and
airline business. Marginal revenue at a low price may be worthwhile if the room or seat would
otherwise be empty, but not if it forces a reduction of the overall price level. So, while tourists
might get a concessionary rate on the weekend, because they book well in advance and have
flexible travel times, this should not undermine the full price charged to business travellers. It
is now commonly accepted that, on any given flight, passengers pay widely varied prices for
their tickets as a result of the use of yield management pricing software.

Pick with a friend a couple of flying destinations that are served by a low-cost carrier (like
Ryanair, EasyJet or Virgin Blue). Agree on a departure date and hour and check the prices of
your flights a couple of times over the next week. Do prices change? By how much and in
which direction? Why do they change? Under what conditions would you find it acceptable if
your friend found a lower price than you for a given flight? What if you both logged on at the
same time and got a different price quote?

Bolton, Warlop and Alba (2003) examine, in ten different studies, how
information about prices, profits and costs influences consumer perceptions
of price fairness. They show that consumers underestimate inflationary
trends, even when provided with explicit quantitative information, and
therefore overestimate the profits that sellers are drawing from price
increases. When comparing the price of the same product in different types
of stores (such as a department store versus a discount store), they tend to
attribute price differences to different profit motives instead of to different
cost levels. Some marketing strategies are considered unfair, even when
they are not under the store’s control. When they are given information
about the cost structure of a firm, consumers tend to focus on the cost of
goods sold and ignore other cost categories. In conclusion, unfavourable
comparisons seem to dominate: consumers apparently have a tendency to
believe that the selling price of a good (or service) is substantially higher
than its fair price.
Framing of Price and Price Reductions
Sometimes prices can be presented in different ways in order to make them
look more attractive, a practice that is referred to as ‘framing’. Likewise,
temporary price reductions can be presented in different formats, such as
‘up to 50% off’, ‘30–50% off’ or ‘buy two get one free’. Rationally
speaking, consumers should be indifferent to different frames that result in
the same cost. However, Chapter 8 showed how framing can affect people’s
judgements. The frequency of use of price framing suggests that it is an
effective pricing tool.

To encourage subscriptions, magazines present their per-issue subscription

price. Membership fees of clubs can be framed in terms of the amount per
day (only $2.50 per day) instead of the actual total amount to pay ($912).
Internet-based suppliers may separate packaging and delivery ($4.95) from
the cost of the product ($24.95). Charitable donations explain the benefits
you can bring to children in poverty ‘for less than a dollar per day’.
Gourville (1998) posits that consumers evaluate unfamiliar single-
alternative transactions by comparing them to known transactions that
involve similar expenses. Different frames therefore foster the retrieval of
different comparison bases and influence the evaluation of the offer and
compliance. The ‘pennies-a-day’ frame, as Gourville describes it, triggers
comparisons to small ongoing expenses like buying a cup of coffee or a
train ticket, which makes the transaction more acceptable.

The effect of the face value of foreign currencies can be considered as an

extreme example of framing. Consumers confronted with prices in foreign
currencies that have an exchange rate that is a multiple of their own
currency tend to underspend because the price looks higher. The reverse
happens when the exchange rate is a fraction of the consumers’ currency
(Raghubir and Srivastava, 2002). This effect occurs even though the
exchange rate is provided.
The Effect of Price Endings
Prices often fall just below a round number, a practice referred to as odd
pricing. A price of, for instance, $10 is converted into $9.99. According to
some counts, between 30 and 65 per cent of all prices end in the digit 9
(Schindler and Kirby, 1997; Stiving and Winer, 1997). The clearest
demonstration of the effect of 9-endings on sales comes from Anderson and
Simester (2003), who varied prices on identical items in different clothing
catalogues that were sent to tens of thousands of randomly selected
customer samples. In the different experiments, demand for the items with
9-ending prices increased by 7 to 35 per cent. The effect was stronger for
new items.

There are two dominant explanations of the phenomenon. A first theory

posits that consumers ignore the right-most digits or at least do not give
them sufficient weight because of left-to-right reading and processing. A
price starting with a 2 therefore looks a lot smaller than a price starting with
a 3, even if the complete price is 2.99. Thomas and Morwitz (2005) indeed
show that the phenomenon only occurs when the leftmost digit drops by a
unit (3 to 2.99) and not when it remains unchanged (3.60 to 3.59).
According to a second theory, the 9-ending signals a good deal or a
promotion. Schindler (2006) analysed retail price advertisements and found
that 99-ending prices are much more often used in advertisements that carry
other lowprice cues. He hypothesizes that consumers’ repeated exposure to
this form of advertising leads to an association between 99-ending prices
and low prices. Manning and Sprott (2009) compared choice situations
between two alternatives with round or 9-ending prices. A round price
difference between 2 and 3 dollars can be presented as (a) 2.00 and 2.99, (b)
1.99 and 2.99 and (c) 1.99 and 3.00. The lower-priced option attracted the
highest choice share with frame (c), and the lowest share with frame (a).

Although odd pricing is very common, recent research by Wadhwa and

Kuangjie (2015) has shown that round prices can also have interest because
they influence product evaluations and preferences through a fluency effect.
A round number (e.g. €200) is more fluently processed and increases the
likelihood of relying on feeling-based inputs, while an unrounded number
(e.g. €198.76) orients consumers to cognitive inputs. In addition, the nature
of the decision as such can be feeling or cognition based. When a purchase
decision is driven by feelings, rounded prices intensify evaluative
judgements; when it is driven by cognition, unrounded prices intensify
cognitive judgements.
Surface Effects
The form in which a price is presented can also affect how it is perceived.
Some presentations can unconsciously induce a perception of higher
magnitude. Coulter and Coulter have examined font size (2005) and the
sound when pronouncing, possibly sub-vocally, the price. Coulter and
Coulter (2013) show the effect of oral rehearsal on the perception of
discounts. Front vowels (e.g. ‘ee’ in ‘three’) lead to an overestimation and
back vowels (‘o’ in ‘two’) to an underestimation of sale prices. The authors
demonstrate the effect both within the same language and across different
languages.
Section 4: Consumer Response to Sales
Promotions
The effectiveness of sales promotion started receiving attention in the 1980s
when the use of promotions increased considerably with the arrival of
scanner data. Before the introduction of checkout scanning, companies had
to rely on store audit data, collected by auditors on a sample of stores,
which were only released on a bimonthly basis. These data did not have the
detail necessary to see the real effect of promotions that are usually run on a
week-to-week basis. When weekly data became available and marketers
realized the full size of the boost to sales, they started investing much more
in promotions. Promotions usually have a clearly identifiable impact on
sales. Graphs of sales over time show a sales spike after a promotion
launch, a phenomenon that is much less apparent for advertising (unless, of
course, the advertisement communicates a promotion).

Promotions can take many forms but promotional campaigns are usually
short-lived (see Box 10.3). The distinction has to be made between
promotions offered by retailers and manufacturers to consumers, and trade
promotions offered by manufacturers to retailers. Examples of trade
promotions are co-op advertising funds and display allowances. The retailer
may or may not pass on cost savings to the consumer (called the ‘pass-
through’).
 Box 10.3 Types of promotion
Direct price reduction – also known as ‘discounts’ or, in the USA, as ‘deals’.
Couponing – refers to the distribution of certificates that can be redeemed for a
discount when purchasing.
Rebates or cashback – price refunds that can only be obtained after the purchase
by mailing an application, for instance.
Display – refers to a special in-store display that attracts attention.
Feature advertising – refers to stand- alone circulars, also known as flyers, that
consumers receive in their postbox. This is a form of cooperative advertising in
which manufacturers pay retailers to feature their products.
Games and contests.
Multibuy – for example, three for the price of two.
Extra quantity – for example, 10 per cent extra length on a chocolate bar.
Bonus offers – buy product X and get product Y free.

Most research focuses on price promotions, feature advertising, display and

couponing. The interest of promotions has been questioned. Initially,
advertising agencies tended to dislike the sudden interest in promotions
since they may take part of the ad budget, although this attitude has
somewhat changed now that many advertising agencies have converted
themselves into full-service agencies that handle all forms of
communication, including promotions. Supplying companies also have
mixed feelings about promotions because they have to carry the costs of
administration, pack changes and the production and inventory costs
associated with peaks and dips in demand. On top of this, a successful
promotion may bring retaliation from competitors, which damages later
profits. In aggregate, the effects of competing sales promotions may cancel
out, leaving a cost that has to be added to the price of goods. Maybe all
players (except the sales promotion agencies) would be better off if
discounting in mature markets was limited (see Exercise 10.3).

Much of our evidence on sales promotions comes from the USA, where
marketing practices are rather different from those applying in Europe. In
the USA, we find that:

more lines are normally offered on deal

discounts are usually larger
promotion periods are usually a week – shorter than in Europe – and
they are often driven by short-life coupons. In other countries, coupons
are less popular.
Exercise 10.3 Coupon war
In autumn 2002, Tesco and Sainsbury became involved in a coupon war in the UK. Sainsbury
had issued coupons on its home delivery service, offering, for instance, discounts of £2 for
every £20 of shopping. Tesco reacted by announcing that it would itself honour these coupons
in Tesco stores. In retaliation, Sainsbury printed thousands of coupons to send to customers in
areas where it was not present, hoping that customers would redeem them with Tesco. Its
objective was to decrease Tesco’s margins. A Tesco spokesperson replied: ‘It is usually our job
to promote shopping with Tesco, but if our competition want to do that as well, that is fine with
us … The risk is that by encouraging consumers to get discounts with us, it actually makes
them more loyal to the Tesco brand.’

Analyse the moves of the two players. Were there alternatives? Under what conditions is there
an interest in this type of escalation and for whom? What would you do next if you were
Sainsbury?
Sources of Extra Sales
Observing a promotional sales spike or bump does not necessarily mean the
promotion was effective. Whether a sales promotion is beneficial, from a
managerial perspective, depends on whether the additional sales can be
attributed to brand switching (usually good news for the manufacturer, but
the retailer’s evaluation depends on the relative margins of the competing
brands), category expansion effects (even better news, both for
manufacturer and retailer) or purchase acceleration and stockpiling (usually
mixed news because this can be considered as ‘borrowing’ sales from the
future). Researchers therefore have built sophisticated econometric models
to decompose the promotional bump into these three sources of sales.

Initial results were conflicting, with some researchers finding that most
promotional sales volume comes from brand switchers (e.g. Gupta, 1988;
Totten and Block, 1987) and others observing that category expansion is a
more important source of additional sales (e.g. Chintagunta, 1993). One
possible explanation of the contrasting findings is product category
differences. When comparing category expansion, brand switching and
purchase acceleration, Pauwels, Hanssens and Siddarth (2002) found a
breakdown of 66/11/23 for a storable product (canned soup) and 58/39/3 for
a perishable product (yoghurt). Van Heerde, Leeflang and Wittink (2004)
found that each source contributes to about one-third of the sales bump on
average; they used four products, two from an American and two from a
Dutch data set.

As already mentioned, stockpiling is considered the least interesting

contributor to promotional sales, because it is assumed that consumers
would otherwise buy the brand later, at the regular price. However,
Ailawadi et al. (2007) demonstrate that stockpiling can also have benefits in
two forms – category consumption can increase because consumers have
(more of) the product at home, and the extra inventory of the promoted
brand may pre-empt the purchase of a competing brand. They observe that
the first benefit is the most important and that, together, the two benefits
offset the downside of stockpiling. Chan, Narasimhan and Zhang (2008)
show that brand-loyal consumers stockpile to benefit from promotions
while brand switchers do not. Heavy users stockpile more.

Price promotions are an important competitive tool for retailers. In a large-

scale experiment over 16 weeks in 86 stores across 26 product categories,
Hoch, Drèze and Purk (1994) compared an everyday low price (EDLP)
strategy (more or less constant low prices) with a Hi-lo strategy (higher
prices but with frequent promotions). For the EDLP condition, they lowered
prices by 10 per cent, while they increased those for the Hi-Lo condition by
10 per cent. As a result, sales increased by 3 per cent in the EDLP
condition, while they decreased by 3 per cent in the Hi-Lo condition. There
were, however, also large differences in profitability, but now in the
opposite direction. EDLP reduced profits by 18 per cent and Hi-Lo
increased them by 15 per cent. The authors explained that, in general, the
results depend on how the customer base is divided among price-sensitive
store switchers and store-loyal customers. The main interest of the EDLP
strategy is that it attracts switchers from other stores. If many consumers are
loyal, this will not work well and the profit level will fall because everyone
will pay a lower price. In the Hi-Lo strategy, loyal customers purchase their
product basket both when the products are on deal and when the price is at
the normal level. On average, they therefore generate more profit than in
the EDLP strategy. Store switchers, on the other hand, will buy more when
prices are lower. Overall, the Hi-Lo strategy is therefore a good price
discriminator, while the drawback of EDLP is that it offers low prices to
everyone, irrespective of price sensitivity.
The Combined Effect of Discount, Display and Ad
Features
Research reports by practitioners and consulting agencies claim that the best
promotion campaigns are built on synergies between price deals, feature
advertising and display. For instance, IRI (1989) have circulated an analysis
of these effects based on their 1988 data on sales in 2,400 grocery stores in
66 markets in the USA. The data are normalized on a price cut of 15 per
cent. The main findings are shown in Figure 10.2. The price cut on its own
increases sales by 35 per cent (an elasticity of –2.3). When the discount is
coupled with an ad feature, the effect is 173 per cent, i.e. 138 per cent more
than the price cut on its own. When the price cut is paired with an in-store
display, the sales gain is 279 per cent, 244 per cent more than the sales gain
alone. Of particular interest is what happens when price cut, ad feature and
display are combined. The sales effects could simply add together, thus:

Figure 10.2 The combined effects of discount, display and ad feature

Figure 10.2 shows instead that there is a gain of 128 per cent above the 417
per cent, which suggests that the three components of promotion act
synergistically.

Later academic research, however, reported mixed findings. Gupta (1988)

observed negative interactions of display and feature with price cuts, which
he attributes to a possible overlap or substitutability among these
promotional instruments. Lemon and Nowlis (2002) found, across different
brands, a mix of small positive and negative interactions of display and price
cut and a negative interaction of feature and price cut. In an experimental
setting, East, Eftichiadou and Williamson (2003) found no synergistic effect
for a double, as opposed to a single, display. Zhang (2006) provided an
explanation of these mixed patterns. She observed first that promotion
markers can serve as a proxy for a price cut, as shown by Inman, McAlister
and Hoyer (1990); a promotion signal is in this case taken as a cue for a
price cut that influences choice, even in the absence of a real reduction in
price. Second, another stream of research has shown that in-store displays
and feature ads can influence the formation of consideration sets (e.g.
Allenby and Ginter, 1995). Zhang therefore builds a model of brand choice
that incorporates both processes: consumers use promotion markers either as
price cut proxies, or for consideration set formation, or for both. If display
and feature have mainly a price cut proxy effect, then negative interactions
are observed in choice models. In contrast, if they help get a product into the
consideration set, they then create a positive interaction. Differences across
past studies can therefore be explained by the mechanism that dominates the
choice process.
Carryover Effects
What happens after the sales promotion has finished? This depends partly not
only on the mix of category expansion, brand switching and purchase
acceleration or stockpiling, but also on the extent to which these effects persist
after the promotion. Consider the possibilities (which are illustrated in Figure
10.3):

Some consumers may buy and consume a discounted brand more with no
effect on later consumption.
Some buyers may switch brands or maintain a raised consumption after
trying a brand on promotion.
Some regular consumers may accelerate purchase and stockpile a brand
on a deal; as a result, they may buy less later. This requires more planning
than is usually found among consumers of frequently purchased goods.

The availability of scanner data has made research on the carryover effect more
feasible and the overall picture shows little carryover effect. The most
comprehensive study in this area is by Ehrenberg, Hammond and Goodhardt
(1994a), using panel data from Britain, Germany, America and Japan on 25
established grocery products. The researchers identified 175 sales peaks of 25
per cent or more for different brands in these product fields and compared sales
levels before and after these promotional episodes. The procedure excluded
cases where the sales pattern was irregular either before or after the peak.

Figure 10.3 Post-promotion sales

The overall outcome of this study was a sales increase of 1 per cent which is
effectively no effect. A check was made by measuring the repeat buying rates
for the 8-week period after the peak; the average was 43 per cent – almost the
same as the 44 per cent inferred from NBD theory (see Chapter 4) which
showed that buying was stationary in the post-promotion period. Differences
between countries were small and there was no evidence that categories
showed a consistent movement when data were available on the same category
from more than one country.

In sum, these studies show little evidence of carryover effect. A lack of

carryover effect is quite difficult to explain. How is it that a spike of extra
sales, often several times the normal level, does not disturb the base-level
sales? One explanation offered by Ehrenberg et al. is that deals touch only a
minority of the brand’s customer base; many regular buyers would not see a
promotion and their behaviour could not be affected. A second explanation
offered by Ehrenberg et al. is that the extra purchasers attracted by the deal
were nearly always people who had bought the brand in the past (see Table
10.1); thus, a promotion does not introduce new buyers to the brand. Those
who respond to promotions are past and current buyers who are already
familiar with the brand characteristics.
A thoughtful article by Neslin and Stone (1996) compared seven different
explanations for the lack of post-promotion effect. The authors suggested that
many people are insensitive to the stock of purchases already made so that
household inventory has little effect on purchase decisions. But the enlarged
inventory will eventually slow purchasing and produces a generally lower sales
level rather than a post-promotion dip. So, the effect of discount sales is to
reduce base-level sales; without promotions the base rate would rise.
Long-Term Effects of Promotions
There has been concern that there may be a long-term negative effect of
promotions that is not revealed in the shorter-term studies that test for
carryover effect. Broadbent (1989) and Ogilvy (1987) are among those who
have suggested that the heavy use of sales promotions will degrade brand
equity.

There are four bases for concern about promotional activity. The first comes
from reference price research. Promotional prices may become integrated in
the reference price and consumers therefore start perceiving the normal,
non-promoted price as high (Kalwani and Yim, 1990). In an interesting
twist to this effect, Palmeira and Srivastava (2013) show that when an
additional product (e.g. a ring) is offered as a promotion along with a
purchase (e.g. a necklace), it is better to give it for free than to charge a low
promoted price. This practice makes it less likely that the value of the free
item is considered and it is therefore not devalued in the eye of the
customer. A second basis for concern about promotional activity comes
from attribution theory, suggesting that consumers make causal attributions
of promotional events; they wonder why the brand is promoted.
Lichtenstein and Bearden (1989), for instance, concluded that consumers’
reactions are more positive when they think that a promotion aims to attract
customers, as opposed to when they think the promotion’s objective is to
get rid of unwanted goods. Third, customers may see the lower price as
evidence of poorer quality. Finally, frequent promotions may condition
consumers to only buy on deal. Bolton (1989) and Raju (1992), for
instance, observed that sales spikes are smaller when promotions are more
frequent.

In a review of research up to 1995 on the possible negative effect of

promotions, Blattberg, Briesch and Fox (1995) conclude that, given the
mixed evidence, the ‘jury is still out’. However, more recent studies, based
on time-series analysis, converge on an absence of positive and a possibility
of negative long-term effects. Dekimpe, Hanssens and Silva-Risso (1999)
find positive effects for only one brand out of 13 in an analysis of four
categories. Nijs et al. (2001) examine the possibility of category expansion
but find this type of effect in only 36 out of 560 product categories.
Pauwels, Hanssens and Siddarth (2002) find no permanent effects of
promotions for purchase incidence or purchase quantity. In only one of the
29 cases do they find permanent effects on brand choice.
Couponing
Couponing is massively used by marketers in the USA. For packaged
goods, a total volume of 305 billion coupons were distributed in 2012, for a
value of $470 billion. However, only 2.9 billion of those were redeemed for
total savings of $3.7 billion (statista.com). Research has mainly focused on
whether and, if so, how, coupons create incremental sales. Coupons can
increase category consumption (Ailawadi and Neslin, 1998), encourage
brand switching (Neslin, 1990), trigger trial (Neslin and Clarke, 1987) and
also reward brand loyalty. Neslin (2002: 49) reports several studies that
show that brand-loyal consumers are more likely than others to redeem
coupons.

Coupons can be considered as an instrument for price discrimination.

Instead of giving a discount to all buyers, coupons reserve the discount for
consumers who make the effort to obtain the saving. Coupons have to be
found, stored, organized and redeemed before expiry for the correct
product. The ‘cost’ associated with this effort, including the opportunity
cost of time, is higher for some individuals than for others. Narasimhan
(1984) develops this argument in more detail and his analyses conclude that
users of coupons are more price-sensitive than non-users of coupons.

American consumers are attached to coupons (see Box 10.4). An

explanation for consumers’ attachment to coupons is that highly coupon-
prone consumers are not only drawn to reduced prices but are also seeking
psychological benefits. For instance, they get a sense of achievement by
purchasing products on deal and view themselves as smart shoppers, an
idea Garretson and Burton (2003) confirmed in a store exit survey.
Alexander, Tripp and Zak (2015) took physiological measures during an
online shopping experiment. Participants who received a coupon show an
increase in oxytocin, which is interpreted as an indication that these
coupons are processed in the brain as if they were a gift and endowed with
social content; increases in oxytocin are typically triggered by positive
social interactions.
 Box 10.4 Life without coupons?
In 2012, newly hired JC Penney CEO Ron Johnson changed the promotional strategy by
getting rid of ‘fake prices’ and replacing them with a ‘fair and square’ everyday-low-
pricing scheme. He explained that he thought that ‘people were just tired of coupons and
all this stuff’. JC Penney’s sales did not respond well to the no-coupon revolution and
declined sharply. In April 2013, Johnson was fired from his position. Apparently, JC
Penney’s shoppers liked couponing and missed seeing the ‘how much you saved today’
amount printed at the bottom of their receipts (Tuttle, 2013). Johnson probably did not
know about or remember the experience of Procter & Gamble who launched in 1996 an
18-month ‘no-coupon’ test in upstate New York, a region where 90 per cent of shoppers
used coupons. Procter & Gamble believed that across-the-board lower prices offered
more efficient savings than coupons. In a survey by the trade journal Supermarket
Business, 80 per cent of manufacturers and 69 per cent of retailers and wholesalers
considered that Procter & Gamble was on the right track and some manufacturers
followed its lead (Partch, 1996). However, consumers reacted fiercely with public
hearings, petition drives and boycotts. Some put up ‘Save Our Coupons’ signs on their
front lawns. A county official claimed that the elimination of coupons by the ‘big guys’
was intended to hurt the ‘average Joe’. The Wall Street Journal observed that ‘coupons, to
many people, are practically an inalienable right’ (Narisetti, 1997: 1). Procter & Gamble
stopped the test after 14 months.

Heilman, Nakamoto and Rao (2002) examine responses to in-store instant

coupons such as electronic shelf coupons and peel-off coupons on product
packages. They show that these ‘surprise’ coupons increase market basket
sales for two reasons – unplanned purchases are made as the result of the
psychological income effect in reaction to the unexpected financial gain,
and the coupons raise the consumers’ mood.

In conclusion, coupons – and promotions in general – are not just forms of

price reduction. As Chandon, Wansink and Laurent (2000) show,
promotions give consumers hedonic benefits (opportunities for value
expression, entertainment and exploration), in addition to utilitarian benefits
(savings, the possibility to upgrade to higher product quality and improved
shopping convenience). Lee and Tsai (2014) even find that price
promotions can also influence the consumption experience. When
consumption takes place immediately after payment, discounts make it
more enjoyable. However, when the consumption is delayed, discounts
reduce enjoyment. This may actually be an additional argument against
their usage in some situations.
Customized Promotions
To increase the effectiveness of promotions, marketers have recently
explored the use of customization, both in online and offline (‘bricks-and-
mortar’) stores. In an offline environment, coupons are usually printed at
the checkout register or sent by mail. They can be redeemed on the next
purchase trip. In an online setting, the store interface can be customized for
each visit and promotions therefore can be offered for the current shopping
occasion. Zhang and Wedel (2009) compare the effect of the degree of
customization in these two settings, using a model of purchase incidence,
product choice and quantity. The depth and frequency of discounts can be
customized at a segment level or at the level of the individual customer. The
authors also compare two types of promotions: ‘loyalty promotions’, aimed
at customers who bought the now promoted product at the previous
purchase occasion, and ‘competitive promotions’, for customers who
previously bought a different brand. The data come from a large
supermarket chain that also sells its products through an Internet store.
After the estimation of the model parameters, they determine the profit-
maximizing promotional strategy in 300 different scenarios. The authors
find that loyalty promotions are more profitable in online stores than in
offline stores, while the reverse holds for competitive promotions. This
result is driven by the fact that customers of online stores are in general
more loyal than those of regular stores. Inducing switching is therefore
easier in an offline environment. Surprisingly, individual-level
customization is only slightly more profitable than segment-based
customization or no customization at all. The effects of customization are
particularly small for offline stores. The main problem here is the very low
redemption rates of targeted coupons. In an online environment,
customization has more potential, but only for promotion-sensitive product
categories.
Summary
Consumers compare observed prices to reference prices in order to
determine their interest in buying a particular product at a particular time
and place. Reference prices can be externally available or recalled from
memory. Price memory is, unfortunately, not reliable for many people. One
reason is that prices are not encoded correctly because of their verbal
length.

When comparing different products, consumers do not necessarily prefer

the product with the lowest price because many associate low price with
low quality. In reality, price is often not a good predictor of quality but that
is not how consumers perceive it.

Market researchers apply different methodologies to assess consumers’

price sensitivity. For hypothetical new products, they often use conjoint
analysis and direct questioning. To examine reactions to large changes in
existing prices or to alternative prices for new products, price experiments
are used. When historical data are available, econometric estimation of
price elasticity is possible.

Consumers react to price changes by adapting their purchase behaviour, but

they may also express their agreement or disagreement in ways that affect a
company’s image and, possibly, its brand equity. There is a clear tendency
to interpret price increases as unfair. Consumers often infer that profit is the
supplier’s motive. Interestingly, the way a price (or discount) is presented
can influence the perception of how costly it is.

Promotions can clearly boost sales temporarily, but long-term positive

effects seem to be absent. Even the obvious short-term effects have to be
interpreted with caution because promotions ultimately risk becoming a
costly zero-sum game among competitors. Everything depends on the
source of the additional sales. If promotions increase switching and bring
new customers to the market, they can be profitable. If they just offer the
product at a lower price to brand-loyal customers, they may be counter-
productive. Customizing promotions (e.g. on the Internet or on mobile
phones) could target them more efficiently, but recent research indicates
that customization is only profitable in certain conditions.
Additional Resources
An excellent overview of pricing research is given in Gijsbrechts (1993).
This overview not only shows what the main findings are but also explains
how they were obtained. It is not restricted to pricing and also covers some
promotion issues. Schindler has published a lot of research on odd pricing
and 9-endings. Schindler (2006) is a nice example of how simple
observations of the marketplace can be very insightful. This chapter
referred several times to meta-analyses. A good example of this type of
analysis can be found in DelVecchio, Henard and Freling (2006). Finally,
Raju and Zhang’s (2010) book, Smart Pricing, examines innovative, mostly
Internet-based, pricing strategies and provides excellent analyses of the
most prominent new business models.
11 The Retail Context
Learning Objectives
When you have completed this chapter, you should be able to:

1. Understand the main ideas about gravity models.

2. Discuss the reasons why and when people use different types of retail store.
3. Report on the factors related to loyal, heavy and compulsive shoppers.
4. Know about the applications that can be made of loyalty data.
5. Discuss the evidence on store atmospheric effects and explain those effects.
Overview
First, we consider how store location affects shopper choice according to aggregate
models. Then, at the individual level, we examine the reasons that individual shoppers
give for their choice of supermarket and consider the way in which store use varies over
the day and the week. After this, we focus on three types of customer: loyal, heavy and
compulsive shoppers. We also briefly review loyalty schemes. In conclusion, we assess
the ways in which the store environment can influence spending, particularly the impact
on shoppers of music, smell, colour and crowding in the store.
Section 1: Shopper Choice
Gravity Models
The number of people who buy from a particular store depends upon nearby
population densities, transport access, store type, and the presence of
competing retail locations. Gravity models are used to analyse retail data
and show how shops, stores, shopping centres and cities draw customers
from the surrounding environment.

Early research in this area focused on the attraction exerted on a shopper by

cities at varying distances. Reilly (1929) argued that trade was attracted in
direct proportion to the population of a city and in inverse proportion to the
square of the distance from the city. The similarity to Isaac Newton’s
explanation for the movement of planets led to theories of this sort being
called gravity models. Reilly’s model implies that there will be a breaking
point between two cities where customers are equally attracted to each city.
The model was tested on 30 pairs of cities and found to be quite accurate,
although it does not allow for the strong appeal of particular stores. In
addition, distance is only an approximate indicator of the travel effort and
cost which really control customer demand, so that Reilly’s model is less
applicable when good transport facilities make distant locations easily
accessible. One other problem is that neighbourhoods may be poor or rich
so that population size is only an approximate indicator of the expenditure
per head that is derived from a location.

In a further development, Huff (1962, 1981) modelled the attraction of a

retail centre within a city. In this model, the attraction of a centre, A1, is a
function of its selling area (S), divided by a power λ (lambda) of the travel
time (T):

A1 = S/Tλ

The probability of a consumer using a shopping area is a function of A1

divided by the sum of the attractions of all the available shopping centres,
i.e.
 p = A1/∑ (A1... An)

Wee and Pearce (1985) found that Huff’s model was well supported and
that the exponent, λ, was approximately 2, as Reilly hypothesized. Huff’s
approach includes all the possible shopping centres, although in practice a
consumer might rule many of these out. Wee and Pearce modified Huff’s
model and used only the shopping centres that the shopper considered. With
this adjustment predictions were better, but even the improved model gave
an R2 of only 0.25, showing that much of location preference remained
unexplained. This is hardly surprising since a gravity model cannot take
account of the detail of a particular shopping environment or of the exact
way in which information is processed. For example, Foxall and Hackett
(1992) found that stores that were located at path junctions were better
remembered. One use of gravity models is to calculate the loss of trade that
would occur in existing stores if a new store were opened in a particular
location.

Because shoppers are attracted to retail centres, retail growth also happens
at these centres. Central place theory (Christaller, 1933; Losch, 1939) uses
the importance of a centre and the economic distance as basic concepts to
explain how centres develop and how retail units tend to cluster together,
with each taking advantage of the custom generated by the others.
Businesses in the same field can benefit from proximity since together they
increase the total custom. We can find examples of this effect in most cities;
one example is the concentration of restaurants in particular locations (see
Box 11.1).
Box 11.1 Retail clusters
Some quite small locations can become specialists in a particular field. In the ancient
Shropshire town of Ludlow, a number of restaurants gained entry to the Michelin Guide
and more high quality restaurants then joined them. People can take a long weekend in
Ludlow and eat excellent cuisine. As a result, Ludlow may draw gourmets from great
distances. A similar effect is found in Ireland, at Kinsale, which is a hub for award-
winning restaurants.
Store Preferences
Gravity models and central place theory provide explanations of retail
attractiveness at an aggregate level. At an individual level, we can ask about
the shoppers’ preferences for store types and assemble the reasons given for
using stores. We can also investigate when the stores are used, and how
frequently. Complete Exercise 11.1 on your personal use of supermarkets.
Exercise 11.1 Supermarket use
Which supermarket do you use most? What is the main reason for this? Do you have a regular
day and time of day for doing your shopping? Many people do have regular times – why do
you think this is so? Would you say that you were loyal to your main supermarket? What does
this mean?

Try to answer these questions now.

Reasons for Using Supermarkets
In the 1990s, surveys of supermarket use were conducted in the UK at
Kingston University in concert with colleagues in the USA and New Zealand.
These surveys sought the reasons why respondents used their main
supermarket (the one where they spent most). Table 11.1 shows the principal
reasons across different years and continents. One message from Table 11.1 is
that the reasons for shopping at supermarkets did not vary that much between
different advanced economies. The two most important criteria were location
and good value; these account for over half of the main reasons given for
patronizing a specific supermarket. When the figures are compared for 1992
and 1994, we find that, in the USA, good value dominates in 1992, probably
because of a recession at that time. Recession hit the UK rather later and we
see an increase in the importance of good value there in 1994. Then, as the
economy became more settled from 1994 to 1998, the importance of good
value eased slightly. This evidence shows that supermarket choice criteria can
change in response to economic conditions. Retailers are alert to such changes
and adjust their offerings and communications accordingly. Thus, with the
onset of the global economic crisis in 2008, there has been an increased
emphasis on price. This can be seen in Figure 11.1 in data provided by Nielsen
for the UK on the two most important reasons for store choice.

Sources: Consumer Research Unit, Kingston University, UK; Debra Perkins, Florida Memorial
University, USA; Phil Gendall, Otago University, New Zealand
We see that, in the economic conditions of 2015, getting good value is
important for the main shopping trip since ‘value for money’, together with
‘low prices’ and ‘in-store promotion’, account for 87 choices out of 200.
Convenient location gets only 24 out of 200 for the main store, while range and
quality criteria combined get 51 out of 200. We should be wary about a direct
comparison between Table 11.1 and Figure 11.1 because of the different
methods used for gathering data, but if we compare top-line findings it appears
that in 2015, compared with earlier data, low prices were considerably more
important, location much the same and quality/range somewhat less. In the UK,
a rise in the deep-discounting stores, Aldi and Lidl, has been one consequence
of this focus on price. In the UK, the big four – Tesco, Asda, Sainsbury’s and
Morrisons – accounted for 75 per cent of all grocery sales in 2010, according to
the Kantar Worldpanel reported in The Guardian (2011). By August 2015, the
share of the big four had fallen to 72 per cent, while the deep discounters had
gained substantially and accounted for 11.5 percent. Over this period, margins
were squeezed as supermarkets fought to retain share. Nielsen also reported on
convenience or ‘top-up’ stores; compared with the main shopping store choice,
location was much more important, and value for money, range and quality less
so.

Figure 11.1 Which factors are most important when choosing where to do your
shopping? (Two reasons selected, data supplied by Nielsen, 2015)
Shopping Trip Patterns
The previous section outlined the reasons shoppers give for their choice of
store. However, it is important to understand not just why and where people
shop but also how often and when. While consumers go shopping for a great
variety of goods, researchers have mostly studied grocery shopping (partly
because of the wealth of detailed data available from consumer panels).
Households tend to have a routine of supermarket shopping that often includes
one weekly main trip and one or more top-up trips. In the USA, McKay (1973),
Frisbie (1980) and Kahn and Schmittlein (1989) found this pattern. In Britain,
Dunn, Reader and Wrigley (1983) also found evidence of weekly trips
supplemented by secondary trips. In recent times, the growth of convenience
stores and small-store formats among the supermarket groups has led to more
secondary trips.

The timing of shopping trips has many managerial implications, ranging from
staffing and stock management to store layout, parking requirements and the
best time to conduct in-store promotions. The store may be little used over
much of the day while, at other times, congestion and delay may reduce the
quality of service delivered. Off-peak periods are useful for staff relaxation,
cleaning, restocking, training and maintenance, as Sasser (1976) notes, but
smoothing demand remains a desirable managerial objective.

Store use varies over the year and can be affected by bad weather. It rises
before holiday periods, including the weekend, and shopping tends to be
heavier later in the week. Figure 11.2 shows the distribution of shoppers over
the week that applied in the UK in 1994 for full-time workers and others.
Figure 11.3 shows the distribution of those shoppers over the day. The figures
show that those who are employed full-time tend to shop later in the week and
later in the day. We can infer from these two distributions that those shopping
late on a Friday or on Saturday are very likely to be employed full-time. This
separation of the employment segments could be useful to store owners; if
these groups prefer different goods, both displays and in-store promotions
could be switched to take account of those preferences. Recent evidence on
daily shopping suggests that Saturday and Sunday supermarket shopping has
become somewhat heavier but that the basic pattern of shopping over the week
is unchanged (sales data are shown for 2015 in Table 11.2, last column).
Figure 11.2 Percentage of shoppers on each day by employment status (East et
al., 1994b)

Figure 11.3 Percentage of shoppers at different times of the day by

employment status (East et al., 1994b)
As noted, there has been a growth in small-store formats as the big supermarket
groups turned their attention to this part of grocery shopping, so that there has
been a growth in sales via these outlets. At the same time, the larger
supermarkets have become less used and some of these have been closed
because they were unprofitable. This turns attention to the many visits of light
shoppers which, in aggregate, account for a substantial part of grocery
shopping. Dunnhumby, which analyses supermarket data for Tesco and other
stores, has supplied us with data on the store visits of light shoppers (Table
11.2). The visit frequencies are very similar in all regions and show little
difference by day of the week, except on Sunday in western Europe which is
half that for other times.
We also obtained data on the use of supermarkets by light shoppers over the
day from Dunnhumby. This is shown in Table 11.3 for the different regions.
Eastern Europe starts shopping earlier and Asia shops later, but otherwise there
is not much variation between regions. Notice that the data are aggregated for
different periods in this table. When allowance is made for this, we see that
shopping visits show two peaks – one in the morning and one in the afternoon.
Thus, current supermarket use by day looks similar to the two-peak pattern
shown in Figure 11.3 for 1994.
The Flexibility of Shopping Times
Store efficiency would be increased if demand was smoother over the day and
over the week. East et al. (1994) investigated flexibility in the times of use of
stores and whether shoppers deliberately avoided busy times. This was tested
in supermarkets in two ways. The first method compared the times of shopping
of the 48 per cent who disliked being held up at the checkout with the 52 per
cent who minded this less. There was no significant difference in the times
when these groups used supermarkets, which suggests that those who disliked
delay took no particular action to avoid congestion. The second method
compared those who claimed to avoid busy times with the rest of the shoppers.
The ‘avoiders’ used quiet times only slightly more; the researchers judged that,
at most, 6 per cent of all shoppers changed times to avoid congestion. This
shows little flexibility in shopping times and East et al. (1994), seeking an
explanation for this, found that over 60 per cent of consumers had routine days
and times for their main grocery shopping, which suggests that habits limit
flexibility. The reasons given for these routines are shown in Table 11.4.

This evidence shows that many shopping trips are related to factors over which
shoppers have little or no control. They cannot choose when they are paid or
when they have to collect their children from school. Because their shopping
times are often linked to such activities, they are fairly inflexible; shoppers
could shop at other times, and they reported this in the survey, but they have
good reasons for their practices and stores would have to offer substantial
inducements at off-peak times to change their customers’ shopping trip
behaviour. It is also clear that the reasons shoppers have for their habits are
relatively unchanging so that any inducements that a store might offer would
have to be sustained. Overall, these findings suggest limited scope for
redistributing demand in supermarkets.
Exercise 11.2 Analysis of store data
On the website associated with this book (www.sagepub.co.uk/east) you will find uk98, which
is the data in an SPSS format from a survey conducted in the UK. Findings from this survey
were used for this chapter. The questionnaire for the survey is supplied as su6. Use SPSS to
compare stores (question 3) in respect of comparative checkout delay, how much the store is
recommended and the relative share of requirements (questions 13, 16 and 21). Use Compare
means, means, to get the means per store.

For all stores, what factors are associated with the rating of the store? To do this, select
potential predictors and run an ordinal regression to predict the answers to question 17. Is
rating related to recommendation and share of requirement? Use a Spearman correlation to test
for this.

How strong are these findings? What alternative explanations are there for the associations you
find?
Section 2: Customer Typologies
Perhaps the most sophisticated classification of shoppers is made by the
Dunnhumby agency in their analysis of Tesco’s loyalty data. Dunnhumby
uses a selection of frequently bought products to classify customers on 30
criteria. For example, a large size of oven-ready chips will indicate
positively for budget consciousness and negatively for health consciousness
or gourmet propensity. Those who purchase organic produce indicate
another type of concern and those who buy Tesco’s own label
disproportionately may be classified as Tesco loyalists. These behaviour-
based classifications are used to ensure that coupons go to those who will
appreciate them, and new products are promoted to customers who are
likely to buy them. In an unpublished 2014 Dunnhumby survey by Julian
Highley, health-committed customers emerge as an attractive target. The
proportion of these people is increasing and they welcome supermarket
efforts to provide healthy food. They tend to be loyal and up-market and are
more responsive to strong promotions than the less health committed. As a
result, serving these customers can yield more profit. In other work,
Dunnhumby has derived the price sensitivity and profitability of customers
from their past purchases. This customer segmentation can be used to direct
promotions to those who will yield more profit and it could even be used to
change displays and promotions over the day as the proportions of different
customers change.

We now consider three other types of store customer: store-loyal, heavy,

and those whose shopping behaviour seems out of control – compulsive
customers. The first two are important for profit; the last indicates a social
problem that is of concern.
Store-Loyal Customers
Like brand loyalty (Chapter 2), store loyalty can be defined in a number of
ways. The attitude to, or satisfaction with, the store provides one measure of
loyalty. As behaviour, loyalty can be measured by share-of-category
requirement (SCR), which is the proportion of spending in a specific store
or group. Interest tends to focus on the SCR relating to the primary store,
called first-store loyalty. Another behavioural measure is retention of the
store in the shopper’s store portfolio; high retention loyalty is when
customers patronize the store for a long time. Despite the assumption that
different loyalty measures have some common basis (discussed in Chapter
2), these measures may show little association. East et al. (2005a) found
that the SCR for the primary store and attitude to the store showed only
weak correlations of 0.13 and 0.15 in UK and New Zealand studies.

Store loyalty can be investigated using the market regularities described in

Chapter 4. We may use models like the Dirichlet to show the relationship at
an aggregate level between SCR requirement and factors such as
penetration, purchase frequency and market share (Kau and Ehrenberg,
1984; Uncles and Ehrenberg, 1990; Uncles and Hammond, 1995; Wright,
Sharp and Sharp, 1998). We may also search for economic, demographic,
behavioural or attitudinal factors that predict SCR or retention at an
individual level. Such variables include age, the frequency of store use, the
accessibility of the store, generalized brand loyalty, shopping on a particular
day, the amount spent in the retail category, income, attitude to the store and
free time. However, to some extent, these treatments of loyalty have
become less interesting to retailers; large stores, which cover a variety of
merchandise, are interested in getting customers to shift types of spending.
Supermarkets want customers to buy toiletries from them rather than from
pharmacies, and televisions from them rather than from electronics
specialists. In this context, they talk of the share of the custom’s wallet.
Box 11.2 Average supermarket loyalty
Average levels of loyalty to different stores depend on the retail category, market
concentration, average store size and the period of time taken for measurement. For
supermarkets in the UK, AGB (1992) reported that over an 8-week period approximately
75 per cent of expenditure took place in the primary supermarket. Also in the UK, Mason
(1991) reported that the SCR for the primary store was 72 per cent over one month, 65
per cent over 12 weeks and 60 per cent over 24 weeks. Loyalties to the store group are
higher than the loyalty to the individual store. In keeping with the double jeopardy rule
(Chapter 4), bigger stores and larger store groups tend to have higher SCRs. However,
with shoppers increasingly using the deep discounters, Aldi and Lidl, and spreading their
spending across different formats, share of category requirement for store groups is now
in decline.
Box 11.3 Purchase and loyalty on the
Internet
In Chapter 1, we drew attention to the rapid growth of Internet purchase. In the early days
of Internet use, it was argued that this channel was particularly suited to ‘bit-based’
products such as music, travel, share purchase, banking, gambling and pornography.
These fields have certainly prospered but the Internet is now frequently used to purchase
a range of physical products. Some of these are familiar, such as groceries, books, CDs
and clothing, while others benefit from the ability to search the Web for unusual products.
For example, if you want to find a specialized product that might not be sold in your
neighbourhood, such as an orthopaedic back support or a home fire-escape ladder, these
are easily located online and delivered to your door. Thus, some use of the Internet is
driven by the convenience of this method of purchase. The ability to deliver relevant
information is a powerful incentive in other cases, such as when purchasing shares, and
this could raise loyalty. From the standpoint of the supplier, the Internet may save costs,
particularly for bit-based products that can be distributed ‘down the wire’. The sale of
music has been revolutionized by this method. The customer can also save because price
comparison is easy on the Internet.

Nielsen (2011a) found that 30 per cent agreed strongly and 37 per cent slightly that you
could save money on groceries by buying on the Internet. However, there are drawbacks
and the Nielsen study showed that the inability to select items for oneself when buying
was the most common problem reported about Internet grocery purchase. Internet grocery
shopping is well established, particularly in the UK, but a MINTEL (2014) report found
that, although about 20 per cent of shoppers did all or most of their grocery shopping
online, this constituted only about 5 per cent of total grocery sales by value. Thus, it
seems that online grocery purchase attracts lighter buyers.

Grocery buying has been studied to see whether brand loyalty is stronger when purchase
is made on the Internet. Degeratu, Rangaswamy and Wu (2001) found that brand names
play a stronger role in choice and that fewer brand switches are made in the online
environment. Another study by Danaher, Wilson and Davis (2003) found this effect for
high market share products but the reverse for small share brands. But these studies
compare those who shop online with those who shop offline and any differences may be
due to the different abilities and interests of the two groups rather than to the channel
used. Arce and Cebollada (2006) compared the online and offline grocery purchasing of
the same people and still found less brand switching online. One explanation for this is
that, on the Web, buyers may work from a list of past purchases and take less advantage
of the discount opportunities available. As a result, they stick to the same brands and have
high loyalty. This effect may be weaker for small share brands that may be bought less
often, which would explain the finding of Danaher et al. (2003).
Heavy Shoppers
According to the ‘heavy half’ principle (Chapter 4), the heaviest 50 per cent
of buyers will make about 80 per cent of all purchases. Retailers want to
identify these heavy buyers and aim to target promotions at these shoppers
in order to recruit and retain them.

It might be thought that those who spend more would use more shops and
therefore have lower first-store loyalty, but Dunn and Wrigley (1984) and
Mason (1991) found that first-store loyalty did not change as total
expenditure in supermarkets increased. Knox and Denison (2000) found a
small positive association (r = 0.24) between store loyalty and total
spending for supermarkets but a negative relationship for other types of
retail outlet. If bigger spenders in supermarkets are more loyal, it follows
that they are doubly attractive to the retailer who gets their main custom.
Not only do they spend more but also a larger proportion of this spending
goes to their favourite store. Stores with loyalty data may be able to identify
heavy buyers directly from their spending and direct promotions to them.
(Note that loyalty data only show purchases at the single store chain but
spending in other stores may be inferred from the goods bought in the
single chain.)

Unpublished data gathered at Kingston University on heavy supermarket

shoppers suggest that, more than other segments, heavy shoppers tend to
have larger incomes and households, be aged under 45, prefer large out-of-
town stores, shop later in the day, less often, and have a regular day for
shopping.
Compulsive Shoppers
Compulsive shoppers buy clothes, shoes and other goods which they do not
need and sometimes never use (Arthur, 1992). Koran et al. (2006) surveyed
shoppers in the USA and found that nearly 6 per cent reported compulsive
shopping behaviour. This behaviour was slightly more common among
women, younger persons and the less well off. Scherhorn, Reisch and Raab
(1990) have reported on compulsive shopping in Germany, and Elliott
(1993, 1994) has researched this problem in the UK. Faber and O’Guinn
(1988) saw compulsive shopping as part of a wider range of compulsive
behaviour, which they describe as: ‘A response to an uncontrollable drive or
desire to obtain, use, or experience a feeling, substance, or activity that
leads an individual to repetitively engage in behaviour that will ultimately
cause harm to the individual and/or others.’

Compulsive shopping is clearly a serious problem, which often causes

financial and psychological distress to the shoppers themselves as well as
their families. One reason for compulsive shopping could be the more
attractive shopping environment of the present day, but to explain why
some people suffer from this compulsion more than others we need to focus
on the behaviour and background of the compulsive shopper. D’Astous
(1990) argues that this type of behaviour is the extreme end of a continuum
and that many people have strong urges to buy that they can barely hold in
check. Compulsive purchasing has been reviewed by Black (2007) who
reports that this pattern this associated with other compulsive behaviour.

Compulsive shoppers tend to be owners of credit cards (d’Astous, 1990)

and they also tend to be younger and to have lower self-esteem (Faber and
O’Guinn, 1992). However, research in this area is by survey and it is
difficult to assign cause and effect; compulsive shoppers may have low self-
esteem because of their behaviour but it seems more likely that this is a
cause, rather than an effect.

Compulsive shoppers appear to get some emotional release, or temporary

‘mood repair’, out of the process of buying. Elliott, Eccles and Gournay
(1996) interviewed 50 compulsive shoppers and probed their thinking and
motivation. The respondents accepted that their behaviour was abnormal
and potentially very damaging but felt they could not easily control it. In
this respect, it functioned rather like a drug and Elliott et al. emphasize this
by describing the behaviour as addictive. They also found that this type of
shopping was often related to unsatisfactory relationships with partners.
Women whose partners worked excessively, ignored them or were
controlling, were more likely to be compulsive shoppers. In some cases,
their behaviour was a form of revenge or was deliberately designed to rile
their partner. In other cases, their treatment by that partner may have
lowered that self-esteem and made compulsive shopping more likely.

Dittmar, Beattie and Friese (1995) suggested that the way in which products
are bought reflects the way in which people see themselves. These
researchers suggested that the sexes might differ in this respect. They found
that men tend to buy instrumental and leisure items impulsively, reflecting
their independence and activity, while women tend to buy symbolic and
self-expressive goods concerned with appearance and the emotional aspects
of self. Dittmar (2005) stresses that materialistic values are one key to
compulsive shopping. Those with low self-esteem can only get mood repair
from shopping if possessions have importance to them (see Figure 11.4).

Figure 11.4 Compulsive shopping as a product of high materialism and low

self-esteem
Theories of Store Loyalty
There may be no overarching explanations for giving most of one’s
shopping spend (high SCR) to one store, or for remaining loyal to a specific
store for many years (high retention). These effects may be the result of a
large number of weak influences that have little in common. However, it is
reasonable to look for one or two mobilizing influences that are responsible
for much of the effect and here there have been two competing theories
about the nature of store loyalty. We report these theories and suggest a
third possibility.

Resource Constraint
The first theory of store loyalty was suggested by Charlton (1973), who
drew on earlier work by Enis and Paul (1970). In this theory, store loyalty is
the outcome of limited resources: those who lack money, time and
transport, or whose environment lacks choice (Tate, 1961), are forced to use
one store much of the time and are therefore obliged to be loyal. Carman
(1970) offered a variation of this model, suggesting that some people had
little interest in shopping and therefore did not use the choice that they had.
Such people had commitments outside the home, full-time work, little home
entertaining and a lack of interest in deals, advertising and shopping. As a
result, they were loyal to both brands and stores because they did not seek
alternatives. However, East et al. (1995) found that shoppers with different
loyalty levels gave similar ratings for the pleasantness of supermarket
shopping, which does not support Carman’s lifestyle theory.

Carman (1970), and Enis and Paul (1970), found that those with low
incomes were more loyal, so resource constraint probably did affect loyalty
at the time when the research was conducted but it is unlikely that this still
applies. Shoppers now have more choice because they have wider access to
stores through car ownership or by shopping online and they can stock
perishable groceries for long periods because of the almost universal
availability of home refrigerators and freezers.
Discretionary Loyalty
Dunn and Wrigley (1984) argued that Charlton’s negative concept of store
loyalty needed review. They suggested that some store loyalty arose from a
pattern of one-stop shopping, often in large supermarkets. We call this
discretionary store loyalty. Discretionary loyalty could be an adaptation to
being time poor and money rich. People can spend less if they use several
stores and cherry pick the bargains (leading to low loyalty) but this takes
time and effort. Those with less time and more money may choose to buy
most of what they need from one outlet because this is easier; as a result,
they will show high loyalty. Mason (1991) found that store loyalty was
higher when the housewife worked and was under 45 years of age (when
family commitments are likely to take a lot of time). Flavián, Martínez and
Polo (2001) found general support for discretionary loyalty in Spain.
However, East et al. (2000) found that neither free time nor income was
significantly related to SCR in a regression analysis covering many factors,
and this leaves some doubt about the basis of discretionary loyalty.

Loyalty to Stores as Habit

An alternative way of thinking about loyalty is that it is a habit. The habit
may be set by the conditions under which a person lives; for example, those
with routine work patterns are free at specific times and may fit shopping at
a particular store into one of these times; after a while, this arrangement
will become routine. Another basis for habit is personal disposition; some
people may find routines more attractive than other people. Either way, we
would expect to find that those with high store loyalty tend to have other
similar habits, such as brand loyalty and a regular day for shopping.
Evidence from East et al. (2000) gave support to this explanation.
The Effects of Loyalty Schemes
Loyalty schemes take a variety of forms. All schemes have some form of
customer incentive. Many schemes, such as supermarket loyalty programmes,
give cash or product incentives. Some schemes raise service levels for a more
valuable part of the customer base, for example the provision of executive
lounges for frequent air travellers. A loyalty scheme has two potential effects.
First, the incentives may raise customer acquisition, share of wallet and
retention by directly influencing customers. Second, when information on the
purchases of different card holders is collected, the store owner may use this
information to target shopper segments more accurately and to produce further
gains in customer acquisition, spending and retention.

Incentives are very expensive. They take about 1 per cent of turnover; thus, if
the store group makes 5 per cent on turnover, the incentives reduce the margin
by 20 per cent and sales must be increased to compensate for this. Furthermore,
competitors can run loyalty schemes which may cancel out any gains. In an
effort to reduce the effect of competitor response, some schemes, such as Air
Miles and Fly Buys, only allow one retailer in each sector. If Shell gives Air
Miles, BP, Esso and other fuel suppliers are excluded. However, competitors
can still run alternative schemes.

A loyalty scheme provides a database covering most customers together with

information about their purchases. Retailers can use this information to ensure
that promotions (often by vouchers) are well targeted. Tesco, with its analysis
agency Dunnhumby, is recognized as a world leader in this field. Tesco is
interested in the types of purchase made – and not made – by a card holder. If
the customer buys no meat, vouchers for meat are never sent, to avoid
offending vegetarians, but if the customer does not buy toothpaste at Tesco,
vouchers might be used in an effort to switch their purchasing of toiletries to
Tesco. A similar inducement might be given to those who appear to buy most
of their wine elsewhere; alternatively, the wine buff may be invited to wine-
tasting events. In other cases, vouchers may be for what the customer already
buys, to act as a reward. The database also reveals customers who have
switched to competitors (they can differentiate this from going on holiday by
the nature of the purchases before the customer leaves). Defecting customers
can be sent vouchers in order to get them back. Customer purchase information
also helps to sell items that never go through the stores, such as gardening
equipment and baby buggies (certain other items purchased, together with a
customer’s age, can indicate when she may be pregnant). The customer
database allows Tesco to launch financial products, such as credit cards and
insurance, with less risk than its competitors. The pattern of demand in stores
allows the management to tailor the inventories for different stores. Tesco can
also sell data to manufacturers because the system shows the relative
performance of brands.

It is probably the information rather than the direct incentive that is the most
important aspect of loyalty schemes but researchers remain interested in the
incentive effect of loyalty cards; this is difficult to measure because other
factors may be involved. For example, Tesco took share from Sainsbury when
the loyalty scheme was introduced in 1995 but much of this change might have
happened in any case because Tesco built more stores, many of which were in
Sainsbury territory (East and Hogg, 1997). Although Tesco gained market
share, the firm’s customers did not show a disproportionate increase in loyalty
measured in aggregate as share-of-category requirement (SCR). The data in
Table 11.5 from Taylor Nelson Sofres show that in 1996, a year after the
introduction of the loyalty scheme, Tesco had a slightly greater market share
than Sainsbury in 1994, but the same SCR. Thus, Tesco’s loyalty scheme
seems to have built a customer base by gaining new customers rather than by
getting its existing customers to buy more. This seems to be the normal pattern
for share gains that was noted in Chapter 4.

Meyer-Waarden (2007) has reviewed studies of incentive effect. In general,

there is little or no effect on SCR from loyalty schemes; for example, Sharp
and Sharp (1997) found that the firms in the Fly Buys scheme in Australia did
not show a significant increase in SCR compared with Dirichlet norms. Meyer-
Waarden found that customers who were holders of one loyalty card showed
single-figure percentage increases in lifetime duration (retention) and more
share of wallet, but customers with several cards did not show these effects.
 Source: Data from Taylor Nelson Sofres

In these studies, loyalty membership may involve self-selection; if those who

join a loyalty scheme are already highly loyal customers, it is difficult to
attribute effects to the loyalty scheme. Leenheer et al. (2007) took account of
self-selection effects in a study of loyalty schemes in Dutch supermarkets.
They found that, if self-selection was ignored, loyalty schemes increased share
of wallet by 30 percentage points, but when self-selection effects were taken
into account, the increase in share of wallet attributable to loyalty programme
membership averaged 4 per cent. The size of the effect varied slightly between
schemes and reduced as customers joined additional loyalty programmes. A 4
per cent gain in customer loyalty is probably not going to cover the cost of the
scheme, but it is a mistake to focus only on share of wallet since customer
acquisition and retention may also give a return. Another study on the long-
term effect of loyalty programmes by Liu (2007) found that these did not affect
heavy buyers but that low- and moderate-weight buyers did increase their
spending. An influential early paper on loyalty schemes was quite sceptical
about their benefits (Uncles, Dowling and Hammond, 2003). This review tends
to support this verdict with regard to their incentive effect, but it appears that
loyalty schemes may be justified when the wider advantages derived from
loyalty data are included.
Section 3: The Store Environment
Store Layout, Location and Space
Actions occur when the environment presents opportunities, stimuli and
rewards. The store layout should therefore be designed to increase spending
opportunities, present purchase cues and make shopping in the store easy
and pleasant. These different considerations do not always coincide; for
example, IKEA uses a layout that requires the customer to move through
the whole store in order to reach the exit; this may raise spending
opportunities but it can be a near-claustrophobic experience for some
customers.

In supermarkets, specific locations in the store are associated with different

rates of purchase, probably because they provide a more powerful stimulus
to shoppers. For example, eye-level shelves sell nearly twice as much as the
lowest shelf. One of the applications of scanner technology, called direct
product profitability (DPP), is to measure the profit from a given stock-
keeping unit (SKU) in a specific location. This technology identifies ‘hot-
spots’ in the store and SKUs can be moved to optimize profit. However,
sales gains for items given more space or better locations are offset by the
lower sales of items that lose space or go to a worse location. Drèze, Hoch
and Purk (1994) showed a potential profit difference of about 15 per cent
when comparing the worst and best configurations of location and space,
but in practice the authors estimated that the feasible changes would
produce a much smaller gain.

The display of goods is important in supermarkets as a large proportion of

grocery purchase decisions are made at the point of purchase (Dagnoli,
1987). Early experimental studies on varying shelf space in grocery
categories show that a doubling of space leads to sales increases in the
region of 20 per cent (Cox, 1970; Curhan, 1972; Kotzan and Evanson,
1969; Krueckeberg, 1969). However, Drèze et al. (1994) found that more
effect came from location than from the amount of space given to an SKU.
Atmospherics

Store Features
The store environment includes the amount of space employed, and the
layout, fittings, colours, aromas, sound and density of customers present.
There are some standard display features used by stores that affect the
impression given by the store. For example, discount stores may sell out of
cases to emphasize low prices, shopping centres create central areas with
entertaining features, and fashion shops use music that suits the age and
taste of their clientele. These features help to define the store offering and to
differentiate it from that of other stores (and from sellers on the Internet). In
this way, store design and the display of goods have parallels with
advertising and packaging.
 Exercise 11.3 Assessing atmospheric features
Describe a shop which you find attractive and stimulating. What is the basis for this? How are
space, colour, sound and odour used? Do these environmental features affect your spending in
the store?

Kotler (1973) suggested that store features create an atmosphere, modifying

the buyer’s knowledge and mood, thus affecting behaviour. Kotler also noted
that atmospherics could be used to de-market; for example, State liquor
stores in some countries are deliberately off-putting environments. There is
no doubt that environmental features can have a strong effect on in-store
behaviour. Labrecque, Patrick and Milne (2013) suggest that colour carries
meaning and can influence consumers’ thoughts, feelings and behaviours.
Bellizzi, Crowley and Hasty (1983) found that people were more aroused by
red than blue or green and suggested that red will speed up behaviour and
might be appropriate where quicker action is more profitable, such as in fast-
food restaurants. Babin, Hardesty and Suter (2003) reported that blue is
associated with greater purchase intention, compared to orange, and found
that this effect was much weaker under subdued lighting. Milliman (1982)
found that fast-tempo music speeded up supermarket customers and this
reduced their purchasing. When customers moved more slowly with slow-
tempo music, they spent 38 per cent more. Areni and Kim (1993) showed
that, compared with popular music, classical background music in a wine
store led to the choice of more expensive wines. North, Hargreaves and
McKendrick (1999) followed up Areni and Kim’s study by using either
French accordion music or German beer cellar music in a supermarket. More
French wine was sold when the French music played, and more German
wine was sold when the German music played, and customers were unaware
of this influence on their behaviour.

Another study by Yalch and Spangenberg (2000) showed that music could
raise sales in a department store. The music had to be appropriate for each
department; for example, music in departments with younger customers had
to be played at high volume. Other work has looked at the way different
stimuli work together. For example, Mattila and Wirtz (2001) showed that,
when music and scent were similar in terms of arousing properties
(congruent), the evaluation of the environment increased. In another
example, a ‘Christmas’ scent raised the evaluation of the store environment
only when accompanied by Christmas music (Spangenberg, Grohmann and
Sprott, 2005). Bosmans (2007) also found that scents that were congruent
with the product had a powerful influence on evaluations. Morrison et al.
(2011) show that the arousal induced by music and aroma results in
increased pleasure levels, which in turn positively influences shopper
behaviours, including time and money spend, approach behaviour, and
satisfaction with the shopping experience. Madzharov, Block and Morrin
(2015) found that ambient scents affect consumers’ spatial perceptions and
feelings of power, along with product preference and enhanced purchasing
behaviour. These effects probably work by association between the stimulus
and the action.

Another explanation for atmospheric effects is that they affect mood which
in turn affects purchasing. Donovan and Rossiter (1982) used a classification
of mood states described by Mehrabian and Russell (1974). They found that
a store’s atmosphere produced mood effects in consumers which could affect
the time and money spent in the store. Figure 11.5 shows this stimulus →
organism → response (SOR) model with the intervening moods of pleasure
and arousal. Donovan and Rossiter expected high arousal to act with
pleasure to raise spending/time in store and with displeasure to reduce
spending/time in store. The first was supported but there were too few
unpleasant environments to test the second effect.

Figure 11.5 The role of moods in mediating atmospheric effects on shopping

behaviour

Smith and Sherman (1992) showed that store image was associated with
mood, which then predicted the amount of time and money spent in the
store. Mehrabian and Russell (1974) suggested that more novel and complex
environments would raise interest, and Gröppel (1993) observed that
supermarkets with high novelty and complexity levels gave more pleasure
and that customers spent more time and bought more in such stores.
Swinyard (1993) argued that only the more elaborate processing of the
highly involved consumer would be affected by mood and that this mood
change would affect shopping intentions. This was supported in a scenario-
based experiment, i.e. only the highly involved shoppers claimed that they
would modify their shopping intentions. Sharma and Stafford (2000) showed
that store design affects the persuasiveness of sales personnel, an effect that
may be mediated by mood. Eroglu, Machleit and Davis (2003) even found
support for mood effects in an online environment.

Donovan et al. (1994) reviewed research in this area. They noted weaknesses
in their earlier article and those of others. In particular, they suggested that it
was necessary to distinguish moods induced by the environment from
emotions associated with purchase. Donovan et al. conducted a further study
which avoided these problems. They found that pleasure did contribute to
time in the store and extra spending, and that arousal did reduce spending in
environments rated as unpleasant, but arousal did not increase spending
when the environment was pleasant. The increase in time without increase in
spending seems to indicate more browsing or window shopping. In some
retail sectors, window shoppers are as common as active shoppers (Nielsen,
2005) and we need to know how valuable window shoppers are: do they buy
later or encourage others to buy through word of mouth?

While mood may act as an intervening variable to modify levels of spending,

there are cases where this explanation is inadequate. In particular, the North
et al. (1999) study indicates that behaviour can be modified directly by the
French and German music cues (there are no French and German moods).
Also, the effect of the tempo of music, observed by Milliman (1982), may be
automatic and not mediated by mood. This fits a stimulus–response (SR)
model rather than an SOR model. Many stimuli affect behaviour without
much awareness; this is part of our low-involvement response to the
environment which helps us to cope with the wide variety of stimuli that
impinge on our senses. By acting in this way, we leave ourselves free to
concentrate on other features of the environment. One SR explanation comes
from North et al. (2004) who looked at congruent elements in advertising;
they suggest that one stimulus will activate or prime related processes so that
a response occurs more easily when a second congruent stimulus is
encountered. This priming process can be seen as readying an individual for
certain types of stimulus but not others, so that thought, feeling and action
are channelled in particular directions. Such an account would explain the
wine store results of Areni and Kim (1993), and North et al. (1999), as well
as the joint effect of congruent stimuli.

The work on atmospherics indicates powerful effects but our explanation of

these effects is limited. Morin, Dubé and Chebat (2007) suggest a dual
model whereby stimuli can affect shoppers’ perceptions by two paths.
Ambient environmental cues, such as smells, operate at an unconscious level
– as pre-attentional signals – and induce limited processing, but focal cues
with more value may be selected for conscious attention. The authors
suggest that some stimuli, such as pleasant music, may influence behaviour
through both channels. One route to influence, which involves conscious
thought, is social. Some retail interiors may cause so much interest that these
places are easily mentioned in conversations and lead more people to visit
the store – for example, the vastness of the Mall of America, and the
aquarium and ski slope in the Dubai Mall. One other explanation for
customer response to atmospheric effects comes from attribution theory. In
this theory, an individual seeks to interpret and control his environment, and
particularly to understand why and how things happen. The theory suggests
some biases in reasoning of which the most important is fundamental
attribution error. This is the tendency to find reasons for behaviour in the
motivational dispositions of people rather than in their environment.
Attribution effects could guide inferences about the purpose of retail
environments, and the behaviour of store staff and other customers. Brown
and Dant (2009) note a number of accounts of retail behaviour that fit this
theory.

Crowding
A common problem in stores, banks, post offices, restaurants and other retail
services is the level of congestion or crowding. People have a complex
response to crowding and, in different contexts, may find it attractive or
aversive. Hui and Bateson (1991) found that an important factor in
determining whether crowding was liked or disliked is the control that
customers feel that they have in the situation. In Hui and Bateson’s study,
high densities of people were associated with increased control in a bar and
reduced control in a bank. People go to banks for instrumental reasons and
bars for recreation, and it seems likely that these different uses affect the
way crowding is perceived, since crowding is more likely to obstruct activity
in banks than in bars. In shops, therefore, people are likely to dislike
densities that impede action, and store designs should aim to reduce such
congestion. However, when shoppers have little to buy and their activity is
more recreational, they may enjoy store congestion. It is also likely that
people are put off when a store appears empty. Wicker (1984) has suggested
that every setting has an optimal number of occupants. For example, some
people feel reluctant to go into a near-empty restaurant. Here, people may
attribute the emptiness of the restaurant to a lack of quality.

Milgram (1970) saw crowding as stressful, making people quicker, less

exploratory and more inclined to omit purchases. The impact of stress is to
narrow concentration so that central tasks may be performed better, but more
complex operations, which require more peripheral perceptions and
memories for their completion, may be performed less efficiently. This
means that the key functions of shopping may be done more efficiently
under crowded conditions but shoppers may forget items that are peripheral
to their needs. Anglin, Stuenkel and Lepisto (1994) found that shoppers who
scored high on measures of stress engaged in more comparison shopping and
were more price-sensitive; these behaviours might be seen as central to
shopping. This is another reason why stress-reducing atmospheric factors
may raise spending. Michon, Chebat and Turley (2005) found that scents
made people more positive about their environment when the shopping
density was at a medium level. This may be because scents tend to reduce
stress. Atmospherics, in general, were found to reduce stress according to
Baker and Wakefield (2012). In psychology, it has been found that there is a
preferred level of stimulation. People seek minimum arousal and this occurs
at an intermediate level of stimulation; they are more aroused when there are
no stimuli (and they are bored) and also when there are many stimuli (and
they cannot process them all). This model is shown in Figure 11.6. It
suggests that store interiors, including the effects of crowding, can be too
stimulating or not stimulating enough and that the optimum will vary across
consumers and types of outlet.

Figure 11.6 There is an optimum level of stimulation

Summary
At the aggregate level, retail use is explained by gravity models, and at the
individual level, by identifying the preferences of customers. Shoppers are
mainly affected by convenience, price, range of choice and quality. In times
of recession, price assumes more importance. Retail use varies by day, and
time of day, and stores are under-used some of the time. Many of the
reasons for use of stores at different times relate to the shopper’s personal
situation and this limits the scope for changing time of use. Fully employed
people shop later in the day and later in the week compared with those who
are not in full-time employment.

A number of typologies have been suggested for shoppers. One important

group for retailers is high spenders. Among grocery shoppers, these heavy
buyers are found to be wealthier, to have larger households and to use a car.
A second group is high-loyalty customers. Behavioural store loyalty, like
brand loyalty, has two basic behavioural forms: share-of-category-
requirement loyalty is the proportion of patronage given to a store in the
retail category, and retention is the duration of patronage. Early studies
suggested that people were more loyal (measured as a proportion) when
they lacked the time, money, transport resources or interest to spread their
custom, but this seems to no longer apply. A second theory, called
discretionary loyalty, was that certain people, who had resources, used them
to concentrate shopping. The latter are time poor and money rich and can
afford to save time by shopping at one store even though this may cost a
little more. Evidence in the UK fitted a third theory, that store loyalty was a
habit which was associated with other habits such as brand loyalty and a
regular time for shopping.

Compulsive shoppers buy things that they do not need and may not use.
This pattern of behaviour has raised concern because of the financial
devastation that uncontrolled spending can produce. These people tend to
have low self-esteem, which is temporarily alleviated by shopping.

In all types of store, spending may be affected by atmospherics, which are

produced by variations in space use, colour, sound, odour and crowding.
Atmospheric changes produce quite strong effects on spending but the
explanations for these effects need more development. In particular, further
studies are needed to understand the interactions between different
atmospheric elements and the impact of multisensory atmospherics on
consumer behaviour (for a review of evidence, see Spence et al., 2014).
Additional Resources
There is a large literature in this area, as well as a major specialist journal,
the Journal of Retailing. For particular topics, see Donovan et al. (1994),
for store atmosphere, Morin, Dubé and Chebat (2007), for the effect of
music, Meyer-Waarden (2007), on loyalty programmes, and Arce and
Cebollada (2006) for work on online versus offline shopping behaviour. For
articles on store atmospherics, refer to the Journal of Business Research.
12 Word-of-Mouth Influence
Learning Objectives
When you have completed this chapter, you should be able to:

1. Discuss the difficulty of conducting research on word of mouth (WOM).

2. Describe how product decisions in different categories are affected by WOM.
3. Report on the relative occurrence and impact of positive and negative WOM in familiar
categories.
4. Describe factors associated with WOM that affect its impact.
5. Report how WOM relates to the current and past usage of brands and to market share.
6. Describe ways in which WOM may build up support for a product in social networks.
7. Suggest how marketers might apply knowledge about WOM.
Overview
Research on WOM has probably been stimulated by online activity since comment on the
Internet (or eWOM) may be similar to face-to-face WOM. Internet comment has the great
advantage of being recorded and researchers have taken advantage of this to study its
effects.

In this chapter, we show how consumers use WOM to choose brands in different
categories. We describe problems in researching this field, report findings on the relative
frequency of positive and negative WOM (PWOM, NWOM) and examine the
circumstances that stimulate people to give WOM. We provide evidence on the impact of
PWOM and NWOM and we explain how different factors contribute to that impact. We
describe how WOM production is related to market share, and review applications of
WOM research.
Section 1: The Nature of Word of Mouth
In Chapter 5, we described the way in which innovations are adopted by
consumers. At the centre of the adoption process is the communication of
information. A person who adopts a new idea or product must find out
about it, either through mass media (advertising, promotions, television,
radio), through personal discovery (e.g. seeing it in a shop) or from other
people (salespersons, other consumers). This chapter is concerned with the
last way of finding out, through the influence of others. Sometimes, this
occurs as observational learning when people see what others do and copy
them, but often they receive advice as word of mouth (WOM). In addition
to guiding the adoption of new products, WOM is involved in the switching
from one brand to another in established markets. Normally, WOM is
focused on conversations between consumers, where, unlike advertising,
the person giving advice usually lacks any commercial interest. This feature
of WOM is one reason why it is influential but another aspect of person-to-
person exchanges is that these are often interactive so that a receiver can
follow up and ask further questions. This helps the receiver to get the
information needed, which makes the advice more influential. WOM advice
may be received via face-to-face exchanges, telephone, text messages, mail,
email, blogs, message boards and social networking websites. WOM also
sometimes occurs as rumour (see Box 12.1).
 Box 12.1 Rumour
Rumours are unverified topical beliefs that circulate between people. Early thinking about
rumour was presented by Knapp (1944) and by Allport and Postman (1947). Rumours
may be based on hope, fear or hatred and may involve claims about conspiracies or
dangers. The Internet now provides a means for the rapid circulation of rumours and
many companies have suffered from this hazard. The financial marketplace is particularly
susceptible to rumour (Kimmel, 2004). Rosnow (2001) argues that uncertainty, credibility
and personal relevance are the primary drivers of rumours, which will spread faster in
contexts of high anxiety (e.g. when investments are at risk). Kimmel and Audrain-
Pontevia (2007) found that roughly three-fifths of rumours were negative, one-fifth
positive and one-fifth neutral. They confirmed that uncertainty, credibility and importance
of the topic were the key factors in rumour transmission. For a review of research on
rumours and their influence, see Kimmel (2010).

WOM may be positive (PWOM, recommendation, advocacy) or negative

(nwom, criticizing, advising against). Some exchanges contain both positive
and negative comment and some are neutral. When it is about a brand,
PWOM usually increases, and NWOM reduces, the receiver’s probability
of purchasing that brand. Some advice occurs in a commercial context, for
example, from sales personnel and on sponsored websites. Commercial
advice is different from C2C advice because it is potentially biased.
However, Carl (2008) studied the responses of consumers to advice from
BzzAgents (people who are given products and asked to talk about these
products to others); he found that three-quarters of respondents trusted the
BzzAgent to give them good advice when they knew that he/she was a
BzzAgent. In fact, the effect of the BzzAgent’s advice was often greater
when their affiliation was known. This indicates that those with a
commercial interest may be quite influential despite their role.

Many of the classic studies on WOM were concerned with innovations and
new categories rather than established brands – for example, Whyte (1954)
on air conditioners, Coleman, Katz and Menzel (1957) on the prescribing of
new drugs by physicians, and Katz (1961) on new farming practices. These
really new products may produce a large amount of comment compared
with well-established products that consumers often know well. However,
sometimes brands will have new features not offered by others and then
choice may be more like the adoption of a new product. For example, a car
manufacturer may offer ‘passive keyless entry’, allowing the car to be
started while the transponder key remains in the driver’s pocket. As the
category matures and brands become more familiar, the reason for choice
may not be an innovation but some simple advantage that can be drawn to a
consumer’s attention. For example, one person might advise another about
the relative cost of mobile phone brands, or their performance in weak-
signal areas. This is useful to a prospective buyer but it does not deal with
an innovative feature of the product. In this chapter, we focus less on the
adoption of new categories and more on brands in mature categories.
Researching Word of Mouth
WOM is difficult to measure. Ideally, we would observe WOM as it occurs
and then monitor its consequences. In practice, WOM occurs too rarely for
it to be observed systematically and usually any consequences occur much
later, so that direct observation of the outcomes of WOM may be
impossible. As a result, other methods have to be used, which are reviewed
below.

Text Mining on the Internet

Although we cannot observe WOM as it happens, we may be able to
measure it as comments posted on the Internet. Such eWOM is not hard to
find in consumer-generated media, but there are problems about relating
this to face-to-face WOM. Those who set up websites may encourage either
more PWOM or more NWOM than is typical in everyday life, and those
who post comments on the Internet, and those who read these comments,
may be different from those who give and receive offline advice. So far,
studies of online comment have produced mixed results. Godes and
Mayzlin (2004b) did not find that the volume of online comment about TV
programmes was predictive of viewing, but Liu (2006) was successful in
predicting box office returns from the volume of online comment about
movies, and Qin (2012) also found that the volume of WOM predicted
movie sales. Interestingly, Liu did not find that the valence of the WOM
(i.e. whether it was positive or negative) was predictive of sales. However, a
later article by Liu (2012), dealing with comments on Twitter, suggests that
the valence is more predictive than the volume of comments. More work is
required here.

Internet research usually deals with aggregate effects. We can count the
posts and obtain data on box office receipts. We may be able to predict
returns from such data but we do not know the specific influences on
individuals and how those individuals reacted, instead we have only the
collective effects. We want to understand as well as predict, and for this we
need individual-level data so that we can connect individual responses to
individual experience. Such individual-level data are obtained in
experiments and surveys.
 Box 12.2 Online advice
The predictive value of online comment depends, in part, on how much this medium is
used, compared to other media. If it is only a small part of total advice on brands, it may
not be a reliable guide to sales. Surveys show that online comment remains a modest part
of the total. In 2006, Keller and Fay found that WOM was:

Face-to-face 70%
Phone 19%
Email 4%
Text message 3%
Online chat or blog 1%
Other 3%

In other surveys, the shares are similar. In 2010, the Keller Fay Group reported that 7 per
cent of WOM was offline in the USA, the UK and Australia, though this rose to 15 per
cent for the teen group in US measurements. In 2011, Keller Fay reported that, in the UK,
81 per cent of WOM was face-to-face, 10 per cent via phone and 9 per cent online
(including email, texting and social networking sites). This suggests that the Internet is
not the dominant channel for advice, though some categories such as restaurants, holidays
and hotels attract much more online comment than others, and it is likely that, in these
fields, the Internet is a more reliable guide to demand. In addition, it is possible that the
design of survey questions leads to the omission of some Internet advice. Fay (2014)
reports a market research study that found a third of referral sales were based on Internet
comment.

There are, however, some differences between Web and face-to-face advice. First, most
online advice is one-way and not interactive. Second, in many contexts, such as online
reviews or Twitter, advice from one source may be received by many others, which is
uncommon for offline advice. Third, the Web may allow a degree of deception – those
reviews on Amazon may include some that are ‘arranged’; because of this, people may be
more suspicious of positive comment on the Web than they are when it occurs face-to-
face. Fourth, offline WOM is more often between close ties whereas, on the Web, a larger
amount of weak-tie contact is likely to occur (e.g. in discussion groups or anonymous
product reviews).

Experiments
A number of experiments have examined the impact of positive and
negative information (e.g. Ahluwalia, 2002; Herr, Kardes and Kim, 1991).
The main problem here is that the artificiality of the laboratory situation
restricts generalization to naturally occurring WOM. This artificiality has
several aspects:
 1. The stimulus is not like real WOM. In experiments, the ‘WOM’ is
often prepared written information rather than spontaneous exchanges
between people (e.g. Herr, Kardes and Kim, 1991). Such prepared
advice cannot be asked for, which is often a feature of real WOM, and
the advice is unlikely to be well-tailored to the needs of the receiver.
2. The response measures may be inappropriate. Experimental studies of
WOM have used attitude towards a product or brand and belief items
to measure impact (e.g. Ahluwalia, 2002); marketers are generally
more interested in the impact on purchase or purchase probability.
Experiments often rush the process by taking measures of effect
shortly after exposure to the stimulus. In natural settings, people who
receive WOM may not act on it for months. In their improved
experimental design, Christiansen and Tax (2000) delayed the
measurement of effects for a week.
3. In experiments, each participant makes an equal contribution to the
outcome. In everyday life, some people say nothing while others give a
lot of WOM.

There may be more flexibility with role-play (simulation) experiments but

these raise a further problem. In a role-play experiment, the subject may be
asked what he or she would do in a specified situation, for example: ‘If
someone asked you about mobile phones, would you recommend/advise
against…?’ There is no guarantee that participants in such role plays would
do as they claim. However, Christiansen and Tax succeeded in devising
quite realistic WOM experiments, using pairs of participants, with one
being required to give advice (in their own terms) to the other about a real
product. Another possibility is to use field experiments. Godes and Mayzlin
(2004a) used a field experiment to compare the extra effect on sales of
WOM from loyal and non-loyal customers. Unfortunately, field
experiments are very resource-intensive and the experimenter may still have
to use a survey method to find out what advice has been given and received.
These problems weaken the value of experiments but they remain popular;
East (2016) has suggested that heuristic mechanisms make the controlled
experimental design more attractive than it deserves.

Retrospective Surveys
In retrospective surveys, respondents have to report on their experience and
these reports may be systematically distorted by recall error. For example, if
NWOM is more easily recalled than PWOM, a measure of relative
frequency will be biased in favour of NWOM (see Box 12.3). A second
concern about surveys relates to the recruitment of the sample, which is
often based on convenience. However, problems about convenience
sampling recede as more data are gathered. If we have 20 studies using
diverse population samples and different categories, and these all show the
same pattern, we can be more confident about the findings. A further
problem arises when survey evidence is interpreted; unlike experiments, it
does not provide causal relationships so that associations between variables
may be explained in a variety of ways.
Exercise 12.1 Questionnaire on word of
mouth
Fill out this questionnaire:

1. Do you own a mobile phone?

No [1]
Yes [2]

2. Which make of mobile phone do you have?

Please write in (Samsung, iPhone, etc.) ………………

3. In the last six months, how many times have you received positive advice about any
mobile phone handset?
Write in number (0, 1, 2, etc. ……)
If you answered 0, then please go to Q.9

4. The last time you received positive advice, did you ask for advice or was it just given?
Just given [1]
Asked for it [2]

5. What was your relationship to the person who last gave you positive advice?
Casual acquaintance [1]
More distant family, friend or colleague [2]
Close family, close friend or colleague [3]

6. About which brand was the last positive advice received?

Please write in (Samsung, iPhone, etc.) ………………

7. Did the last positive advice that you received affect your handset choice or intended
handset choice?
No [1]
Yes [2]

8. How strongly expressed was the last negativeadvice?

Hardly at all strongly [1]
Moderately strongly [2]
Fairly strongly [3]
Very strongly [4]

9. In the last six months, how many times have you received negative advice about any
mobile phone handset?
Write in number (0, 1, 2, etc. ………..)
If you answered 0, then please go to Q.15

10. The last time you received negative advice, did you ask for advice or was it just given?
Just given [1]
 Asked for it [2]

11. What was your relationship to the person who last gave you negative advice?
Casual acquaintance [1]
More distant family, friend or colleague [2]
Close family, close friend or colleague [3]

12. About which brand was the last negative advice received?
Please write in (Samsung, iPhone, etc.) ………………

13. Did the last negative advice received affect your handset choice or intended handset
choice?
No [1]
Yes [2]

14. How strongly expressed was the last negative advice?

Hardly at all strongly [1]
Moderately strongly [2]
Fairly strongly [3]
Very strongly [4]

15. In the last six months, how many times have you given negative advice about any
mobile phone handset?
Write in number (0, 1, 2, etc. ……)
If you answered 0, then please go to Q.17

16. About which brand did you last give positive advice?
Please write in (Samsung, iPhone, etc.) ………………

17. In the last six months, how many times have you given positive advice about any mobile
phone handset?
Write in number (0, 1, 2, etc. ……)

18. About which brand did you last give negative advice?
Please write in (Samsung, iPhone, etc.) ………………

The purpose of this exercise is to show how aspects of WOM may be measured in a
retrospective survey. From the responses of a group of people to these questions, it is possible
to find out:

how much PWOM is received compared with NWOM

how much PWOM is given compared with NWOM
whether those who give more PWOM also give more NWOM
whether people mostly give PWOM about their current brand
whether advice on their current brand affects their behaviour more than advice on other
brands
how much WOM comes from close ties
whether most WOM is requested or not
what proportions of the sample claim to have had their choice affected by the PWOM
and NWOM received
how different factors relate to impact; for example, are people more influenced by
WOM when they have requested it and does WOM have more impact when it is about
 the currently owned brand?

When no method is satisfactory, researchers may give up and investigate

something else. This has probably led to a lack of research on WOM in the
past. But this is something that we can ill afford. In many categories, WOM
appears to be the most powerful influence on consumption and, outside the
commercial arena, WOM is involved in many social changes. In these
circumstances, even weak findings should be put into the public domain.

To some extent, the problems that affect measurement in this area may be
offset by using multiple methods and measures, and a wide range of
categories. A second strategy is to estimate measurement distortions so that
errors can be corrected (Box 12.3).
Box 12.3 Measuring recall bias
East et al. (2013) measured how the volume of WOM recalled is related to the time
elapsed since the WOM was given. Respondents were asked about the volume of PWOM
and NWOM that they gave in the week after using a service such as a hotel. They were
also asked how long ago the service was used. Data on six categories were collected. It
was anticipated that people would forget more instances of WOM when the time lapse
was longer. However, there was a tendency for the recalled volume of both PWOM and
NWOM to increase with the time elapsed. When the ratio of PWOM to NWOM was
measured in relation to time lapse, there was no significant trend. This finding suggests
that ratios gathered over different periods are comparable.
Section 2: The Occurrence of Word of Mouth
How Does Word of Mouth Affect Brand Choice in
Different Categories?
WOM from others provides helpful information and this is particularly true in
the case of services that cannot be tested before a choice is made. For example,
a person who has to find a new dentist has few sources of relevant information
on a dentist’s competence. Because of this, advice from other people is
probably the best way of finding a dentist that they will be happy with. The
adviser provides a kind of second-hand experience. WOM will be less
important in the case of goods that can be inspected and tested, and when
information can be gathered from advertising or online search before buying.
This means that the need for WOM will vary between categories. It is often
said that WOM reduces risk – and this is true – but risk is highest when there is
a lack of information on a product and little opportunity to gather information
by direct experience.

In early work, WOM was credited with very large effects. Dichter (1966)
claimed that advice figured in as many as 80 per cent of brand decisions. Katz
and Lazarsfeld (1955) claimed to show that WOM was seven times as effective
as newspapers and magazines, four times as effective as personal selling and
twice as effective as radio advertising in influencing consumers to buy
products. These early studies applied more to the adoption of new categories
than to brand switching, so these claims may not tell us much about brand
choice in familiar categories. However, WOM clearly has an impact on brand
choice; Keaveney (1995) found that about 50 per cent of service provider
replacements occurred primarily through WOM. A study by East et al. (2005b)
showed that sales impact depended on the category. Table 12.1 shows the
results of this work.

The main source of information was classified into recommendation, personal

search, advertising and ‘other’. The ‘other’ category included non-commercial
editorial advice in the mass media and situations in which people had no choice
because of contracts, gifts or other circumstances that were compelling. At the
base of the table, we see that recommendation was the main influence in about
one-third of the brand choices. In this research, each respondent was asked
about two or three categories, and the data are grouped accordingly in Table
12.1. Within each grouping, you can see that categories differ in the way they
get their customers. For example, in the first grouping, coffee shops and mobile
airtime providers are more often chosen on recommendation than credit cards.
The evidence from Table 12.1 shows that WOM is less often a source of
information for cars (13 per cent) and for retail services such as supermarkets
(10 per cent and 9 per cent). This is not surprising since durables can be tried
out and supermarkets are well known and both are the objects of substantial
advertising.

Marketers are also interested in the value of customers acquired by different

methods. It is thought that customers recruited by WOM will understand and
appreciate the product better than those acquired by advertising and may spend
more, recommend more and be better retained, making them more valuable.
Trusov, Bucklin and Pauwels (2009) and Villanueva, Yoo and Hanssens (2008)
found that customers derived from WOM were more valuable than those found
by conventional marketing activity. An analysis by Uncles et al. (2013) found
that referral customers did recommend more and were better retained but they
did not spend more; their findings suggested that it was the referral customer’s
longer retention that made the main contribution to their value.
Two out of three customers come to us by word of mouth
How Does WOM Occur?
There is a widespread belief in marketing that PWOM comes from satisfied
and NWOM from dissatisfied customers (see Box 12.4), but think back to
the last advice that you gave. Was it driven by satisfaction or dissatisfaction,
or were you trying to provide information that would help someone else with
their decision? Our satisfaction or dissatisfaction with a product may be the
main basis for giving advice, but often we will be influenced by other
factors.

The fact that advice may be unrelated to satisfaction is indicated by a study

conducted by Anderson (1998), who used the Swedish Customer
Satisfaction Barometer and the American Customer Satisfaction Index,
which covered many industries in each country. The results were very
similar for the two countries and Figure 12.1 illustrates the data for Sweden.
We see that there is a little more WOM when people are very satisfied or
very dissatisfied, but that when they are neutral about an issue WOM is still
produced at about 80 per cent of the maximum level. This indicates that
satisfaction and dissatisfaction need not be involved in the production of
WOM and implies that other circumstances are relevant.

Figure 12.1 Frequency of WOM in relation to satisfaction and

dissatisfaction in Sweden (adapted from Anderson, 1998)
 Box 12.4 Comparing satisfied and
dissatisfied consumers
A study of a frequently purchased product by Goodman and Newman (2003) found that
NWOM from dissatisfied customers occurs about twice as frequently as PWOM from
satisfied customers. Anderson’s (1998) comprehensive study also showed greater WOM from
those who were very dissatisfied compared with those who were very satisfied but the effect
was small. Anderson commented: ‘The widespread belief in a high degree of word of mouth
by dissatisfied customers may be unwarranted. In fact, in a sizable proportion of cases, the
difference between the two is probably not significant’ (Anderson, 1998: 15).

People may confuse the WOM from satisfied and dissatisfied customers with WOM in
general. This may be why Silverman (2001: 134) claims that studies have shown that most
WOM is negative. To establish the ratio of all PWOM to all NWOM, we need studies on the
general occurrence of PWOM and NWOM, not just those where the WOM is based on
dis/satisfaction. Mangold, Miller and Brockway (1999) established a typology of the triggers
of word of mouth about services as judged by the receiver of WOM. East et al. (2015) used
this typology in surveys to find out what most stimulates PWOM and NWOM. The results are
shown in Table 12.2, where we can see that satisfaction (PWOM) and dissatisfaction
(NWOM) are the most important factors but that a variety of other stimuli induce WOM.
These results are for services and a study has now been done on durable goods (at the time of
writing, not yet published). The findings for durables are quite different. Advertising has the
strongest effect on PWOM and little effect on NWOM. Satisfaction is an important stimulus
of PWOM but dissatisfaction had little effect on NWOM, perhaps because modern durables
are so reliable that they rarely cause dissatisfaction.

Some products generate more WOM than others. Berger and Schwartz (2011)
point out that some categories are more interesting than others – one would
expect phone apps to get more comment than soap. Also, some products are
more visible than others because of their usage or because they are frequently
present in the environment. Such products (e.g. mobile phones) may stimulate
more WOM than those that drop from sight after use, such as films. Berger and
Schwartz found that the cued and visible products received more WOM than
interesting products with limited visibility. The latter were talked about a lot
immediately after consumption but WOM then fell away. In a field experiment,
the authors found that WOM could be increased by linking a product to a
recurrent feature of the environment. The authors also suggest that advertising
could be more effective when this sort of linkage is used. A study of the decay
in the production of WOM after product experience found wide variation
between products (East et al., 2014b).

Motives for Giving WOM

Rather different from the triggers of WOM studied by Mangold et al. (1999)
are the motives for giving advice. There are circumstances in which people
deliberately go out of their way to give advice, most notably when they write
product assessments and give advice online, and it is worth understanding why
they choose to do this. In an early study before online advice existed, Dichter
(1966) reported on the motivation to talk about products. He analysed
interviews and found that people gave advice because this gave them standing
with the receiver. The recommendations which people gave were based on
experience, involvement with the product, the needs of the receiver and public
information (such as ads) on the product. Also using interviews, Sundaram,
Mitra and Webster (1998) found that PWOM was motivated by altruism,
product involvement, self-enhancement and assisting the company producing
the product. The motives for giving NWOM were altruism, anxiety reduction,
vengeance and as a response to others seeking advice. Hennig-Thurau et al.
(2004) turned attention to online advice. Using an online questionnaire, they
found that advice was motivated, as in offline, by wanting to vent negative
feelings, concern for other consumers, social benefits, economic incentives,
helping the company and advice seeking, but there were two other factors
related to the online environment – one was a form of self-enhancement
coming from expressing positive feelings and a second was the assistance
provided by the platform to express advice. Related to the second point, Berger
and Iyengar (2012) found that more interesting topics are covered online
because the medium gives more time to consider topics and write reviews.
However, the relative permanence of online comment may affect the
willingness to express it. A study by Eisingerich et al. (2014) found that people
were more reluctant to express comments on a social media site than face-to-
face because they felt more social risk. A further question on motivation is
whether giving advice has any effect on future advice giving. In a scenario
experiment, Chawdhary and Dall’Olmo Riley (2015) found that giving PWOM
on a service provider enhances the merits of the provider in the mind of the
sender, who then claims to be more likely to give PWOM in the future.

Although motives are important, we should also take account of situational

determinants of WOM, as Mangold et al. (1999) do in their typology.
Is there More PWOM than NWOM?
Naylor and Kleiser (2000) studied users of a health and fitness resort and
found more positive comment than negative. Chevalier and Mayzlin (2003)
found that the majority of book reviews on two websites were positive.
Godes and Mayzlin (2004b) studied TV comment on websites and found
that positive appraisals occurred nearly twice as often as negative
appraisals. Romaniuk (2007) found four times as much PWOM as NWOM
when assessing advice about television programmes. The Keller Fay Group
conducts surveys of ‘branded’ conversations that have taken place over the
last 24 hours in the USA and the Group has provided us with data for 2009:
65 per cent of these conversations were mainly positive, 8 per cent mainly
negative, 15 per cent mixed and 12 per cent neutral. If we assume that
people hearing mixed comment on a brand would state that they have
received both PWOM and NWOM, these data indicate that there is more
PWOM than NWOM with a ratio of about 3.5 to 1. East, Hammond and
Wright (2007) examined the ratio of PWOM to NWOM in 15 different
studies, covering all the brands in a range of widely used categories (mostly
services). In every case, the PWOM incidence exceeded the NWOM
incidence and the average ratio was 3.1 to 1. This work was conducted by
asking respondents about the PWOM and NWOM that they had given in the
last six months. In follow-up studies, respondents were asked about the
WOM they had received in the last six months and the WOM they would
give, if asked. These follow-up studies gave PWOM:NWOM ratios of 2.4 to
1 and 3.4 to 1 respectively. From this evidence, it is clear that there is more
PWOM than NWOM, though the ratio varies by category.

Why is there more PWOM? One explanation may be that there are not
many negative things to say about goods and services. Mostly, people are
satisfied with what they get, according to Peterson and Wilson (1992),
whose work suggested an average 10:1 ratio of satisfied to dissatisfied
(Chapter 9). A second possibility is that PWOM is seen as more useful.
Most consumer choices are about selecting one from many brands. NWOM
may eliminate an option but this does not settle the choice if more than one
option remains. By contrast, PWOM may be used by a receiver to make a
final decision. Thus, if people are trying to help others with their advice,
saying what is good may be more constructive than saying what is bad.
How Much Do People Talk about Their Current
Brand?
East, Romaniuk and Lomax (2011) investigated whether the brand that was
referred to in PWOM and NWOM was currently used, previously used or
never used (Table 12.3). Across 15 studies, they found that, on average, 71
per cent of PWOM was about the currently used brand, 22 per cent about a
previously used brand and only 7 per cent about a never-used brand. For
NWOM, 22 per cent was about the currently used brand, 55 per cent about
a previously used brand and 22 per cent about a never-used brand.
Wangenheim (2005) also found that NWOM was often about previously
owned brands, and Winchester and Romaniuk (2008) found that, when
people expressed negative beliefs about brands, these brands were often
previously owned. Table 12.3 also shows that people are more willing to
give NWOM than PWOM on brands they have not used and this suggests
that, sometimes, brands become widely discussed because of their
deficiencies. This is a serious worry for managers. Note that the previous
work showing more PWOM than NWOM was about all the brands in a
category. Individual brands could be the object of more NWOM than
PWOM.
How Does the Occurrence of WOM Relate to the
Market Share of the Brand?
Bigger brands, with more users, will get more recommendations because, as we
see above, most recommendations are about the current main brand. As a
result, the volume of recommendation will tend to relate to market share.
NWOM on previously owned brands will reflect the market share that applied
at an earlier time, and if the market has not changed much this will
approximate to the current market share. This means that NWOM volume will
also relate to market share, but less so than PWOM because it relates to an
earlier market structure. This was tested by Uncles, East and Lomax (2010a).
They analysed data from 13 surveys and found an average correlation between
market share and PWOM volume of 0.92. This was significantly greater than
the corresponding correlation for NWOM which was 0.73.
This evidence shows that if one brand gets more PWOM than another, it is not
necessarily performing better. To do well, a brand must get more PWOM and
less NWOM than would be expected on the basis of the market-share norm.
Sometimes, new brands may get much more PWOM than their market share
warrants; for example, some unpublished evidence at Kingston University
showed that, when smart phones first arrived, iPhone was well above the norm
for PWOM and its subsequent success has vindicated this early interest.
Factors Associated with Word-of-Mouth
Production
In studies at Kingston University, we have found that the volume of
recommendation is often related to:

The relative attitude to the brand. Relative attitude is the rating of the
brand compared with other available brands and is similar to relative
satisfaction.
Whether a person was recruited to the brand by recommendation or
not. In the main, those who are recruited by recommendation tend to
give more recommendations themselves. This was found by
Wangenheim and Bayón (2004) when they investigated German utility
customers and Uncles et al. (2013). This effect is likely to depend on
the size of a person’s circle of friends. Those who interact more with
others have more opportunity both to receive and to give advice.
Related to this, Godes and Mayzlin (2004a) incentivized PWOM and
found that the extra sales that resulted were related to the size of a
person’s social circle.
Whether the communicator recommends other categories. People who
give advice across a wide range of products are called mavens (Feick
and Price, 1987).
Age. The pattern here is that people tend to give and receive less
WOM as they age, particularly when they are over 65 (East et al.,
2014a). This may depend on opportunity, since there is likely to be a
loss of social contact as people age, stop work and their children leave
home. (Factors associated with age are covered in more detail in
Chapter 6.)
Whether a brand owner has heard others recommend their brand. (We
discuss this in Section 4.)

Other factors relating to WOM production depend on the categories:

Customer tenure (duration of time as a customer of the brand). The

relationship between tenure and PWOM was described in Chapter 2. In
 brief, East et al. (2005a) found that recommendations fell as tenure
increased in the case of credit cards, bank accounts, motor insurance
and supermarkets, but recommendation rose in the case of car
servicing and fashion shops. In other categories, there was no
significant effect.
Weight of purchase. Heavy buyers quite often give more WOM but not
always. Perhaps, in some categories, they habituate to the brand and
then become less interested in talking about it.

Interestingly, share-of-category requirement is not usually related to

PWOM. High-share customers, by their nature, have a more restricted
experience of brands other than their main one and this may limit their
ability to give advice when this involves comparison of brands.
Section 3: The Impact of Word of Mouth
What is the Impact of Positive and Negative Word
of Mouth on Brand Choice?
NWOM may be less common than PWOM, but perhaps it has more impact
when it does occur? There seems to be a belief among marketers that an
instance of NWOM has more effect than an instance of PWOM, and there is
some evidence suggesting that this might be true. Arndt (1967) showed
twice as much impact on purchase from NWOM than from PWOM, but he
studied only one brand. Also, a series of studies has shown a ‘negativity
bias’ – that negative information has more impact on attitudes than positive
information (Anderson, 1965; Chevalier and Mayzlin, 2003; Fiske, 1980;
Herr, Kardes and Kim, 1991; Mittal, Ross and Baldasare, 1998; Mizerski,
1982). Negative information is less common than positive information so
attitudes tend to be positive, reflecting the generally positive thinking. This
means that negative information usually differs from the prevailing attitude
more than positive information (Fiske, 1980). It is this gap between the
position supported by the message and the position currently held by the
receiver that is the basis for negativity bias. The larger gap when the
message is negative gives more room for a change in attitude. For example,
evidence that a brand is unreliable might have more effect than evidence
that the brand is reliable because most people assume that modern products
are reliable. Exceptionally, when the receiver’s expectation is negative and
the information received is positive, there could be a ‘positivity bias’.
Research on negativity bias is reviewed in detail by Skowronski and
Carlston (1989). However, some work has not supported negativity bias.
For example, Ahluwalia (2002) compared responses to written positive and
negative information when participants were familiar or unfamiliar with the
brand. When the brand was familiar, there were no significant differences in
the impact of positive and negative information.

This work on negativity bias has used measures of impact based on change
in attitude or thinking. However, in marketing, impact may be better
measured as a change in purchase or purchase propensity. People might
receive NWOM and change their attitude but not change their intention to
purchase. This could happen if, prior to the NWOM, they had zero
probability of purchase. East, Hammond and Lomax (2008) used the shift in
purchase probability to measure the impact of WOM; they showed that
positive advice has more effect if the receiver has a low likelihood of
purchase before the PWOM is received because this leaves more room for
change. Conversely, NWOM has more effect when the initial probability of
purchase is high. This applies the gap notion to intention rather than
attitude, as in the negativity bias research.

In Table 12.4, we show East et al.’s average results from 19 studies. In this
work, respondents were asked what their probability of purchase was before
and after hearing WOM, using the Juster (1966) scale (see Chapter 7). The
mean probability of purchase before WOM was 0.43 for those who received
PWOM and 0.40 for those receiving NWOM, so that there was somewhat
more ‘room for change’ in the purchasing probability for the receivers of
PWOM (0.57) than NWOM (–0.4). The impact of PWOM was
correspondingly greater in magnitude than that of NWOM (0.20 versus –
0.11). A study by Sweeney, Soutar and Mazzarol (2014) has also found that
PWOM has more effect on purchase intention than NWOM. These findings
suggest that PWOM usually has more impact than NWOM when impact is
measured as a change in intention. However, as we stated at the beginning
of this chapter, it is difficult to study WOM effects, and estimates of past
probabilities of purchase could easily be biased by selective recall. For this
reason, we should be cautious about these research findings. In a recent
article, East et al. (2016) review how impact should be measured.
What Variables Affect the Impact of WOM?
East et al. (2008) measured how six variables affected WOM impact, where
impact was measured as change in the intention to buy. These were: the
prior probability of purchase; how strongly expressed the WOM was;
whether the WOM was about the main brand; the closeness of the
communicator and receiver (that is, whether a close friend/relative or not);
whether the WOM was sought or not; and how much advice the respondent
reported giving on the category that was studied. These factors were used in
a regression analysis to predict impact. Table 12.5 shows the output from
the analysis. We see that the prior probability of choice is the most
significant factor, supporting the gap argument in the previous section. For
PWOM, the greater the prior probability, the less the change (and the
reverse for NWOM). The strength of WOM expression, a variable noted as
important by Mazzarol et al. (2007), is a strong determinant of impact.
Also, PWOM about the currently used main brand has more effect than
PWOM on other brands, while NWOM on the main brand has less impact
than NWOM on other brands. The closeness of the communicator and
whether the advice was sought are only significant for PWOM, and the
amount of WOM given by the respondent is only significant for NWOM.
Previous work has shown that close ties have more effect than distant ties
(Brown and Reingen, 1987) and that sought advice is more influential than
advice that is unsought (Bansal and Voyer, 2000; East et al., 2005a). The
weak associations shown in Table 12.5 may relate to the method of analysis.
When multiple regression is used, other variables that are associated with
both the predictor and outcome variables can assume part of the
explanation.

An interesting feature of Table 12.5 is the similarity in the magnitude of the

different determinants, as shown by the beta coefficients. This indicates that
PWOM and NWOM are similar in kind. We might expect this since they
are both advice that is often intended to help the recipient.
The Effect of Brand Commitment
East et al. (2008) analysed the shift in purchase probability against the
probability of purchase prior to receiving WOM. The result is shown in Figure
12.2. For most of the range, there is a close relationship between impact,
measured as shift in purchase probability, and prior probability of purchase.
However, people who are very likely to buy a brand give less weight to
NWOM on that brand and people who are very unlikely to buy a brand give
less weight to PWOM on the brand, perhaps because they intend to buy another
brand. Thus, Figure 12.2 shows how commitment to brands can make people
resistant to advice about alternatives. Figure 12.2 is useful because it helps us
to see how consumers differ in their response to information, depending on
their prior probability of purchase.
Section 4: Wom in The Social Network
One stimulating development on WOM has been work by Watts and Dodds
(2007), who cast a critical eye over the two-step flow model of mass media
influence which was proposed by Lazarsfeld, Berelson and Gaudet (1944) and
refined by Katz and Lazarsfeld (1955). In this model, mass-media
communications are processed by a small group of opinion leaders, who pick
up ideas from the mass media, interpret them, selectively pass them on, and
thus promote or oppose change. Sometimes opinion leaders will enrol further
opinion leaders to spread the word, as shown in Figure 12.3. This is a one-way
process akin to the spread of a disease. Watts and Dodds (2007) suggest that
the opinion leader is usually only modestly more influential than average. They
argue that influence may flow in both directions in networks of individuals,
and in the more fluid social network that they propose innovations take off
when a critical mass of easily influenced individuals has been reached. When
this occurs, there is a large-scale cascade of adoption.

Figure 12.2 Shift in probability of purchase (impact) as a function of the

probability of purchase before receiving WOM
Figure 12.3 Flow of influence via opinion leaders
Unfortunately, Watts and Dodds do not really explain how transmission can go
in both directions. One possibility is that when existing users of a brand hear
their brand recommended, they raise their level of recommendations. In some
unpublished work at Kingston University, we asked respondents whether they
had heard others recommending a service that they had recently used, and we
also asked them how many times they had recommended the service
themselves in the preceding four weeks. We found that, on average, those who
had heard others recommend a service gave nearly twice as many
recommendations compared with those who had not heard their brand
recommended. Even after removing the effect of some covariates, the effect
remained. One explanation for this is that people can easily repeat
recommendations that they hear. This provides a mechanism whereby influence
can travel over the network of existing users in any direction, though a variety
of other mechanisms are likely to be involved.

Such ideas may also help us understand how advertising can affect WOM. It is
known that the level of WOM on a product rises when it is advertised (Bayus,
1985; Graham and Havlena, 2007). According to Keller and Fay (2009), 20 per
cent of WOM discussions refer to paid advertising content. The ad may
increase the salience of the brand so that previously used PWOM scripts are
more likely to be expressed, or the ad may supply a script that a receiver can
repeat – this seems more likely for print and radio ads where information is
already in a verbal form that can be passed on. This suggests that ads should be
tested to see whether they do promote WOM.
Section 5: Applications of Word-of-Mouth
Research
Net Promoter Score
The Net Promoter Score (NPS) is intended to measure the number of people
who are positive about a brand/company (promoters) and the number who
are negative (detractors) (Reichheld, 2003). The score is computed by
subtracting the number of detractors from the number of promoters, as
shown in Figure 12.4. The NPS asks about the intention to recommend, but
Romaniuk, Nguyen and East (2011) found that intentions reflect the WOM
that responders have given in the recent past; apparently, when asked what
they will do, people will check on what they have done. Past behaviour is
likely to be a guide to future behaviour, and in Chapter 2 this was supported
with respect to purchase. Thus, it might be better to measure past WOM
directly rather than use a question about what responders will do.

Figure 12.4 Measuring the Net Promoter Score

In the NPS, detractors are meant to give much of the NWOM on a brand.
This seems doubtful; those who give little PWOM may just be disinclined to
give WOM in any form. This was tested by East, Romaniuk and Lomax
(2011). They identified promoters and detractors in three categories and
found out how much PWOM and NWOM these people claimed to produce
in the last six months. The detractors produced very little NWOM, and in
two of the categories they gave more PWOM than NWOM. Bear in mind
that the NPS is based on customers and much of the NWOM on a brand is
produced by ex-customers who were included in the East et al. study but are
not included in the NPS measure. It appears that the NPS provides a good
indication of PWOM, but not NWOM because of the focus on current
customers. Also, if we wish to evaluate the way that WOM supports a brand,
it would be better to measure the amount of WOM received (the NPS is
based on the amount of WOM given).

One of Reichheld’s contentions is that the NPS is a better measure of

company performance than satisfaction. The best-known measure of
satisfaction is the American Customer Satisfaction Index (ACSI), first
developed in Sweden by Fornell (1992) and discussed in Chapter 9. The
predictions of the ACSI and the NPS have been compared (e.g. Keiningham
et al., 2007; Morgan and Rego, 2006) and generally the ACSI has been
superior. However, because both the NPS and the ACSI are restricted to
recent customers, there is potential for a better measure that covers all
consumers.
Influentials or Current Users?
In the two-step flow model, advertising is relayed by a limited group of
‘influentials’ or opinion leaders who recommend widely. Thus, a popular
strategy is to identify these influentials and direct communications to them.
As we have seen, Watts and Dodds (2007) criticized the two-step flow
model and suggested that influence was more widely spread in the network.
Related to this, Balter and Butman (2005) argued that WOM was more
effective when it was delivered by ordinary people. Furthermore, research
by Goodey and East (2008) showed that those who scored high on the
mavenism index (Feick and Price, 1987) did not give much more WOM
than those who scored low, so it may be difficult to identify truly influential
people.

To some extent, the best strategy depends on cost. If costs are low (as when
the Internet is used), it makes sense to target all those on a customer
database. However, the messages need to differ between current users
(responsible for most of the PWOM) and ex-users (responsible for most of
the NWOM). If costs are high, it may pay to focus on the influentials; this is
what happens when BzzAgents are given products to talk about. However,
whether all users or just influentials are targeted, there is a need for research
to find out what sort of information people pass on and what impact
different forms of information have on receivers.
Promoting Positive and Stopping Negative
Comments
A popular method of promoting PWOM from current customers is the
referral programme. This is a managed intervention designed to add to
naturally occurring PWOM. Often, there is a reward for the person making
a successful referral and sometimes an incentive for the person referred.
There is evidence that the customers acquired through such campaigns are
more valuable than those acquired by other means (Schmitt, Skiera and Van
den Bulte, 2011). Referral programmes use customer databases to
communicate with those who are likely to support a brand, but another use
of such databases is to direct information to groups who could be criticizing
the product. In Table 12.3, we showed that about one-fifth of negative
advice comes from current customers and more than half from ex-
customers and both are likely to be accessible via the database. When there
is dissatisfaction with the brand, it should be possible to design messages to
reduce it. By doing so, both NWOM and defection may be reduced.
Information, Not Hearsay
When there are widespread misunderstandings about a topic, there is a
danger that strategies will be misjudged. Many beliefs about WOM appear
to have been mistaken. It is not true that NWOM is more common than
PWOM, according to the evidence that has now accumulated, and it does
not appear that there is much evidence that NWOM has more impact. The
role of satisfaction or dissatisfaction in the genesis of WOM has probably
been over-emphasized. Nor is it generally true that long-term customers
usually recommend more than short-term customers. More research
findings are needed to displace such hearsay and to inform evidence-based
marketing strategies.
Summary
PWOM and NWOM are powerful influences on consumer choice but they
are difficult to study. Internet research deals with only a moderate fraction
of those giving advice, experimental research lacks relevance to natural
settings, and survey research is prone to bias. In the absence of good
evidence, some misunderstanding has occurred. It now appears that
dis/satisfaction, though often important, may not be needed for WOM.
Therefore, comparisons between satisfied and dissatisfied customers are
inappropriate for determining the occurrence and impact of PWOM and
NWOM.

PWOM tends to be about the communicator’s current main brand and

NWOM about previously owned brands. These patterns produce a strong
association between the volume of PWOM and market share, and a
somewhat weaker association between NWOM volume and market share.
Market share thus provides a norm for the amount of PWOM and NWOM
that brands should receive on average and this allows for measurement of
better or worse performance for individual brands.

Research evidence shows that PWOM is more common than NWOM and
that, in general, PWOM has somewhat more impact on the probability of
purchase than NWOM. Impact is related to the probability of purchase
before the WOM is received, the strength of expression of WOM, and
whether the WOM is about the current main brand or not. Those people
who are very likely to buy a brand give less weight to NWOM on this
brand, and those who are very unlikely to purchase a brand give less weight
to PWOM on it.

There is uncertainty about the process whereby influence passes through the
social structure. The two-step flow model, in which the mass media affect
opinion leaders who then pass the message on to followers, has been
criticized by Watts and Dodds (2007), who argue that influence is more
dispersed and bi-directional. One suggestion is that the WOM production of
product users is increased when they hear others recommend their brand,
and this would make influence omni-directional and more dependent on
ownership than opinion leadership. Managers, seeking to influence WOM,
may target opinion leaders or they can try to influence the whole customer
base. The best strategy depends on costs; when these are low, it is better to
target the whole customer base.

The Net Promoter Score (Reichheld, 2003) is a measure of WOM, but it is

based on customers and it is ex-customers who express most of the negative
sentiment about brands. In consequence, the Net Promoter Score is a poor
measure of NWOM.
Additional Resources
To see how word-of-mouth impact varies between categories and for a
further review of the literature, see Eastet et al. (2008). It is also worth
checking the websites for the WOM agencies mentioned earlier:
www.bzzagent.com/ and www.womma.org. One marketing text has tried to
apply the new thinking covered in this chapter, namely Allan Kimmel’s
(2010) Connecting with Consumers. Some interesting new work is coming
from Wharton – check the Internet.
13 The Response to Advertising
Learning Objectives
When you have completed this chapter, you should be able to:

Give examples of effective advertising.

Discuss the issues relating to the effective frequency and concentration of ads.
Discuss the mechanisms by which advertising can increase sales or change behaviour.
Discuss which consumer segments and which types of brands are likely to be more
responsive to advertising.
Consider how developments in digital media, biometrics and neuroscience will affect
advertising.
Exercise 13.1 What do you expect?
Go through the objectives and decide what you think will be reported in the following pages.
This helps you to take in the evidence presented in this chapter.
Overview
Advertising sometimes has a powerful impact on consumption, but its effect is variable
and quite often there is no discernible outcome. We discuss the ways in which advertising
can be effective, and give examples of successful campaigns.

The advertising budget is allocated through media planning and this involves trade-offs
between reaching more buyers and having a greater number of exposures to a smaller
number of buyers. We review research on the frequency and concentration of ad
exposures and how these affect the impact of advertising. There is evidence that high
concentrations of ads may break through the resistance established by competitor brands.

The mechanisms by which advertising operates are complex. Sales effects resulting from
advertising appear to have two components: a short-term effect that is largely dissipated
within a few weeks, and a long-term effect which can last for more than a year but which
only occurs if there is a short-term effect. We discuss the bases for these effects. We also
introduce recent thinking that applies consumer memory theory to explain how
consumers take account of advertising.

Ads can have more effect on certain consumer segments and on certain types of brands.
We illustrate these effects and consider which segments provide most extra sales volume.

The media landscape continues to evolve, offering new ways of communicating with
consumers, while biometrics and consumer neuroscience are being increasingly promoted
as methods of evaluating advertising effectiveness. We discuss some impacts of these
changes.
Section 1: Effective Advertising
Ads may affect our knowledge and attitudes, but ultimately they must
change or reinforce behaviour if they are to be useful. In social applications,
ads can reduce accidents, increase voting rates, promote healthy eating, or
get people to report suspicious behaviour that might indicate terrorist
activity. In commercial applications, advertising can increase purchase and
subscription, or maintain purchase rates when the price goes up. Sometimes
the profit-making behaviour occurs at the end of a chain of prior actions,
and the links in this chain may be strengthened by advertising, such as by
getting consumers to go to a showroom or to check a product on the
Internet.

Ambler and Broadbent (2000) discuss how advertising campaigns vary in

effectiveness (how much change they achieve) and in efficiency (how much
change they achieve for a given cost). Effectiveness depends in part on
decisions about copy, such as whether to write informative or emotional
ads, the prominence given to the brand, whether to include a unique selling
proposition, the use of a celebrity endorsement, the length of the ad, the
creative idea, and so forth. For example, Zhang et al. (2014) found that
emotional appeals were more effective for experiential services, while
rational appeals were more effective for credence services. Efficiency
depends on the quality of the ad and on media planning and buying, which
should ensure that ads reach the target audience in a cost-effective manner.
Decisions about content and media schedules depend on theories about how
viewers consume media and how they process advertisements, so different
theories will lead to different recommendations. The optimal approach may
vary with the product category, whether the brand is already familiar to the
consumer, and whether the purpose is to communicate an offer, remind
people to keep buying the brand, or to stimulate a substantial change in
behaviour.

In the commercial arena, three particular outcomes may be derived from

advertising, which we discuss in more detail in the following sections.
These are:
 Price support: buyers pay more per unit and so the profit on a sale is
increased.
Sales support: buyers buy more than they would have done without the
advertising. This is the most widely used criterion of ad effectiveness.
Cost saving: costs are reduced as advertising stabilizes sales and
makes intermediaries more compliant.

These outcomes can occur via a number of intermediate mechanisms.

Advertising may:

be based on a new analysis of the product range and consumer

segments (a relaunch) – for example, Virgin focused more on its
younger customers when it relaunched the Australian mobile telephone
service in 2004 (Effective Advertising 8, 2006)1
induce word of mouth and media comment that eventually results in
purchase – Murray-Burton, Dyke and Harrison (2007) reported on a
live Monopoly game, which could be played over the Internet and
which generated measurable WOM
increase retailer stocking, raising the opportunity to purchase – the
Felix cat food campaign (Broadbent, 2000) boosted distribution so that
a third of the extra sales effect came from this source; in many cases,
expanded distribution produces a sales gain without the help of
advertising
raise demand for scarce items such as property and shares – in the case
of the One2One telephone company (now absorbed into EE), ads lifted
share values so that capital could be raised at a lower cost (Kendall,
1998).
Price Support
Some ads make consumers aware of discounts and the possibility of saving
money. This price-related advertising appeals to more price-sensitive
consumers (Bolton, 1989; Kaul and Wittink, 1995), so any benefits must
come via increased sales rather than higher margins. By contrast, most
brand advertising is designed to raise perceptions of brand quality, and this
tends to reduce price sensitivity, allowing the brand owner to raise or
maintain margins. For example, Broadbent (2000) showed that price
sensitivity about Lurpak butter dropped in regions that received more
advertising for this brand; Hamilton, East and Kalafatis (1997) found that
well-advertised brands usually had either slightly lower price elasticity or
were more highly priced than others. This combination of high adspend and
high price is a common pattern for leading brands (Farris and Reibstein,
1991). Sometimes it is suggested that the high level of advertising could be
a consequence of the brand’s success, but several case histories provide
evidence that the reduction in price sensitivity follows the advertising. For
example, in the Stella Artois case, the ads seemed to lead the sales (see Box
13.1). Binet and Field (2009) found that ad campaigns that tried to reduce
price sensitivity were more effective than those that tried to increase sales.
 Box 13.1 Stella Artois
Price support has been demonstrated by the success of Stella Artois advertising in Britain
(Baker, 1993). Stella was advertised as ‘reassuringly expensive’, to imply high quality. It
attracted a large proportion of lager drinkers despite a trade price premium of 7.5 per
cent. Publicans more than recovered this premium when they sold Stella at its higher
retail price. In 1999, Stella was priced 14 per cent above the premium lager average and
the profit increment was estimated at six times the ad cost (Broadbent, 2000).

Mela, Gupta and Lehmann (1997) studied the impact of brand advertising
and sales promotion on price sensitivity over an 8-year period. They
focused on a mature product where life-cycle effects were minimal and
found that reductions in brand advertising were associated with increased
price sensitivity. Most of this effect occurred among less loyal customers,
showing that this segment is an important target for brand advertising
campaigns.

Squeezing the Retailer

Steiner (1973, 1993) found that advertising in the toy industry could both
reduce the price to the customer and raise manufacturer margin. He showed
that advertising created a consumer demand for products that compelled
retailers to stock them, so they had to pay the manufacturer’s asking price.
At the same time, competition between retailers forced them to reduce the
selling price. As a result, consumers and manufacturers did well at the
expense of retailers. This effect is likely to be particularly strong when the
product is a ‘must have’, such as the final Harry Potter book (see Box 13.2).
Farris and Albion (1980) reviewed this subject and concluded that
advertising exerted pressure on retailers’ margins and that the net effect of
such advertising often lowered the price to the consumer. When retailers
have great power, as in the case of the leading UK supermarket groups, this
effect may be less apparent, but even in groceries there is evidence that
manufacturers have adjusted production to emphasize the stronger brands
where they have more leverage on price. In the 1990s, Procter & Gamble
and Unilever dropped a large number of small brands and focused on the
power brands which supermarkets had to stock.
Box 13.2 The trouble with Harry (from
The Guardian, 4 May 2007)
Waterstone’s owner HMV yesterday defended its decision to sacrifice profits and offer
the forthcoming Harry Potter book at half price, suggesting a price war had left it with
little choice.

HMV chief executive Simon Fox said the whole market for the final instalment of the boy
wizard’s tale would be at half price, and cited Ottakar’s, now owned by HMV, as an
example of the price to be paid for not joining in a Harry Potter price battle: ‘Not being
price competitive on the book seemed to set a perception that the store was high price.
There are very few books that have that level of publicity,’ said Mr Fox. ‘If we try to be
anything other than half price we are setting the Waterstone’s brand off as high price and
that’s something we are trying to change’.
Sales Support
The cases in the Advertising Works (UK) and Effective Advertising
(Australia) series demonstrate that advertising can increase sales. On rare
occasions, the effect is large, as in the case of Levi 501 jeans in the UK.
Here, campaigns from 1984 to 1987 raised sales 20 times (Feldwick, 1990).
But the Levi 501 case was quite exceptional. Even when the best campaigns
are reviewed, sales gains of 100 per cent or more are uncommon and tend to
go to small brands, which can sometimes increase share substantially as they
start from a lower base.

In Advertising Works 15 (Green, 2007), a small-volume brand, Actimel,

secured a year-on-year sales gain of 426 per cent; the much bigger company
O2 gained about 35 per cent on contract customers and 100 per cent on pre-
pay customers as a result of a successful campaign in a rising market.
However, the payback of a campaign can be much more substantial when the
brand is bigger. The O2 payback was as much as 80 times the ad cost when
all possible benefits were included. The Actimel payback was much lower,
at about 1.7 times, because of the small size of the brand. In other examples
from Effective Advertising 8 (2006), the Sunbeam electric blanket brand
gained 83 per cent but the payback was only 1.8 as the market was relatively
small; compare this with a campaign for Australian lamb that raised sales by
about 25 per cent over five years, but gave a payback in this big market of
53 times the ad cost. So, ad campaigns can give very good returns for big
brands, but it is hard to show a large payback for small brands, even when
market share is substantially increased.

Sometimes, even static sales are an achievement if, without the advertising,
there would have been a decline. In Australia, Hahn Premium light beer
expected to lose its leading position because of a build-up of intense
competition but its campaign successfully countered the attack and the brand
even gained a little share (Effective Advertising 8, 2006). The Australian
lamb advertising corrected a long-term decline in lamb sales, so the 25 per
cent gain on sales from the start of that campaign is probably an
underestimate of the advertising achievement.
One very successful campaign in the UK was conducted by the ad agency
TBWA for Wonderbra. This ad probably drew attention because it was
puzzling. The model (Eva Herzigova) was unconventionally provocative and
the quote was mysteriously enigmatic to those unfamiliar with the work of
the film star, Mae West. The cost of the initial four-month campaign was
only £330,000 as billboards were the predominant medium used (Baker,
1995). Over a two-year period, a gain in sales of 120 per cent was achieved,
even though Wonderbra was selling above the price of many other brands.
The key to this success was almost certainly the substantial editorial
comment and word of mouth (WOM) that the advertising provoked,
including discussion of how advertising could distract drivers and cause
accidents. Although this is an extreme example, WOM is an important
secondary effect of advertising. Keller and Fay (2009) found that 32 per cent
of online WOM and 21 per cent of offline WOM referred to paid
advertising, and that this involved more recommendation than other WOM.

The Advertising Works and Effective Advertising cases have been selected
here because they are successful. Most advertising for established brands
produces far less sales response. This is illustrated in a report by Riskey
(1997) on 23 Frito-Lay ad campaigns. This study compared brand sales
when ads were running with a no-ad control condition. The study was
conducted using the BehaviorScan method of Information Resources Inc.
(IRI), which is described in more detail in Box 13.3. Twelve campaigns
showed measurable effects, and these cases produced an average sales
increase of 15 per cent.
 Box 13.3 BehaviorScan
Information Resources Inc. (IRI) uses cable TV in specific towns to test ads. Households
are recruited to a panel and agree to receive television that may be modified by IRI. The
Behavior Scan technology swaps commercials so that some households receive trial ads
or extra exposures of normal ads when compared with other households. The former
allows copy tests to be conducted, the latter weight tests. Members of the panel show an
identification number when they buy groceries in town. IRI finances the scanners in the
town’s stores and downloads sales information each night from those scanners. The
system allows sales to be tied to households receiving different frequencies of
advertising.

This system permits experimental tests but suffers from some weaknesses:

Members of the household may not be watching a TV set even though it is on.
Out-of-town purchases (out-shopping) are missed.
The tests exclude trade response. National advertising may generate more retailer
stocking and competitor advertising than in the test communities.
The brands that are tested are chosen for commercial reasons and this may bias the
sampling.
There may be a ‘hothouse’ effect if panellists guess that commercials are on test
and, as a consequence, take more interest in them.
Word of mouth between test and control panellists may compromise the
experiment.

As a method of testing copy, the BehaviorScan procedure takes a long time

and is expensive. The ARS Persuasion Measure is a cheaper, quicker, but
less reliable method – this uses the shift in intention to purchase after
exposure to ads (Blair and Rabuck, 1998).
Cost Saving
In some cases, advertising produces efficiencies that reduce cost. For
example, Volkswagen saved on storage costs when extra demand meant that
they had fewer cars unsold (Kendall, 1998). Costs may also be saved when
advertising is accurately targeted so that irrelevant inquiries are avoided.
Internet job advertising can get replies from anywhere in the world and
such ads should be designed to cut out applicants who cannot be appointed
by virtue of their location or nationality. Kendall (1998) showed the value
of well-targeted advertising in the campaign to recruit personnel to the
British Army. In 1994, one person from every 6.7 inquirers was enlisted.
Following the advertising campaign, the conversion ratio improved to 1 in
3.4. In the analysis, it was estimated that this change in ratio saved the
Army £16 million after deducting the cost of the advertising. In addition, it
appeared that better recruits were enlisted since they were less likely to
drop out during the period of initial training.
The Effects of Social Advertising
Large paybacks are quite often found in social applications of advertising.
For example, a £31 million campaign to raise rear seatbelt usage in the UK
gave a directly quantifiable return of £18 million, and when further
assumptions were made about the costs of injury and death the return was
£73 million (Broadbent, 2000). Another campaign in Australia achieved a
drop in smoking of over 7 per cent, equivalent to 190,000 fewer smokers.
The healthcare saving was estimated at $24 million (Effective Advertising 6,
2001). Often, social advertising has no opposing advertising but has to work
against consumer inertia (e.g. energy saving) or self-indulgence (e.g. eating
less).

Advertising Works 20 (Snow, 2011) gives the example of Marie Curie

Cancer Care changing their traditional annual advertising campaign to focus
on distribution, and soliciting people to act as collectors rather than as
donors. For an outlay of £184,000, they recruited an additional 5,219
collectors, who generated an extra £634,583 in donations. Evidence from
Hanssens (2009) and Ataman et al. (2010) shows that distribution has six to
ten times the sales impact of advertising. Thus, to the extent that advertising
promotes additional distribution of this sort, its effect is greatly multiplied
for both social and commercial advertising.
Box 13.4 Successful ads
The wide variation in the effectiveness of ads has led to research about which elements of
an ad make it effective. Armstrong (2010) summarizes findings from 3,000 studies into
around 200 normative principles on how to write effective ads. For example, Principle
3.1.1 notes ‘Do not mix rational and emotional appeals’, and Principle 6.9.2 notes ‘Attack
ads should employ objective information, not emotion’. Armstrong et al. (2016) show that
copy that follows these principles is more effective. Similar results, using a different
copywriting codebook, have been found by Harnett et al. (2016).

While principles for developing effective copy might help to avoid mistakes, creativity is
still required to produce ads that gain attention and engagement from the audience. By its
nature, creativity cannot be anticipated. But, after its creation, we can see features of an
ad that help to make it successful. For example, the Wonderbra campaign was noticed
because it created curiosity at a number of levels. Among these were the enigmatic and
challenging character of the model, the oddity of putting such ads on billboards, and in
the ad shown, uncertainty about the origin of captions such as ‘or are you just pleased to
see me?’ This comes from a line Mae West says to Cary Grant in the film She Done Him
Wrong. The full line is ‘Is that a pistol in your pocket, or are you just glad to see me?’
Section 2: Advertising Frequency and
Concentration
Schedules
Advertising is presented in media according to a schedule. Traditionally, for
TV, radio or print media, ad exposures may be continuous (delivered at a
steady frequency per month) or in bursts (e.g. one month on and two months
off). When the bursts are short-interval (e.g. a week), this pattern may be
called pulsing. Sometimes, a low level of advertising or drip is maintained in
the gaps between bursts. The choice of schedule should be determined
primarily by its sales impact on consumers. Continuous schedules spread the
advertising across a larger number of people but each person tends to see
fewer exposures compared with bursting. As the time for the burst is
reduced, there is more concentration (the exposures occur over a shorter
period) but fewer buyers may be reached by the ad during this reduced time.

Figure 13.1 Concave and S-shaped responses to exposures

In order to choose the most effective schedule, we need to know how

individuals respond to each additional ad exposure and how they react to the
same number of exposures when these are concentrated into different time
intervals. If each additional exposure produces a smaller sales effect than the
last, the response is concave to the x-axis and is an example of diminishing
marginal returns (see Figure 13.1). When this occurs, the most cost-effective
number of exposures (known as the effective frequency) is one per person
and it is best to use a continuous schedule that spreads the advertising across
the target population as widely as possible. In this way, more people are
reached at lower frequencies. But if additional exposures produce increasing
and then decreasing increments in sales, which is an S-shaped response
curve, then the best strategy is to use a schedule that takes the audience
quickly to the point where their sales response is steepest. This strategy
sacrifices ad penetration for ad frequency. It is achieved best by bursting and
pulsing schedules.
S-Shaped or Concave Response?
Simon (1979) analysed data from a study by Zielske (1959) and found that
the second and later exposures gave a concave response (diminishing
returns). In further evidence, Simon and Arndt (1980) reviewed 37
advertising studies and established that the great majority showed a concave
response function. Roberts (1996) found that the response curve was
concave in 15 out of 17 cases of mature brands. McDonald (1995) also
argued that the true pattern of sales response to advertising exposures is
concave for mature brands; his earlier work (McDonald, 1970) supported
this when it was re-analysed. McDonald was influenced in part by work
done by Jones on single-source data (Jones, 1995a, 1995b, 1995c), which
showed little gain after the first exposure. In studies by Adams (1916) and
by Burnkrant and Unnava (1987), three showings of different ads for a
brand were compared with three showings of the same ad. There was more
effect on purchase propensity when the ads were different. This indicates
that the second and third exposures of the same ad had less effect than a
new ad; the curve was concave.

Despite this evidence, there is practitioner support for the S-shaped

response function. This is partly because the difficulties of research in this
field make results uncertain. In particular, there is a problem posed by the
decay of any ad effect as the period of time between exposure and
measurement increases. Also, the time between exposures can vary, thus
varying the concentration. Broadbent (1998) pointed out that there is very
little agreement among researchers and practitioners on the period over
which exposures should occur when frequency effects are studied; three
exposures could occur in a day, a week or a month.

There are good reasons why the response to exposures should be S-shaped.
An important argument is based on the idea of breakthrough, getting over a
threshold of attention so that the audience cannot miss the message
(Broadbent, 1998). An S-shaped response is also implicit in Krugman’s
(1972) three-hit theory. Krugman argued that on first exposure, viewers are
curious, on the second the meaning of the ad may become clear and they
endorse or reject the message, and on the third and subsequent exposures
they are reminded of the message again and may take action. In this
account, the second and third exposures are more effective than the first. An
influential book by Naples (1979) endorsed the three-hit theory.

Aggregate evidence gives some support for an S-shaped response function.

Lodish and Lubetkin (1992) analysed IRI data and found that both new and
established products did better when the advertising was initially
concentrated, rather than spread over time. This suggests that exposures
need to exceed some threshold level if they are to have optimum effect.
Schmidt and Eisend (2015) summarized the effects of many previous
studies of advertising response, and found that ad and brand recall increased
linearly up to eight exposures, while brand attitudes continued improving
up to ten exposures. This supports the arguments against a concave
response function but fails to show the threshold effect expected for an S-
shaped function. However, Schmidt and Eisend’s (2015) summary relies on
laboratory experiments rather than field studies, so the relevance of these
results to practical marketing is in doubt.

This evidence has left considerable uncertainty, which may be resolved by a

detailed discussion of work on concentration and breakthrough.2
Concentration
Roberts (1999) provided evidence on the way concentration affects the
outcome of repeated exposures to ads. He used a UK data set on 750
households from a Taylor Nelson Sofres Superpanel, which was gathered by
TVSpan. Each household was equipped with a TV ‘setmeter’ so that
viewing, including ad exposures, could be recorded. The study covered 113
brands from 10 categories advertised over a two-year period. Roberts
controlled for two major co-variates: concurrent sales promotions and
weight of television viewing. He compared respondents who had received
exposures on a brand with these same respondents when, over another
period, they had not received any ad exposures on the brand for 28 days.

Figure 13.2 Percentage increase in sales for different frequencies and

concentrations (Roberts, 1999) Recency, frequency and the sales effects of
TV advertising, Admap, February, 40–44. © Warc.

Roberts (1999) conducted a number of analyses. Here, we focus on the

effects of three exposures in three different intervals: one day, three days and
28 days. Figure 13.2 shows the sales recorded after the end of the exposure
interval. When the exposures occur over 28 days, the additional effect of the
second and third exposures is small and follows the familiar concave pattern.
When all three exposures occur in the same day (and in practice this is often
over a few hours), the effect of the second and third exposures is large and
produces a convex sales response, which could be the lower part of an S-
curve. Over three days, the pattern is more linear. There appears to be an
interaction between exposures when these occur over a short period but not
over a long period. Thus, multiple exposures, when close together, assist
each other to achieve breakthrough.

Roberts’ evidence has not been corroborated by other research and it should
be closely scrutinized. First, we note that the data are gathered in Britain
where ad clutter is relatively low. This raises the possibility that stronger
effects from ad concentration might be observed in high-clutter
environments. Second, the study is restricted to groceries. Third, doubts may
be felt about the effectiveness of the control comparison used in this work
(purchases made by the same respondents when they had not been exposed
to the advertising for 28 days). However, in the absence of other evidence
Roberts’ results remain important.
Breakthrough
Breakthrough relates to the psychological concept of interference, which
occurs when learning about one concept reduces the ease with which a
related concept is processed or recalled. Thus, learning more about Brand A
interferes with the processing or recall of the competitive Brand B, and vice
versa. Breakthrough occurs when the concentration of exposure to ads for
one brand overcomes the interference set up by past learning about other
brands.

Burke and Srull (1988) demonstrated interference effects in the recall of

advertising; in their study, as competitive brand advertising was increased
the recall of target brand details decreased. This evidence adds support to
the argument that Roberts’ results were obtained because the greater
concentration overcame interference from competitive brand ads. A
weakness of Burke and Srull’s study is that it was based on recall and
lacked any observation of sales effect; but the effect has now been found in
market data by Danaher, Bonfrer and Dhar (2008), who examined the
impact on sales of a focal brand when competitors advertised within a
week. Danaher et al. found a substantial reduction compared with the sales
that would have occurred without competitor advertising.

Practitioners have always been concerned to measure share of voice – the

proportion of advertising exposure that a brand gets compared to its
competitors. The research by Danaher et al. suggests that this simplifies the
issue too much, as it takes insufficient account of the way interference from
competitive advertising occurs. The effects of interference will depend on
the timing of ads, as well as on the closeness and strength of the links in the
relevant associative network, as discussed in Chapter 3. As a result,
interference will vary between competitive brands – some competitors are
more similar in their associative network and so their advertising will
present a greater threat of interference. There is also a background level of
interference set up by all the advertising received that inhibits the response
to any ad. Concentrated exposures may help to overcome this clutter, thus
increasing the attention given to the focal brand.
Section 3: A Model of Advertising Effect
Some ads are aired a great number of times. This suggests that an
incremental effect continues after many exposures – this is likely because
the brand is kept salient relative to competitors through repetition of the
brand name and its associated elements (e.g. logo, characters, colour, jingle;
Sharp, 2010). Therefore, we need a model of advertising effect that describes
the consequence of early exposures, as well as the continuing effect from
what is often a large number of subsequent exposures. Figure 13.3 illustrates
such a model in which advertising has two types of primary effect on an
audience.

Under this model, the first few presentations of the ads may secure
additional interest and may sometimes modify thinking about a brand. This
thoughtful initial response is likely to have a substantial sales effect, when it
occurs. Once consumers become accustomed to the ad, they will no longer
give it much thought; however, a second primary effect may continue to
operate, produced by the low-involvement automatic mechanisms that help
to maintain brand awareness. Mechanisms of this sort produce weak effects
but may continue to work over many repetitions as they refresh the
associative network described in Chapter 3. This dual-process account has
some parallel in work on persuasion by Fazio (1990) and Petty and Cacioppo
(1986). Even if existing consumers of a brand do not pay much attention to a
new ad, they may still be affected by these passive low-involvement
processes.

Figure 13.3 also shows secondary processes that may occur later. These
secondary effects generally produce weak effects on sales, but because they
are sustained over long periods may contribute substantially to the total sales
benefit from advertising. Studies of the connection between primary and
secondary effects consistently show that a secondary effect occurs only
when there has been a primary effect on sales. Abraham and Lodish (1990)
reported that, if advertising tests do not show an effect after six months, they
are unlikely to show any effect later. Jones (1995a) found that no one-year
ad effect was observed if a short-term effect was not detected in the first
seven days after exposure. Riskey (1997) observed that longer-term effects
in 12 ad campaigns occurred only when there were shorter-term effects.
Lodish et al. (1995b) found no delayed effects from ads that were ineffective
in their test year. This evidence shows that any secondary effect is an
outcome of the primary effect; however, a primary effect does not guarantee
a secondary effect because it could be countered by competitor
communications.

Figure 13.3 Primary and secondary responses to advertising

The secondary effects could arise in a number of ways. There may be

framing when ads modify thinking to produce a persistent change in the way
that a product or brand is perceived. For example, ads may persuade
customers to think of their phone as a replacement for their camera. Usually,
the changes produced by advertising are more modest and adjust the
positioning of the brand, rather than its categorization. For example, ads may
establish that a car model is more economical than previously thought. These
shifts in thinking, whether dealing with the category or the attribute, are
likely to be first produced by thoughtful primary processes rather than by
automatic mechanisms.

Another process that could produce a secondary effect is purchase

reinforcement. When the short-term effect of an ad is to induce extra
purchase, this additional purchase experience may strengthen the propensity
to buy the brand in the future. Purchase reinforcement could occur because,
after buying a brand, it comes to mind more easily, as experience-based
information is more easily recalled (Fazio, 1986; Fazio and Zanna, 1981;
Fazio, Powell and Herr, 1983). Purchase reinforcement could also be based
on knowledge of where to buy the brand – which store and the location in
the store. One objection to the purchase reinforcement explanation for
advertising is that there can be no such effect in the case of consumer
durables; people who have just bought a new cooker are not in the market
for another one. Because of this, Givon and Horsky (1990) suggest that
advertising induces a different effect for durables. They suggest that those
who receive the advertising may become potential adopters and then may
find out more about the product from others, particularly those who have
already acquired it.

A third effect occurs when advertising raises distribution and reduces stock-
outs so that the product is more easily purchased. More efficient stocking
may occur either because retailers anticipate demand and order more when
they are advised about forthcoming ad campaigns, or because the extra
demand from an ad campaign forces retailers to increase stock. Retailers
may maintain the higher stock level after the advertising has finished, thus
further boosting sales.

A fourth process is social influence, either as positive word of mouth when

consumers recommend the product, or as observational learning when public
use of products catches the attention of others, as with fashion goods. The
idea that new usage and brand switching are often started by
recommendation is undeniable, while the role of social influence is explicit
in theories about the diffusion of new ideas and products (Chapter 5). In
Chapter 12, we reported evidence that word of mouth is increased by
advertising (Bayus, 1985; Graham and Havlena, 2007) and discussed
mechanisms that might bring this about.
The Decay of Advertising Effect
The extra propensity to buy produced by advertising declines over time and
this decay process will affect both the primary and the secondary effects of
advertising. Decay usually shows an exponential pattern, like that of
radioactivity; the rate of decay is constant but, since it applies to a
diminishing quantity, the change in the whole becomes less and less as it
approaches some base level. Such patterns are usually described by their
half-life – the period of time required for activity to decay to half its
original level. Some of the decay may be due to forgetting and some to
interference effects from competitive brands.

Short-Term Decay
Early studies estimated advertising half-lives on the assumption that there
was a single process of ad decay. On this basis, Broadbent (1984) claimed
that, for most brands, half-lives were in the region of 4–6 weeks. A meta-
study of 70 brands by Clarke (1976) indicated half-lives in the range of 4–
12 weeks. Subsequently, Broadbent and Fry (1995) suggested that ad decay
had both short-term and long-term components. Roberts (1999) measured
average short-term ad decay for frequently purchased brands. He found that
this fitted an exponential curve with a half-life of 16 days. This means that,
on average, an exposure loses 4.4 per cent of its sales effect each day and
72 per cent after 28 days. This rapid loss of effect has practical implications
(see Exercise 13.2).
 Exercise 13.2 On which day should you
advertise?
Groceries have an uneven pattern of purchase over the week. Spending is heavier on Thursday, Friday
and Saturday when compared with Sunday, Monday, Tuesday and Wednesday. The Nielsen figures
are shown below.

Weekly UK supermarket sales by day of the week (percentage of total, supplied to us by Nielsen in
2015):

Which day should you advertise if the impact of your ads decays each day?

What factors beside decay might affect your decision?

Long-Term Decay
Lodish and Lubetkin (1992) investigated long-term decay using IRI data on
upweight tests. In their study, 44 brands received 50–100 per cent extra
advertising during a test year and approximately half of the brands showed a
sales increase that year. Lodish and Lubetkin analysed the extra sales achieved
for these brands over the next two years, after ad spending had returned to
normal. They compared the results with a control condition where there had not
been any extra advertising. Table 13.1 shows the extra sales for the upweight
group compared to the control group. The extra sales in years 2 and 3, after the
upweight had finished, were roughly equal to the extra sales during the test year.
This work was criticized for excluding cases where there was no sales impact
from increased weight, since these cases might have shown a response in later
years, but Lodish et al. (1995b) checked and found that there was no such later
response.

Roberts (2000) investigated long-term effects by dividing consumers into those

who had – and had not – been exposed to advertising for grocery brands in the
previous 28 days. His analysis took account of weight of viewing, concurrent
promotions and brand size. He found that repeat purchase of the brand in the
subsequent 12 months was higher if the ad had been seen. Roberts assessed the
extra sales in the year as 5.6 times the short-term increase in sales over a month.
When these extra sales are taken into account, advertising will often pay off over
a year. Hanssens, Parsons and Schultz (2001) also found evidence of a
substantial long-term ad effect. Wood and Poltrack (2015) showed that for
television ads, short-term effects in the first 12 weeks were about half of the total
effect over the first year.

Figure 13.4 How primary and secondary effects of advertising may combine
with base sales

Figure 13.4 shows both short-and long-term effects of advertising. Here, a burst
of advertising produces a short-term primary effect. The long-term secondary
effect is based on the extra sales generated by the short-term effect; it has a
smaller amplitude but lasts much longer. The combined effect sums the base
sales, the long-term and the short-term effects.
Box 13.5 New thinking on effective
advertising
Past work on effective advertising has often emphasized the measurement of aggregate
sales responses. New approaches give more emphasis to understanding consumer
memory networks. An example of such an approach is provided by Wright (2016), who
proposed a model of advertising effect based on experimental evidence about how
memory and emotion operate (this could be seen as an evolution of Figure 13.3). His
model is born of concerns that existing approaches are incomplete, over-generalize, or
have no obvious association with known brain processes. Wright suggests that effective
advertising has four requirements:

1. ‘An opportunity to see, or being present in the media to which the consumer is
exposed.’

 This is the role of media planning, to maximize the opportunities to see within
the advertising budget available. The exact selection of ad placements will depend
on the advertiser’s views about effective frequency, and on the desirability of using
a variety of media platforms to create campaign synergies (Jeannes and Beal,
2012) or to achieve concentration by making use of multiple touchpoints within a
single day (e.g. morning TV, drive-time radio, outdoor advertising, digital media).

2. ‘Attention, drawing eyes-on or ear-on and then triggering either conscious or

subconscious presence in a transient area of consumer memory.’

Work on attention to advertising has tended to examine focused visual attention.

This neglects the role of automatic processing that may occur for brand logos,
characters, colours or jingles (Sharp, 2010) or emotional content (Heath, 2012). It
also provides an incomplete description of how peripheral content may be drawn in
to sensory memory and prime the recall of brand-related information. For example,
consumers may peripherally notice some purple swirls on a merchandising display
in-store, with the result that they are primed to think of Cadbury chocolate when
they feel hungry, even though they never paid much attention to the display.

3. ‘Activation, being some kind of processing of ad content in long-term memory,

either through the activation of existing memory networks, conscious elaboration
of the content of the ad, or subconscious development of mild emotional
attachments to the advertised brand.’

For advertising to work, it must be present in long-term memory in some form so

that its effects can be evoked at the point of purchase at a later time. The move into
long-term memory typically occurs through the dual processes discussed earlier in
Figure 13.3, with either thoughtful initial processing or ongoing low-involvement
automatic processing that strengthens the associative memory networks described
in Chapter 3.
 Wright suggests a third mechanism to supplement these dual processes – the
creation of a mild emotional attachment. He treats this as different from automatic
processing as he believes that it relies on a different process within the brain;
specifically, that mild emotional attachments arise from a primitive area of the
brain responsible for emotional processing. This area of the brain may overlay an
emotion on a memory to produce what Damasio (1994) calls a somatic marker.
Assuming that the emotion is positive, as is usually the case with advertising, this
will result in a long-term propensity to slightly favour a particular concept or
situation – a mild emotional attachment. Wright’s suggested mechanism is
consistent with Heath’s (2012) work on low-involvement emotional processing of
advertising, as well as with Damasio’s (1994) work on brain structures and
decision-making.

4. ‘Retrieval, so that some of the results of this mental processing are evoked at the
point of purchase, with a resultant effect on consideration and choice.’

The past processing of ad content will have no effect unless the relevant memories
(or mild emotional attachments) are brought to mind at the point of purchase. The
chances of such retrieval will be affected by the stimuli that consumers are exposed
to just before they make a purchase decision. In other words, even though
advertising is processed, it runs the risk of being lost in the vast store of consumer
memory unless it is specifically evoked. Wright comments that we know relatively
little about how past processing of advertising is evoked at the point of purchase,
and encourages more research in this area.

Wright (2016) is a critic of the exclusive use of sales response to evaluate advertising
effect. This is partly because such evaluations usually take little account of stimuli
encountered at the point of purchase and so can never provide a full explanation of
advertising effect, but also because the sales response is often relatively small in
magnitude and so easily masked by the natural variations in sales common in most
markets. Thus, Wright argues for the use of recall-based measures of advertising
effectiveness to supplement the analysis of sales response.
Exercise 13.3 Reconciling models of
advertising effectiveness
1. This chapter has discussed advertising effects such as supporting higher prices,
encouraging more sales and building distribution. Yet Wright (2016) suggests that
advertising operates through consumer memory. Discuss how a stronger presence in
consumer memory may or may not support higher prices, encourage more sales or build
distribution.
2. Other mechanisms discussed in this chapter include memory interference, social
influence and breakthrough. Discuss how this can be understood in terms of consumer
memory. That is, how can these effects be explained by the ways that advertising is
processed into memory?
Section 4: Specific Effects
Whatever the mechanisms that underlie advertising effect, the response may
be greater for some types of consumer segments or brands than for others.
Prior knowledge of these differences may offer a substantial practical
advantage to advertisers.
High and Low Loyalty Segments
Raj (1982) found the maximum sales response to advertising occurred
among those with a share-of-category requirement (SCR) loyalty of 50–70
per cent. The purchases of other brands were not much affected when the
focal brand gained – the extra sales in the 50–70 per cent loyalty group
came mainly from an increase in category volume. Tellis (1988a) also found
that ads had more sales effect on buyers with an SCR greater than 50 per
cent. Those who had not bought the brand before showed only a small
volume sales response to the advertising compared with loyal buyers.

Roberts (1999) looked at percentage increase rather than volume increase.

He found that, on average, the low-loyalty segment (SCR less than 10 per
cent) showed the greatest percentage increase and the high-loyalty group
(SCR more than 50 per cent) the least. Baldinger and Rubinson (1996)
found that there were three times as many buyers with an SCR of less than
10 per cent compared with buyers with an SCR greater than 10 per cent.
Although the extra volume per individual low-loyalty buyer is small, the
aggregate volume gain from the large segment of low-loyalty buyers can be
very substantial when compared with the aggregate volume gain from the
smaller number of buyers in the high-loyalty segment. Using the segment
sizes found by Baldinger and Rubison, we calculate that the low-loyalty
segment (SCR less than 10 per cent) gave a larger increment in volume than
that from all the rest of the buyers.

Wood and Poltrack (2015) found that, for supermarket products, most of the
incremental revenue gained by advertising came from previous brand
buyers who had switched back to buy the advertised brand. Some of these
buyers went on to show high levels of loyalty over the subsequent year. So,
while the low-loyalty segment is an important advertising target, these
buyers do not necessarily stay low loyalty.

This helps us to understand the rather different roles performed by one-to-

one direct marketing and media advertising. Sales growth may more easily
arise from low-loyalty customers as there are more of them and they have
the most scope to increase purchasing. Media advertising provides effective
access to this large number of low-loyalty buyers. Some low-loyalty buyers
may evolve into high-loyalty buyers, and advertising can start this process.
However, in direct marketing, it pays to focus on those with high initial
loyalty if an approach to a buyer is expensive, and promotion should be
widely directed. If costs are low, this constraint does not apply, as the lower
return from the mass of low-loyalty buyers will still be worthwhile. Wood
also points out that the sales responsiveness of different loyalty segments
sometimes varies considerably (Leslie Wood, personal communication, 15
December 2015); so if marketers are able to allocate more of the media
spend to the most responsive segments, say through direct marketing, then
advertising efficiency may increase.
Heavy and Light Buyer Segments
Some of the discussion about low- and high-loyalty segments also applies
to light and heavy buyers. In Chapter 4, we saw that the distribution of
buyers’ purchase frequencies follows a Gamma distribution. This means
that there are many more light than heavy buyers so that the small volume
changes made by many light buyers (and non-buyers) can aggregate to a
much bigger increase in sales than the large volume changes made by few
heavy buyers. This is what we see: sales growth comes principally from the
light buyer/non-buyer segment. Advertising can reach light buyers and non-
buyers and is usually involved when a brand gains share. The parallel with
high and low loyalty is not exact, as some light buyers are 100 per cent
loyal – all those who buy only once, for example, must be 100 per cent
loyal by definition. Nonetheless, a similar argument applies to both
situations – sales growth principally comes from those who are not yet
heavy buyers of the brand.
Heavy and Light Television Viewer Segments
Heavy TV viewers have been found to be relatively less responsive to
television ads. Roberts (1999) divided households into the 40 per cent with
the lightest viewing pattern and the rest. When he compared these light and
heavy viewers, the light viewers had over twice the sales response of the
heavy viewers. Roberts suggests that the greater responsiveness of light
viewers is a ‘share-of-mind’ effect, i.e. the light viewer experiences less
clutter so that each ad has more effect. Light viewers tend to watch during
the peak evening period – it is their extra viewing that makes it a peak
period.
Big and Small Brands
Brands are big mainly because they have more buyers rather than because
each buyer buys them more frequently (Ehrenberg, 1988). If the sales
response from advertising depends mainly on the number of buyers, big
brands will get more total volume uplift from an ad campaign, other things
being equal. But the evidence often shows stronger gains for smaller
brands. Riskey (1997) suggested that this may be because the small-brand
ads tend to report some advance in the brand and these ‘newsy’ ads were
more effective. Sometimes, the buyer base of the small brand is actually
quite large, even though purchases in a given period are relatively low;
long-established small brands can have a large number of past buyers who
may be activated by advertising. Another factor against big brands is that
they may have reached a ceiling where there are few potential recruits;
when this applies, advertising serves to retain buyers rather than to increase
sales. In addition, the advertising of small brands is often infrequent and
this may make it more effective when it does occur. An example of a small
brand that benefited from advertising is the Co-operative Society
(Broadbent, 2000). Here, an old and relatively small store brand, which was
very well known and had not advertised for some time, did very well when
it returned to advertising. So although big brands usually have more buyers,
a small brand may get a good response from advertising if it has not
advertised for a while, has something new to say and has a long history.
New and Existing Brands
New brands are often introduced by advertising. Unless the new brand can
be directly marketed, it must use media advertising to let potential
customers know that it exists. The responsiveness to ads, measured as
advertising elasticity, is high for new brands. Advertising elasticity is the
ratio of the proportional increase in sales to proportional increase in
advertising. Lodish et al. (1995a) found an elasticity of 0.05 for established
brands and 0.26 for new brands. The novelty of a product may help it gain
attention, but even though a new brand shows a high elasticity the gain in
volume can still be very modest because there are few base sales.
Furthermore, buyers of new brands may not stay. If they are willing to try
new brands, they may later move to try yet newer brands.
Section 5: The Evolving Media Landscape
The growth of new digital channels continues to change the face of
advertising. Blogs, message boards, emails and personal websites allow
large numbers of people to put ideas into the public domain with great ease.
As a result, an increasing proportion of consumer choice is directed by
consumer-generated media, over which suppliers have limited control. As
the Internet is used so widely, more and more of the ad budget goes to this
medium. Much of this spending is related to search engines because of the
exceptional precision of search-related advertising, and because the act of
search pre-qualifies a customer as more likely to buy. Apps for mobile
devices are now also growing rapidly, with data available from 2014
showing that the Facebook app then had about 113 million unique users in a
month, followed by YouTube at 84 million, Google Search at 78 million,
Facebook Messenger at 70 million, and Pandora Radio, Google Maps and
Gmail in the 60+ million range (Fulgoni, 2015). Other major types of
Internet activity include expenditure via display or banner ads, the use of
websites, social media and other portals to promote customer engagement
and interaction with the brand.

Apps also enable companies to interact with customers through micro-

websites, games and other applications. Advertising Works 20 (Snow, 2011)
provides the example of a charity for the homeless, Depaul UK, developing
the iHobo game app, allowing users to look after a virtual hobo and
providing the opportunity to donate in-game. Their modest £6,000 budget
was extended by seeding the app to two technology bloggers for a viral
campaign, leading to 600,000 downloads, 95 times as many donors as in
previous campaigns, and 1,021 young people being added to their donor
database.

Blogs and discussion sites have much of the character of personal

communication, even though the communicators and receivers are usually
not personally acquainted. Generally, comment on the Internet is treated as
an electronic form of word of mouth (eWOM), which may be passed on
from consumer to consumer, either like face-to-face WOM or by giving the
link in emails. Digital media companies such as Facebook and Twitter
promote WOM through sharing functions within their online social
networks. By making recommendation easy and relevant, these companies
create a forum for social engagement that they control, enabling them to
inject advertising messages into personal interactions. Other companies take
advantage of these sharing functions by seeding videos that they hope will
be virally spread to a very large audience, or providing brand content in
tweets, in the hope that their followers will retweet this content to their own
networks.

The factors underlying digital sharing may vary between the media. Araujo,
Neijens and Vliegenthart (2015) analysed 19,343 global brand messages
posted on Twitter over three years and found that, for Twitter, informational
cues are the primary cause of sharing. While emotional content and the use
of hashtags amplified the effects of informational cues on retweeting, they
had no detectable impact on their own. This differs from Facebook videos,
where strong emotions that cause a physiological reaction are the main
driver of viral sharing (Nelson-Field, Riebe and Newstead, 2013). However,
Araujo et al. (2015) and Nelson-Field et al. (2013) studied different
behaviours and used slightly different methods; Araujo et al. identified
emotional cues from Twitter text, whereas Nelson-Field et al. measured
emotional reactions directly. So, it is possible that variation in the factors
underlying sharing is due to differences in study methods rather than
differences in media.
 Exercise 13.4 Where will new media take us?
How is advertising in new media going to develop? How is this going to affect other forms of
advertising?

Fulgoni and Morn (2009) examined the effect of Internet advertising.

Despite low click-through rates, they found that mere exposure to display
ads and search marketing (such as Google Adwords) increased site visits,
search behaviour and both online and offline purchasing. Dinner, Van
Heerde and Neslin (2014) further quantified these effects, and found that
online advertising was about twice as effective as traditional advertising;
while some people exposed to the ad were stimulated to click through to
purchase, others instead visited a local retail outlet to buy the advertised
product, increasing the total effect of the online ad. Olbrich and Schultz
(2014) investigated effects in the opposite direction, from offline to online.
They found that print ads did not drive more traffic to the advertiser’s
website, but were associated with higher rates of conversion from website
visits to sales.

Television, as an advertising medium, faces competition from digital media,

from channel fragmentation due to satellite and terrestrial digital television
offerings, and from viewer use of recording devices to zip through ads.
Surprisingly, to date TV advertising has remained relatively unaffected by
these challenges. Average channel shares have dropped, due to the larger
number of channels, but total viewing has not declined and viewer
behaviour follows the same patterns it has done for decades (Sharp, Beal
and Collins, 2009). Nor has TV ad effectiveness declined (Lynch, 2015;
Rubinson, 2009). Further, fast-forwarding of ads does not seem to affect ad
recall (du Plessis, 2009), perhaps because fast-forwarding requires that the
viewer attend to the ad break, instead of engaging in ad-avoidance
behaviour.

At the time of writing, the changes that previously played out between TV
advertising are manifesting again in the growth of in-app advertising on
mobile devices. Fulgoni (2015) found that these new digital channels tend
to add to, rather than cannibalize, existing digital channels. That is, the rise
of mobile devices has not reduced the use of desktop computers, just as the
rise of desktop computers did not reduce TV viewing. Fulgoni noted that
advertising on mobile devices was cheaper, more effective and less affected
by clutter, leading him to recommend that media planners ensure they
include in-app ads in their ad schedules. The advantages of mobile apps will
likely erode over time as more advertisers participate in this medium, but
this does demonstrate the value that can sometimes be gained from
exploiting a new media channel. For now, mobile in-app advertising seems
to have simply added an additional medium that operates in a familiar
manner.
Section 6: Biometrics and Consumer Neuroscience
Another recent trend is the upsurge of interest in the use of biometrics (eye
tracking, heart rate, breathing, facial expression, skin conductivity) and
neuroscience (EEG or electroencephalograms and fMRI or functional
magnetic resonance imaging) to detect whether people are paying attention
to and processing ads. Some marketers believe that biometrics can directly
measure attention to ads, and that brain scans (EEG and fMRI) can reveal
whether advertisements are being processed either through encoding into
memory or by increasing the desire for products.

The Advertising Research Foundation conducted a large study in 2010 to

evaluate the predictions made by the vendors of biometrics and consumer
neuroscience products. Varan et al. (2015) noted that there was little
agreement between the predictions of the different products, and there were
many issues of representativeness and validity that needed to be carefully
managed – just as there are with any form of survey research. This might
seem disappointing to advertisers wanting to use these latest techniques, but
it is important to realise that this kind of research into advertising is still in
its infancy, and one can reasonably expect that performance of these novel
measures will improve over time.

The Advertising Research Foundation then sponsored a further study that

did not rely on ‘black box’ proprietary techniques from the commercial
vendors of biometric/neuroscience approaches. Instead, marketing scientists
applied the relevant methods under controlled scientific conditions so that
they could gain full information on their effectiveness. Venkatraman et al.
(2015) reported the results. They examined traditional pen-and-paper tests
of advertising effectiveness, as well as a wide range of
biometric/neuroscience tests, and evaluated the effectiveness of these tests
by modelling the relationship with in-market advertising elasticities (a
version of ad effectiveness).

Surprisingly, the traditional pen-and-paper measures (liking, change in

intent to purchase, recognition) outperformed most of the
biometric/neuroscience techniques in predicting ad elasticities. The only
detectable improvement over traditional measures arose from taking
account of the activation of the Ventral Striatum, observed through fMRI.
This is an area of the brain associated with rewards and desirability. The
best performing model of advertising effectiveness showed a large effect on
ad elasticities from activation of the Ventral Striatum, and also a large effect
on ad elasticities from ad recognition.

What does this mean for marketing? It is consistent with the theories of
Wright (2016) in that it shows the importance of recognition (demonstrating
that the ad is processed into memory), while activation of the Ventral
Stratium is consistent with establishment of a somatic marker that yields an
ongoing mild emotional attachment to the advertised brand.

However, it does not show any evidence of a variety of other possible

mechanisms (including rational elaboration of content). This may simply be
a matter of detectability. As Wright (2016) noted, sales effects are often a
poor measure of advertising effectiveness in practice as they contain so
much natural variability and do not take account of the effects of stimuli
encountered just before purchase.

As with all such studies, the work of Venkatraman et al. (2015) is limited by
sampling and by the context of their research. Venkatraman’s method also
analyses all 30 seconds of a TV ad together, whereas methods such as
biometrics and EEG can track the reactions to individual scenes at
particular points within an ad. The more detailed tracking of viewer
reactions provided by biometrics and EEG offers opportunities to assist
copy development, as advertisers may be able to use these methods to
choose which scenes to insert at particular points within an ad.

So, as work in this area continues, other results may well emerge.
Exercise 13.5 Personal reflection on
advertising effects
When you next sit with a housemate and see a TV advertisement for a bank, ask that person (i)
if they recognize the advertisement (i.e. whether they remember seeing it before), (ii) whether
they like it, and (iii) whether it makes them more likely to use that bank. On the following day,
reflect on the advertisement yourself and think about whether the advertised bank easily comes
to mind, and whether you have a mild emotional attachment to that bank, or to any bank.
Summary
Commercial advertising targets behaviours that affect profit, such as
purchase and rental. Social advertising may target behaviours such as
smoking and dangerous driving. The benefit from commercial advertising
occurs through more sales, greater margins or lower costs; however, it will
only work if it is processed into memory in some way, either consciously or
subconsciously. Ad campaigns usually have quite small effects on sales that
are hard to detect in the short term; in large markets, the payback on
advertising can nonetheless be many times the expenditure.

We now understand better how individuals react to different frequencies

and concentrations of advertising. Extra exposures tend to give increasing
sales effects when concentrated into a short interval; otherwise, extra ads
have a diminishing effect. This is probably because concentrated exposures
give better breakthrough.

We present a model in which ads may have two sorts of primary effect: one
is thoughtful, and the effect is quite strong if it occurs; the other is
automatic and weak. Thoughtful processes occur early in the sequence of ad
exposures if they occur at all; automatic mechanisms can occur at each
exposure so ads can continue to have some effect for a long period. Wright
(2016) proposes a third mechanism – mild emotional attachment; this
differs from other automatic mechanisms as it arises from a different brain
process. The result is to establish a somatic marker that overlays other
memories, rather than simply adding further weak associations between
existing memories.

When primary effects occur, there may be consequential secondary effects

that include framing (persisting change in thinking about the brand),
purchase reinforcement (later purchase is facilitated by the occurrence of
the earlier purchase), better distribution, word of mouth and copying by
others.

Corresponding to these primary and secondary effects, sales appear to have

short-term and long-term decay rates. In one study, the short-term
component had an average half-life of 16 days. The long-term component
can persist for a year or more after the advertising has finished and can
double the effect of advertising compared to the short-term effect alone.

While low-loyalty and light buyers show less individual volume response to
advertising, the large number of these buyers ensures that these segments
provide more sales gain. Advertising is effective at reaching light and non-
buyers and contrasts with direct marketing, which is better when there is a
need to focus on a particular segment of existing buyers. In their reaction to
television advertising, light viewers are more responsive than heavy
viewers.

While the evolution of new media channels presents superficial challenges

to traditional advertising methods, research shows that TV advertising
continues to be effective and that the increasing use of mobile devices has
not reduced the use of desktop computers. Internet advertising generates the
kinds of offline sales lifts that have been traditionally expected from other
media. However, the growth of recommendation on the Internet may create
greater changes over time.
Additional Resources
You should read at least one of the cases in the Advertising Works series (or
the equivalent in another country). These cases indicate the concerns of ad
agencies and the difficulties of measuring advertising effects. If you are
more interested in the decisions that have to be made in ad agencies, one
textbook that is based on advertising research is Rossiter and Bellman
(2005). If you are interested in Armstrong’s rules, you can read his book or
visit his site www.advertisingprinciples.com, where the rules can be
downloaded free of charge.
Notes
1 The Advertising Federation of Australia (AFA) supports the publication
Effective Advertising, and in the UK the Institute of Practitioners in
Advertising (IPA) supports Advertising Works. These publications are
dedicated to showing how competent and creative advertising can bring
returns to the advertiser.

2 There is more agreement between researchers and practitioners on ad

frequency when the advertising or product is new or complex. It is widely
held that a linear or S-shaped response may be expected under these
conditions. Roberts (1996) found that the response was effectively linear in
five out of seven cases where the brand was new or relaunched.
References

Aaker, D.A. (1991) Managing Brand Equity: Capitalizing on the Value of a

Brand Name, New York: The Free Press.

Aaker, D.A. and Keller, K.L. (1990) Consumer evaluations of brand

extensions, Journal of Marketing, 54(1), 27–41.

Aaker, J.L. and Williams, P. (1998) Empathy versus pride: the influence of
emotional appeals across cultures, Journal of Consumer Research, 25(3),
241–61.

Abraham, M.M. and Lodish, L.M. (1990) Getting the most out of
advertising and promotion, Harvard Business Review, 68(3), 50–60.

Adams, H.F. (1916) Advertising and its Mental Laws, New York:
Macmillan.

AGB (1992) Revealed: the nation’s shopping habits, SuperMarketing, 14

August.

Ahluwalia, R. (2002) How prevalent is the negativity effect in consumer

environments Journal of Consumer Research, 29 (Sept.), 270–9.

Ahmad, R. (2002) The older or ageing customer in the UK: are they really
different? International Journal of Market Research, 44(3), 337–60.

Ailawadi, K.L. and Neslin, S.A. (1998) The effect of promotion on

consumption: buying more and consuming it faster, Journal of Marketing
 Research, 35(3), 390–8.

Ailawadi, K.L., Gedenk, K., Lutzky, C. and Neslin, S.A. (2007)

Decomposition of the sales impact of promotion-induced stockpiling,
Journal of Marketing Research, 44(3), 450–67.

Ailon, G. (2008) Mirror, mirror on the wall: culture’s consequences in a

value test of its own design, Academy of Management Review, 33(4),
885–904.

Ajzen, I. (1971) Attitude vs. normative messages: an investigation of the

differential effects of persuasive communications on behavior,
Sociometry, 34(2), 263–80.

Ajzen, I. (1985) From intentions to actions: a theory of planned behavior. In

J. Kuhl and J. Beckmann (eds), Action-Control: From Cognition to
Behavior, Heidelberg: Springer, 22–39.

Ajzen, I. (1991) The theory of planned behavior. In E.A. Locke (ed.),

Organizational Behavior and Human Decision Processes, 50, 179–211.

Ajzen, I. (2002) Perceived behavioral control, self-efficacy, locus of control

and the theory of planned behavior, Journal of Applied Social
Psychology, 32(4), 665–83.

Ajzen, I. and Driver, B.L. (1992) Application of the theory of planned

behavior to leisure choice, Journal of Leisure Research, 24(3), 207–24.

Ajzen, I. and Fishbein, M. (1969) The prediction of behavioral intentions in

a choice situation, Journal of Experimental Social Psychology, 5(4), 400–
16.
Ajzen, I. and Fishbein, M. (1972) Attitudinal and normative beliefs as
factors influencing behavioral intentions, Journal of Personality and
Social Psychology, 21(1), 1–9.

Ajzen, I. and Fishbein, M. (1977) Attitude–behavior relations: a theoretical

analysis and review of empirical research, Psychological Bulletin, 84,
888–918.

Ajzen, I. and Fishbein, M. (1980) Understanding Attitudes and Predicting

Social Behavior, Englewood Cliffs, NJ: Prentice-Hall.

Aksoy, L., Cooil, B., Groening, C., Keiningham, T. and Yalçın, A. (2008)
The long-term stock market valuation of customer satisfaction, Journal of
Marketing, 72(4), 105–22.

Alexander, V., Tripp, S. and Zak, P.J. (2015) Preliminary evidence for the
neurophysiologic effects of online coupons: changes in oxytocin, stress,
and mood, Psychology & Marketing, 32(9), 977–86.

Ali, M. (1977) Probability and utility estimates for racetrack bettors,

Journal of Political Economy, 85, 803–15.

Allais, M. (1953) Le comportement de l’homme rationel devant le risque:

critique des postulats et axiomes de l’ecole americaine, Econometrica,
21(4), 503–46.

Allenby, G.M. and Ginter, J.L. (1995) The effects of in-store displays and
feature advertising on consideration sets, International Journal of
Research in Marketing, 12(1), 67–80.
Allport, G.W. (1935) Attitudes. In C. Murchison (ed.), A Handbook of
Social Psychology, Worcester, MA: Clark University Press, 798–844.

Allport, G.W. and Postman, L. (1947) The Psychology of Rumor, New

York: Holt, Rinehart and Winston.

Ambler, T. and Broadbent, S. (2000) A dialogue on advertising

effectiveness and efficiency, Admap, July/August, 29–31.

Anderson, E.I. and Simester, D. (2003) Effects of $9 price endings on retail

sales: evidence from field experiments, Qualitative Marketing and
Economics, 1(1), 93–110.

Anderson, E.W. (1998) Customer satisfaction and word of mouth, Journal

of Service Research, 1(1), 5–17.

Anderson, E.W. and Mansi, S.A. (2009) Does customer satisfaction matter
to investors? Findings from the bond market, Journal of Marketing
Research, 46(5), 703–14.

Anderson, E.W. and Mittal, V. (2000) Strengthening the satisfaction–profit

chain, Journal of Service Research, 3(2), 107–20.

Anderson, E.W., Fornell, C. and Lehmann, D.R. (1994) Customer

satisfaction, market share and profitability, Journal of Marketing, 58(3),
53–66.

Anderson, E.W., Fornell, C. and Mazvancheryl, S.K. (2004) Customer

satisfaction and shareholder value, Journal of Marketing, 68(4), 172–85.
Anderson, K. and Sharp, B. (2010) Do growing brands win younger
consumers? International Journal of Market Research, 52(4), 433–41.

Anderson, N.H. (1965) Averaging versus adding as a stimulus-combination

rule in impression formation, Journal of Experimental Psychology, 70,
394–400.

Andreasen, A.R. (1985) Consumer responses to dissatisfaction in loose

monopolies, Journal of Consumer Research, 12(2), 135–41.

Andreasen, A.R. (1988) Consumer complaints and redress: what we know

and what we don’t know. In E.S. Maynes (ed.), The Frontier of Research
in Consumer Interest, Columbia: University of Columbia and American
Council of Consumer Interest, 675–721.

Andreasen, A.R. and Manning, J. (1990) The dissatisfaction and

complaining behavior of vulnerable consumers, Journal of Consumer
Satisfaction, Dissatisfaction and Complaining Behavior, 3, 12–20.

Anglin, L.K., Stuenkel, J.K. and Lepisto, L.R. (1994) The effect of stress on
price sensitivity and comparison shopping. In C.T. Allen and D.R. John
(eds), Advances in Consumer Research, 21, 126–31.

Araujo, T., Neijens, P. and Vliegenthart, R. (2015) What motivates

consumers to re-tweet brand content? The impact of information,
emotion and traceability on pass-along behaviour, Journal of Advertising
Research, 55(3), 284–95.

Arce, M. and Cebollada, J. (2006) The role of loyalty in online and offline
shopping behaviour: an empirical application to the grocery industry,
Conference Proceedings of the European Marketing Academy, Athens.
Areni, C.S. and Kim, D. (1993) The influence of background music on
shopping behavior: classical versus top-forty music in a wine store. In L.
McAlister and M.L. Rothschild (eds), Advances in Consumer Research,
20, 336–40.

Armitage, C.J. and Conner, M. (2001) Efficacy of the theory of planned

behaviour: a meta-analytic review, British Journal of Social Psychology,
40(4), 471–99.

Armstrong, J.S. (1985) Long-range Forecasting: From Crystal Ball to

Computer, 2nd edition, London: Wiley.

Armstrong, J.S. (2010) Persuasive Advertising: Evidence-Based Principles,

Basingstoke: Palgrave Macmillan.

Armstrong, J.S., Du, R., Green, J. and Graefe, A. (2016) Predictive validity
of evidence-based principles: an application of the index method,
European Journal of Marketing, 50.

Arndt, J. (1967) The role of product-related conversations in the diffusion

of a new product, Journal of Marketing Research, 4 (Aug.), 291–5.

Arthur, C. (1992) Fifteen million Americans are shopping addicts,

American Demographics, March, 14–15.

Asugman, G. (1998) An evaluation of negative word-of-mouth research for

extensions, European Advances in Consumer Research, 3, 70–5.

Ataman, M.B., Mela, C.F. and Van Heerde, H.J. (2008) Building brands,
Marketing Science, 27(6), 1036–54.
Ataman, M.B., Van Heerde, H.J. and Mela, C.F. (2010) The long-term
effect of marketing strategy on brand sales, Journal of Marketing
Research, 47(5), 866–82.

Babacus, E. and Boller, G.W. (1992) An empirical assessment of the

SERVQUAL scale, Journal of Business Research, 24(3), 253–68.

Babin, B.J., Hardesty, D.M. and Suter, T.A. (2003) Color and shopping
intentions: the intervening effect of price fairness and perceived affect,
Journal of Business Research, 56(7), 541–51.

Bagozzi, R.P. (1981) Attitudes, intentions and behavior: a test of some key
hypotheses, Journal of Personality and Social Psychology, 41(4), 607–27.

Bagozzi, R.P. (1984) Expectancy–value attitude models: an analysis of

critical measurement issues, International Journal of Research in
Marketing, 1(4), 295–310.

Bagozzi, R.P. (1992) The self-regulation of attitudes, intentions and

behaviors, Social Psychology Quarterly, 55, 178–204.

Bagozzi, R.P. and Kimmel, S.K. (1995) A comparison of leading theories

for the prediction of goal-directed behaviours, British Journal of Social
Psychology, 34(4), 437–61.

Baker, C. (1993) Advertising Works 7, Henley-on-Thames: Institute of

Practitioners in Advertising, NTC Publications.

Baker, C. (1995) Advertising Works 8, Henley-on-Thames: Institute of

Practitioners in Advertising, NTC Publications.
Baker, J. and Wakefield, K. (2012) How consumer shopping orientation
influences perceived crowding, excitement, and stress at the mall, Journal
of the Academy of Marketing Science, 40(6), 791–806.

Baldinger, A.L. and Rubinson, J. (1996) Brand loyalty: the link between
attitude and behavior, Journal of Advertising Research, 36(6), 22–34.

Baldwin, M.W. and Holmes, J.G. (1987) Salient private audiences and
awareness of the self, Journal of Personality and Social Psychology,
52(6), 1087–98.

Balter, D. and Butman, J. (2005) Grapevine: The New Art of Word-of-

Mouth Marketing, New York: Penguin Portfolio.

Bansal, H.S. and Voyer, P.A. (2000) Word-of-mouth processes within a

services purchase decision context, Journal of Service Research, 2(3),
166–77.

Bargh, J.A., Chaiken, S., Govender, R. and Pratto, F. (1992) The generality
of the automatic activation effect, Journal of Personality and Social
Psychology, 62(6), 893–912.

Bartlett, F.C. (1932) Remembering, Cambridge: Cambridge University

Press.

Bass, F.M. (1969) A new product growth model for consumer durables,
Management Science, 15(5), 215–27.

Bass, F.M. (1995) Empirical generalizations and marketing science: a

personal view, Marketing Science, 14(3), Part 2 of 2, G6–18.
Bass, F.M., Givon, M.M., Kalwani, M.U., Reibstein, D. and Wright, G.P.
(1984) An investigation into the order of the brand choice process,
Marketing Science, 3(4), 267–87.

Bass, F.M., Jeuland, A.P. and Wright, G.P. (1976) Equilibrium stochastic
choice and market penetration theories: derivation and comparisons,
Management Science, 22(10), 1051–63.

Batra, R., Lenk, P. and Wedel, M. (2010) Brand extensions strategy

planning: empirical estimation of brand-category personality, Journal of
Marketing Research, 47(2), 335–47.

Bayus, B.L. (1985) Word of mouth: the indirect effects of marketing efforts,
Journal of Advertising Research, 25(3), 31–9.

Bayus, B.L. (1991) The consumer durable replacement buyer, Journal of

Marketing, 55 (January), 42–51.

Beales, H.J., Mazis, M.B., Salop, S.C. and Staelin, R. (1981) Consumer
search and public policy, Journal of Consumer Research, 8(1), 11–22.

Beatty, S.E. and Smith, S.M. (1987) External search effort: an investigation
across several product categories, Journal of Consumer Research, 14(1),
83–95.

Bellizzi, J.A., Crowley, A.E. and Hasty, R.E. (1983) The effects of color in
store design, Journal of Retailing, 59(1), 21–44.

Bemmaor, A.L. (1995) Predicting behavior from intention-to-buy measures:

the parametric case, Journal of Marketing Research, 32(2), 176–91.
Benartzi, S. and Thaler, R.H. (1995) Myopic loss aversion and the equity
premium puzzle, Quarterly Journal of Economics, 110(1), 73–92.
Reprinted in Kahneman, D. and Tversky, A. (2000) Choices, Values, and
Frames, New York: Russell Sage Foundation/Cambridge University
Press, 301–16.

Benterud, T. and Stø, E. (1993) TV shopping in Scandinavia: consumer

satisfaction, dissatisfaction and complaining behavior, Journal of
Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 6,
196–203.

Bentler, P.M. and Speckart, G. (1979) Models of attitude–behavior

relations, Psychological Review, 86(5), 452–64.

Bentler, P.M. and Speckart, G. (1981) Attitudes ‘cause’ behaviors: a

structural equation analysis, Journal of Personality and Social
Psychology, 40, 226–38.

Berger, J. and Iyengar, R. (2012) How interest shapes word of mouth over
different channels, available at
http://rady.ucsd.edu/faculty/seminars/papers/berger.pdf

Berger, J. and Schwartz, E.M. (2011) What drives immediate and ongoing
word of mouth? Journal of Marketing Research, 48(Oct.), 869–80.

Berlyne, D.E. (1965) Structure and Direction in Thinking, London: Wiley.

Berlyne, D.E. and McDonnell, P. (1965) Effects of stimulus complexity and

incongruity on duration of EEG desynchronisation,
Electroencephalography and Clinical Neurophysiology, 18(2), 156–61.
Bernoulli, D. (1738) Specimen Theoriae Novae de Mensura Sortis,
Comentarii Academiae Scientiarum Imperiales Petropolitanae, 5, 175–
92. Translated by L. Sommer in Econometrica, 1954, 22(1), 23–36.

Berry, L.L. (1983) Relationship marketing. In L.L. Berry, G.L. Shostack,

and G.D. Upah (eds), Emerging Perspectives on Service Marketing,
Chicago: American Marketing Association, 25–8.

Berthon, P., Pitt, L., Parent, M. and Berthon, J.-P. (2009) Aesthetics and
ephemerality: observing andpreserving the luxury brand, California
Management Review, 52, 45–66.

Bettman, J.R. (1979) An Information Processing Theory of Consumer

Choice, Reading, MA: Addison-Wesley.

Biel, A.L. (1991) The brandscape: converting brand image into equity,
Admap, October, 41–6.

Bijmolt, T.H.A., van Heerde, H.J. and Pieters, R.G.M. (2005) New
empirical generalizations on the determinants of price elasticity, Journal
of Marketing Research, 42(2), 141–56.

Binet, L. and Field, P. (2009) Empirical generalizations about advertising

campaign success, Journal of Advertising Research, 49(2), 130–3.

Bird, M. and Ehrenburg, A.S.C. (1966) Intentions-to-buy and claimed brand

usage, Operational Research Quarterly, 17, 27–46.

Bird, M., Channon, C. and Ehrenberg, A.S.C. (1970) Brand image and
brand usage, Journal of Marketing Research, 7, 307–14.
Bitner, M.J., Booms, B.H. and Tetreault, M.S. (1990) The service
encounter: diagnosing favourable and unfavourable incidents, Journal of
Marketing, 54(1), 71–84.

Black, D.W. (2007) A review of compulsive buying disorder, World

Psychiatry, 6(1), 14–18.

Blackett, T., Boad, B., Cowper, P. and Kumar, S. (1999) The future of co-
branding. In Co-Branding, pp. 113–23. London: Palgrave Macmillan.

Blair, M.H. and Rabuck, M.J. (1998) Advertising wearin and wearout: ten
years later – more empirical evidence and successful practice, Journal of
Advertising Research, 38(5), 7–18.

Blattberg, R.C., Briesch, R. and Fox, E.J. (1995) How promotions work,
Marketing Science, 14(3), G122–32.

Bogomolova, S. and Romaniuk, J. (2009) Brand defection in a business-to-

business financial service, Journal of Business Research, 62(3), 291–6.

Bolton, L.E., Warlop, L. and Alba, J.W. (2003) Consumer perceptions of

price (un)fairness, Journal of Consumer Research, 29(4), 474–91.

Bolton, R.N. (1989) The relationship between market characteristics and

promotional price elasticities, Marketing Science, 8(2), 153–69.

Bolton, R.N. (1998) A dynamic model of the duration of the customer’s

relationship with a continuous service provider: the role of satisfaction,
Marketing Science, 17(1), 45–65.
Bornemann, T. and Homburg, C. (2011) Psychological distance and the dual
role of price, Journal of Consumer Research, 38(3), 490–504.

Bosmans, A. (2007) Scents and sensibility: when do (in)congruent ambient

scents influence product evaluations? Journal of Marketing, 71(3), 32–
43.

Bottomley, P.A. and Holden, S.J.S. (2001) Do we really know how

consumers evaluate brand extensions? Empirical generalizations based
on secondary analysis of eight studies, Journal of Marketing Research,
38(4), 494–500.

Briesch, R.A., Krishnamurthy, L., Mazumdar, T. and Raj, S.P. (1997) A

comparative analysis of reference price models, Journal of Consumer
Research, 24 (Sept.), 202–14.

Brislin, R.W. (1970) Back-translation for cross-cultural research, Journal of

Cross-Cultural Psychology, 1(3), 185–216.

Broadbent, S. (1984) Modelling with Adstock, Journal of the Market

Research Society, 26(4), 295–312.

Broadbent, S. (1989) The Advertising Budget: The Advertiser’s Guide to

Budget Determination, Henley-on-Thames: NTC Publications.

Broadbent, S. (1998) Effective frequency: there and back, Admap, May,

34–8.

Broadbent, S. and Fry, T. (1995) Adstock modelling for the long term,
Journal of the Market Research Society, 37(4), 385–403.
Broadbent, T. (2000) Advertising Works 11, Henley-on-Thames: Institute of
Practitioners in Advertising, World Advertising Research Centre
(WARC).

Brown, G. (1986) Monitoring advertising performance, Admap, 22(3), 151–

3.

Brown, G.H. (1953) Brand loyalty, Advertising Age, 24, 28–35.

Reproduced in A.S.C. Ehrenberg and F.G. Pyatt (eds), Consumer
Behaviour, Harmondsworth: Penguin Books, 28–35.

Brown, J.J. and Reingen, P.H. (1987) Social ties and word-of-mouth referral
behavior, Journal of Consumer Research, 14(3), 350–62.

Brown, J.R. and Dant, R.P. (2009) The theoretical domains of retailing
research: a retrospective, Journal of Retailing, 85(2), 113–28.

Buday, T. (1989) Capitalizing on brand extensions: lessons of success and

failure, Journal of Consumer Marketing, 6(4), 27–30.

Budd, R. (1986) Predicting cigarette use: the need to incorporate measures

of salience in the theory of reasoned action, Journal of Applied Social
Psychology, 16, 663–85.

Burke, R.R. and Srull, T.K. (1988) Competitive interference and consumer
memory for advertising, Journal of Consumer Research, 15(1), 55–68.

Burnkrant, R.E. and Unnava, H.R. (1987) Effects of variation in message

execution on the learning of repeated brand information. In M.
Wallendorf and P. Anderson (eds), Advances in Consumer Research, 14,
173–6.
Burns, D.J. and Perkins, D. (1996) Accounts in post-purchase behavior:
excuses, justifications and meta-accounts, Journal of Satisfaction,
Dissatisfaction and Complaining Behavior, 9, 144–57.

Buttle, F. (1996) SERVQUAL: review, critique, research agenda, European

Journal of Marketing, 30(1), 8–32.

Buzzell, R.D. and Gale, B.T. (1987) The PIMS Principles: Linking Strategy
to Performance, New York: The Free Press.

Buzzell, R.D., Quelch, J.A. and Salmon, W.J. (1991) The costly bargain of
trade promotion, Harvard Business Review, 68(2), 141–9.

Camerer, C.F. (2000) Prospect theory in the wild. In D. Kahneman and A.

Tversky (eds), Choices, Values and Frames, New York: Russell Sage
Foundation, 288–300.

Campbell, M.C. (1999) Perceptions of price unfairness: antecedents and

consequences, Journal of Marketing Research, 36(2), 187–99.

Caplovitz, D. (1967) The Poor Pay More, 2nd edition, New York: The Free
Press.

Cardozo, R.N. (1965) An experimental study of consumer effort,

expectation and satisfaction, Journal of Marketing Research, 2(3), 244–9.

Carl, W.J. (2008) The role of disclosure in organized word-of-mouth

marketing programs, Journal of Marketing Communications, 14(3), 225–
41.
Carman, J.M. (1970) Correlates of brand loyalty: some positive results,
Journal of Marketing Research, 7(1), 67–76.

Carman, J.M. (1990) Consumer perceptions of service quality: an

assessment of the SERVQUAL dimensions, Journal of Retailing, 66(1),
33–55.

Castensen, L.L., Isaakowitz, D.M. and Charles, S.T. (1999) Taking time
seriously: a theory of socioemotional selectivity, American Psychologist,
54(March), 165–81.

Chan, H., Wan, L.C. and Sin, L.M. (2009) The contrasting effects of culture
on consumer tolerance: interpersonal face and impersonal fate, Journal of
Consumer Research, 36(2), 292–304.

Chan, T., Narasimhan, C. and Zhang, Q. (2008) Decomposing promotional

effects with a dynamic structural model of flexible consumption, Journal
of Marketing Research, 45(4), 487–98.

Chandon, P., Morwitz, V.G. and Reinartz, W.J. (2005) Do intentions really
predict behavior? Self-generated validity effects in survey research,
Journal of Marketing, 69(2), 1–14.

Chandon, P., Wansink, B. and Laurent, G. (2000) A benefit congruency

framework of sales promotion effectiveness, Journal of Marketing, 64(4),
65–81.

Chang-Hyeon, J., Arentze,T. and Timmermans, H. (2006) Characterisation

and comparison of gender-specific utility functions of shopping duration
episodes, Journal of Retailing and Consumer Services, 13(4), 249–59.
Channon, C. (1985) Advertising Works 3, London: Holt, Rinehart and
Winston.

Charlton, P. (1973) A review of shop loyalty, Journal of the Market

Research Society, 15(1), 35–51.

Chatfield, C. and Goodhardt, G. (1975) Results concerning brand choice,

Journal of Marketing Research, 12(1), 110–13.

Chawdhary, R. and Dall’Olmo Riley, F. (2015) Investigating the

consequences of word of mouth from a WOM sender’s perspective in the
services context, Journal of Marketing Management, 31(9–10), 1018–39.

Cheng, Y. and Tsai, Y. (2014) Train delay and perceived wait time:
passengers’ perspective, Transport Reviews, 34(6), 710–29.

Chevalier, J.A. and Mayzlin, D. (2003) The effect of word of mouth on

sales: online book reviews, Journal of Marketing Research, 44(3), 345–
54.

Chintagunta, P.K. (1993) Investigating purchase incidence, brand choice

and purchase quantity decisions of households, Marketing Science, 12(2),
184–208.

Christaller, W. (1933) Central Places in Southern Germany, translated by

C.W. Baskin (1966), Englewood Cliffs, NJ: Prentice-Hall.

Christiansen, T. and Tax, S.S. (2000) Measuring word of mouth: the

questions of who and when, Journal of Marketing Communications, 6,
185–99.
Churcher, P.B. and Lawton, J.H. (1987) Predation by domestic cats in an
English village, Journal of Zoology, 212, 439–55.

Churchill, G.A., Jr and Surprenant, C. (1982) An investigation into the

determinants of customer satisfaction, Journal of Marketing Research,
19(4), 491–504.

Churchill, H. (1942) How to measure brand loyalty, Advertising and

Selling, 35(24), 34–50.

Clare, J.E. and Kiser, C.V. (1951) Preference for children of a given sex in
relation to fertility. In P.K. Whelpton and C.V. Kiser (eds), Social and
Psychological Factors Affecting Fertility, New York: Milbank Memorial
Fund, 621–73.

Clarke, D.G. (1976) Econometric measurement of the duration of

advertising effect on sales, Journal of Marketing Research, 13(4), 345–
57.

Clemmer, E.C. and Schneider, B. (1989) Towards understanding and

controlling dissatisfaction with waiting during peak demand times. In
M.J. Bitner and L.A. Crosby (eds), Designing a Winning Service
Strategy, Chicago: American Marketing Association, 87–91.

Cole, C., Laurent, G., Drolet, A., Ebert, J., Gutchess, A., Lambert-
Pandraud, R., et al. (2008) Decision making and brand choice by older
consumers, Marketing Letters, 19(3/4), 355–65.

Cole, C.A. and Houston, M.J. (1987) Encoding and media effects on
consumer learning deficiencies in the elderly, Journal of Marketing
Research, 24(1), 55–63.
Coleman, J., Katz, E. and Menzel, H. (1957) The diffusion of an innovation
among physicians, Sociometry, 20(4), 253–70.

Collins, M. (1971) Market segmentation: the realities of buyer behaviour,

Journal of the Market Research Society, 13(3), 146–57.

Collins, M., Beal, V. and Barwise, P. (2003) Channel use among multi-
channel viewers: patterns in TV viewing, Report 15 for Corporate
Sponsors, Adelaide, Australia: Ehrenberg-Bass Institute for Marketing
Science.

Conner, M., Lawton, R., Parker, D., Chorlton, K., Manstead, A.S.R. and
Stradling, S. (2007) Application of the theory of planned behaviour to the
prediction of objectively assessed breaking of posted speed limits, British
Journal of Psychology, 98(3), 429–53.

Copeland, M.T. (1923) Relation of consumers’ buying habits to marketing

methods, Harvard Business Review, 1, 282–9.

Coulter, K.S. and Coulter, R.A. (2005) Size does matter: the effects of
magnitude representation congruency on price perceptions and purchase
likelihood, Journal of Consumer Psychology, 15(1), 64–76.

Coulter, K.S. and Coulter, R.A. (2013) Small sounds, big deals: phonetic
symbolism effects in pricing, Journal of Consumer Research, S224–37.

Cox, K.K. (1970) The effect of shelf space upon sales of branded products,
Journal of Marketing Research, 7(1), 55–8.

Crawford, C. and Di Benedetto, A. (2006) New Products Management, 8th

edition, New York: McGraw-Hill/Irwin.
Crawford, C.M. and Di Benedetto, C.A. (2014) New Products
Management,11th edition, New York: McGraw-Hill.

Crocker, J., Fiske, S.T. and Taylor, S.E. (1984) Schematic bases of belief
change. In J.R. Eiser (ed.), Attitudinal Judgement, New York: Springer-
Verlag.

Cronin, J.J., Jr and Taylor, S.A. (1992) Measuring service quality: a re-
examination and extension, Journal of Marketing, 56(3), 55–68.

Crosby, L.A. and Stephens, N. (1987) Effects of relationship marketing on

satisfaction, retention, and prices in the life insurance industry, Journal of
Marketing Research, 24(4), 404–11.

Cross, S.E. and Madson, L. (1997) Models of the self: self-construals and
gender, Psychological Bulletin, 122(1), 5–37.

Cunningham, R.M. (1956) Brand loyalty: what, where, how much? Harvard
Business Review, 34 (Jan./Feb.), 116–28.

Curhan, R.C. (1972) The relationship between shelf space and unit sales in
supermarkets, Journal of Marketing Research, 9(4), 406–12.

Customer Care Measurement & Consulting (CCMC) (2013) The 2013

Customer Rage Study: Will we ever learn? The sad state of customer care
in America. Available at: www.customercaremc.com/the-2013-customer-
rage-study/

D’Astous, A. (1990) An inquiry into the compulsive side of normal

consumers, Journal of Consumer Policy, 13(1), 15–32.
Dabholkar, P.A., Thorpe, D.I. and Rentz, J.O. (1996) A measure of service
quality for retail stores, Journal of the Academy of Marketing Science,
24(1), 3–16.

Dacin, P.A. and Smith, D.C. (1993) The effect of adding products to a brand
on consumers’ evaluations of new brand extensions. In L. McAlister and
M.L. Rothschild (eds), Advances in Consumer Research, 20, 594–8.

Dagnoli, J. (1987) Impulse governs shoppers, Advertising Age, 5 October,

p. 93.

Dall’Olmo Riley, F.D. (2010) Brand Management, London: Sage.

Dall’Olmo Riley, F.D., Ehrenberg, A.S.C., Castleberry, S.B., Barwise, T.P.

and Barnard, N.R. (1997) The variability of attitudinal repeat-rates,
International Journal of Research in Marketing, 14(5), 437–50.

Dall’Olmo Riley, F.D., Pina, J.M. and Bravo, R. (2015) The role of
perceived value in vertical brand extensions of luxury and premium
brands, Journal of Marketing Management, 31(7–8), 881–913.

Dall’Olmo Riley, F.D., Singh, J. and Blankson, C. (eds) (2016) The

Routledge Companion to Contemporary Brand Management, London:
Routledge.

Damasio, A.R. (1994) Descartes’ Error: Emotion, Reason, and the Human
Brain, New York: Avon.

Danaher, P.J., Bonfrer, A. and Dhar, S. (2008) The effect of competitive

advertising interference on sales for packaged goods, Journal of
Marketing Research, 45(2), 211–25.
Danaher, P.J., Wilson, I.W. and Davis, R.A. (2003) A comparison of online
and offline consumer brand loyalty, Marketing Science, 22(4), 461–76.

Davidson, A.R. and Jaccard, J.J. (1975) Population psychology: a new look
at an old problem, Journal of Personality and Social Psychology, 31,
1073–82.

Davis, F.D. (1989) Perceived usefulness, perceived ease of use, and user
acceptance of information technology, MIS Quarterly, 13(3), 319–39.

Dawar, N. and Anderson, P.F. (1993) Determining the Order and Direction
of Multiple Brand Extensions, INSEAD Working Paper 93/17/MKT.

Dawes, J., Mundt, K. and Sharp, B. (2009) Consideration sets for financial
services brands, Journal of Financial Services Marketing, 14, 190–202.

Day, D., Gan, B., Gendall, P. and Esslemont, D. (1991) Predicting purchase
behaviour, Marketing Bulletin, 2(May), 18–30.

Day, G.S. (1969) A two-dimensional concept of brand loyalty, Journal of

Advertising Research, 9, 29–35.

Day, R.L. (1984) Modeling choices among alternative responses to

dissatisfaction. In T.C. Kinnear, (ed.), Advances in Consumer Research,
11, 496–9.

Day, R.L. and Landon, E.L. (1976) Collecting comprehensive complaint

data by survey research. In B.B. Anderson (ed.), Advances in Consumer
Research, 3, 263–8.
De Langhe, B., Van Osselaer, S.J., Puntoni, S. and McGill, A.L. (2014)
Fooled by heteroscedastic randomness: local consistency breeds
extremity in price-based quality inferences, Journal of Consumer
Research, 41(4), 978–94.

De Matos, C.A., Henrique, J.L., Vargas, R. and Carlos, A. (2007) Service

recovery paradox: a meta-analysis, Journal of Service Research, 10(1),
60–77.

de Mooij, M. (2010) Global Marketing and Advertising: Understanding

Cultural Paradoxes, 3rd edition, London: Sage.

de Mooij, M. (2011) Consumer Behavior and Culture: Consequences for

Global Marketing and Advertising, 2nd edition, London: Sage.

de Mooij, M. and Hofstede, G. (2010) The Hofstede model: applications to

global branding and advertising strategy and research, International
Journal of Advertising, 29(1), 85–110.

Degeratu, A.M., Rangaswamy, A. and Wu, J. (2001) Consumer choice

behavior in online and traditional supermarkets: the effects of brand
name, price, and other search attributes, International Journal of Research
in Marketing, 17(1), 55–78.

Dekimpe, M.G., Hanssens, D.M. and Silva-Risso, J.M. (1999) Long-run

effects of price promotions in scanner markets, Journal of Econometrics,
89(1/2), 269–91.

DelVecchio, D., Henard, D.H. and Freling, T.H. (2006) The effect of sales
promotion on post-promotion brand preference: a meta-analysis, Journal
of Retailing, 82(3), 203–14.
DeSarbo, W.S., Huff, L., Rolandelli, M.M. and Choi, J. (1994) On the
measurement of perceived service quality. In R.T. Rust and R.L. Oliver
(eds), Service Quality: New Directions in Theory and Practice, London:
Sage, 201–22.

Deutsch, M. and Gerard, H.B. (1955) A study of the normative and

informational influences upon individual judgment, Journal of Abnormal
and Social Psychology, 51(3), 629–36.

DeWulf, K., Odekerken-Schroder, G. and Iacobucci, D. (2001) Investments

in consumer relationships: a cross-country and cross-industry
exploration, Journal of Marketing, 65 (Oct.), 33–50.

Dichter, E. (1966) How word-of-mouth advertising works, Harvard

Business Review, 44(6), 147–66.

Dick, A.S. and Basu, K. (1994) Customer loyalty: towards an integrated

framework, Journal of the Academy of Marketing Science, 22(2), 99–
113.

Dickson, P.R. and Sawyer, A.G. (1990) The price knowledge and search of
supermarket shoppers, Journal of Marketing, 54(3), 42–54.

Dimson, E., Marsh, P. and Staunton, M. (2004) Low-cap and low-rated

companies, Journal of Portfolio Management, Summer, 1–12.

Dinner, I.M., Van Heerde, H.J. and Neslin, S.A. (2014) Driving online and
offline sales: the cross-channel effects of traditional, online display, and
paid search advertising, Journal of Marketing Research, 51(5), 527–45.
Dittmar, H. (2005) Compulsive buying: a growing concern? An
examination of gender, age, and endorsement of materialistic values as
predictors, British Journal of Psychology, 96(4), 467–91.

Dittmar, H., Beattie, J. and Friese, S. (1995) Gender identity and material
symbols: objects and decision considerations in impulse purchases,
Journal of Economic Psychology, 16(3), 491–512.

Donovan, R.J. and Rossiter, J.R. (1982) Store atmosphere: an

environmental psychology approach, Journal of Retailing, 58(1), 34–56.

Donovan, R.J., Rossiter, J.R., Marcoolyn, G. and Nesdale, A. (1994) Store

atmosphere and purchasing behaviour, Journal of Retailing, 70(3), 283–
94.

Dowling, G.R. and Uncles, M.D. (1997) Do customer loyalty programs

really work? Sloan Management Review, 38(Summer), 71–82.

Doyle, P. (1989) Building successful brands: the strategic options, Journal

of Marketing Management, 5(1), 77–95.

Drèze, X. and Nunes J.C. (2009) Feeling superior: the impact of loyalty
program structure on consumers’ perception of status, Journal of
Consumer Research, 35(6), 890–905.

Drèze, X. and Nunes, J. (2011) Recurring goals and learning: the impact of
successful reward attainment on purchase behavior, Journal of Marketing
Research, 48(2), 268–81.

Drèze, X., Hoch, S.J. and Purk, M.E. (1994) Shelf management and space
elasticity, Journal of Retailing, 70(4), 301–26.
Driesener, C. and Romaniuk, J. (2006) Comparing methods of brand image
measurement, International Journal of Market Research, 48(6), 681–98.

Drolet, A., Schwarz, N. and Yoon, C. (eds) (2010) The Aging Consumer:
Perspectives from Psychology and Economics, New York: Routledge,
Taylor and Francis Group, Ch. 4.

du Plessis, E. (2009) Digital video recorders and inadvertent advertising

exposure, Journal of Advertising Research, 49(2), 236–9.

Du, R.Y. and Kamakura, W.A. (2011) Measuring contagion in the diffusion
of consumer packaged goods, Journal of Marketing Research, 68(Feb.),
28–47.

Dubé-Rioux, L., Schmitt, B.H. and Leclerc, F. (1989) Consumers’ reactions

to waiting: when delays affect the perception of service quality. In T.K.
Srull (ed.), Advances in Consumer Research, 16, 59–63.

Dunn, R.S. and Wrigley, N. (1984) Store loyalty for grocery products: an
empirical study, Area, 16(4), 307–14.

Dunn, R.S., Reader, S. and Wrigley, N. (1983) An investigation of the

assumptions of the NBD model as applied to purchasing at individual
stores, Applied Statistics, 32(3), 249–59.

Eagly, A.H. and Chaiken, S. (1993) The Psychology of Attitudes, Orlando,

FL: Harcourt, Brace, Jovanovich.

Eagly, A.H. and Chaiken, S. (2005) Attitude research in the 21st century:
the current state of knowledge. In D. Albarracín, B.T. Johnson and M.P.
 Zanna (eds), The Handbook of Attitudes, Mahwah, NJ: Erlbaum, 743–
67.

East, R. (1973) The duration of attention to alternatives and re-evaluation in

choices with two and three alternatives, European Journal of Social
Psychology, 3(2), 125–44.

East, R. (1992) The effects of experience on the decision making of expert

and novice buyers, Journal of Marketing Management, 8(2), 167–76.

East, R. (1993) Investment decisions and the theory of planned behaviour,

Journal of Economic Psychology, 14(2), 337–75.

East, R. (2000) Complaining as planned behavior, Psychology and

Marketing, 17(12), 1077–95.

East, R. (2016) Bias in the evaluation of research methods, Marketing

Theory, 16(2), 219–31.

East, R. and Hammond, K. (1996) The erosion of repeat-purchase loyalty,

Marketing Letters, 7(2), 163–72.

East, R. and Hogg, A. (1997) The anatomy of conquest: Tesco versus

Sainsbury, Journal of Brand Management, 5(1), 53–61.

East, R., Eftichiadou, V. and Williamson, M. (2003) Point-of-purchase

display and brand sales, International Review of Retail, Distribution and
Consumer Research, 13(1), 127–34.
East, R., Gendall, P., Hammond, K. and Lomax, W. (2005a) Consumer
loyalty: singular, additive or interactive? Australasian Marketing Journal,
13(2), 10–26.

East, R., Grandcolas, U., Dall’Olmo Riley, F.D. and Lomax, W. (2012)
Reasons for switching service providers, Australasian Marketing Journal,
20(2), 164–70.

East, R., Hammond, K. and Gendall, P. (2006) Fact and fallacy in retention
marketing, Journal of Marketing Management, 22(1–2), 5–23.

East, R., Hammond, K., Harris, P. and Lomax, W. (2000) First-store loyalty
and retention, Journal of Marketing Management, 16(4), 307–25.

East, R., Hammond, K. and Lomax, W. (2008) Measuring the impact of

positive and negative word of mouth on brand purchase probability,
International Journal of Research in Marketing, 25(3), 215–24.

East, R., Hammond, K., Lomax, W. and Robinson, H. (2005b) What is the
effect of a recommendation? Marketing Review, 5(2), 145–57.

East, R., Hammond, K.A. and Wright, M. (2007) The relative incidence of
positive and negative word of mouth: a multi-category study,
International Journal of Research in Marketing, 24(2), 175–84.

East, R., Harris, P., Willson, G. and Hammond, K. (1995) Correlates of

first-brand loyalty, Journal of Marketing Management, 11(5), 487–97.

East, R., Harris, P., Willson, G. and Lomax, W. (1995) Loyalty to

supermarkets, International Review of Retail, Distribution and Consumer
Research, 5(1), 99–109.
East, R., Lomax, W. and Narain, R. (2001) Customer tenure,
recommendation and switching, Journal of Consumer Satisfaction,
Dissatisfaction and Complaining Behavior, 14, 46–54.

East, R., Lomax, W. and Willson, G. (1991) Factors associated with service
delay in supermarkets and post offices, Journal of Consumer Satisfaction,
Dissatisfaction and Complaining Behaviour, 4, 123–8.

East, R., Lomax, W., Willson, G. and Harris, P. (1992) Demand Over Time:
Attitudes, Knowledge and Habits that Affect When Customers Use
Banks and Building Societies, Working Paper, Kingston Business School.

East, R., Lomax, W., Willson, G. and Harris, P. (1994) Decision making and
habit in shopping times, European Journal of Marketing, 28(4), 56–71.

East, R., Romaniuk, J. and Lomax, W. (2011) The NPS and the ACSI: a
critique and an alternative metric, International Journal of Market
Research, 53(3), 327–46.

East, R., Uncles, M. and Lomax, W. (2014a) Hear nothing, do nothing: the
role of word of mouth in the decision making of older consumers, Journal
of Marketing Management, 30(7/8), 786–801.

East, R., Uncles, M., Romaniuk, J. and Dall’Olmo Riley, F. (2015) Factors
associated with the production of word of mouth, International Journal of
Marketing Research, 57(3), 439–58.

East, R., Uncles, M., Romaniuk, J. and Hand, C. (2013) Distortions in

retrospective word-of-mouth measurement, International Journal of
Market Research, 55(4), 477–86.
East, R., Uncles, M., Romaniuk, J. and Hand, C. (2014b) The decay of
positive and negative word of mouth after product experience.
Australasian Marketing Journal, 22(4), 350–5.

East, R., Uncles, M., Romaniuk, J. and Lomax, W. (2016) Measuring the
impact of positive and negative word of mouth: a reappraisal,
Australasian Marketing Journal, 24(1), 54–58.

Edwards, W. (1954) The theory of decision making, Psychological Bulletin,

51(4), 380–417.

Effective Advertising 6 (2001) Sydney: Advertising Federation of Australia

Advertising Effectiveness Awards.

Effective Advertising 8 (2006) Sydney: Advertising Federation of Australia

Advertising Effectiveness Awards.

Ehrenberg, A.S.C. (1959) The pattern of consumer purchases, Applied

Statistics, 8, 26–41.

Ehrenberg, A.S.C. (1969) The discovery and use of laws of marketing,

Journal of Advertising Research, 9(2), 11–17.

Ehrenberg, A.S.C. (1986) Pricing and brand differentiation, Singapore

Marketing Review, 1, 5–15.

Ehrenberg, A.S.C. (1988) Repeat Buying: Theory and Applications, 2nd

edition, London: Charles Griffin & Co. (first published in 1972 by North
Holland).
Ehrenberg, A.S.C. (1993) If you’re so strong, why aren’t you bigger?
Admap, 28(1), 13–14.

Ehrenberg, A.S.C. and England, L.R. (1990) Generalising a pricing effect,

Journal of Industrial Economics, 39(1), 47–68.

Ehrenberg, A.S.C and Goodhardt, G.J. (1979) Essays on Understanding

Buyer Behavior, New York: J. Walter Thompson Co. and Market
Research Corporation of America.

Ehrenberg, A.S.C. and Goodhardt, G.J. (2001) New brands: near-instant

loyalty, Journal of Targeting, Measurement and Analysis in Marketing,
10(1), 9–16.

Ehrenberg, A.S.C., Goodhardt, G.J. and Barwise, T.P. (1990) Double

jeopardy revisited, Journal of Marketing, 54(3), 82–90.

Ehrenberg, A.S.C., Hammond, K.A. and Goodhardt, G.J. (1994a) The after-
effects of price-related consumer promotions, Journal of Advertising
Research, 34(4), 11–21.

Ehrenberg, A.S.C., Uncles, M.D. and Carrie, D. (1994b) Armed to the

Teeth: An Exercise in Brand Management, Cranfield, UK: European
Case Clearing House (ref. M94-005:594-039-1/594-039-4/594-040-
1/594-040-4).

Ehrenberg, A.S.C., Uncles, M.D. and Goodhardt, G.J. (2004)

Understanding brand performance measures: using Dirichlet benchmarks,
Journal of Business Research, 57(12), 1307–25.
Eisingerich, A.B., HaeEun, H.C., Yeyi, L., He (Michael), J. and Bell, S.J.
(2014) Why recommend a brand face-to-face but not on Facebook? How
word-of-mouth on online social sites differs from traditional word-of-
mouth, Journal of Consumer Psychology [online].

Elliott, R. (1993) Shopping addiction and mood repair. In M. Davies et al.

(eds), Emerging Issues in Marketing: Proceedings of the Marketing
Education Group, Loughborough: Loughborough University Press, 287–
96.

Elliott, R. (1994) Addictive consumption: function and fragmentation in

postmodernity, Journal of Consumer Policy, 17, 159–79.

Elliott, R. and Jobber, D. (1990) Understanding organizational buying

behaviour: the role of cognitions, norms and attitudes, Proceedings of the
23rd Marketing Education Group Conference, Oxford Polytechnic, 402–
23.

Elliott, R., Eccles, S. and Gournay, K. (1996) Man management? Women

and the use of debt to control personal relationships, Proceedings of the
29th Marketing Education Group Conference, Strathclyde, available on
compact disc, ‘Buyer Behavior’ track.

Elliott, R., Jobber, D. and Sharp, J. (1995) Using the theory of reasoned
action to understand organizational behaviour: the role of belief salience,
British Journal of Social Psychology, 34(2), 161–72.

Enis, B.M. and Paul, G.W. (1970) Store loyalty as a basis for marketing
segmentation, Journal of Retailing, 46(3), 42–56.

Ennew, C.T. and Binks, M.R. (1996) The impact of service quality and
service characteristics on customer retention: small businesses and their
 banks in the UK, British Journal of Management, 7, 219–30.

Eroglu, S.A., Machleit, K.A. and Davis, L.M. (2003) Empirically testing a
model of online store atmospherics and shopper responses, Psychology
and Marketing, 20(2), 139–50.

Eskin, G. (1973) Dynamic forecasts of new product demands using a depth

of repeat model, Journal of Marketing Research, 10(2), 115–29.

Euromonitor (2006) Boomers: Now they are Sixty – Changing

Consumption Habits of 40–60-Year-Olds to 2010. Euromonitor, Global
Report – Strategy Briefing.

Euromonitor (2011) The World’s Oldest Populations. Euromonitor, Special

Report.

Faber, R.J. and O’Guinn, T.C. (1988) Compulsive consumption and credit
abuse, Journal of Consumer Policy, 11, 97–109.

Faber, R.J. and O’Guinn, T.C. (1992) A clinical screener for compulsive
buying, Journal of Consumer Research, 19(3), 459–69.

Fader, P.S., Hardie, B.G.S. and Huang, C.Y. (2004) A dynamic changepoint
model for new product sales forecasting, Marketing Science, 23(1), 50–
65.

Fader, P.S., Hardie, B.G.S. and Zeithammer, R. (2003) Forecasting new

product trial in a controlled test market environment, Journal of
Forecasting, 22(5), 391–410.
Farris, P.W. and Albion, M.S. (1980) The impact of advertising on the price
of consumer goods, Journal of Marketing, 44(3), 17–35.

Farris, P.W. and Reibstein, D.J. (1991) How prices, ad expenditures, and
profits are linked, Harvard Business Review, 57(6), 173–84.

Fay, B. (2014) Landmark study values consumer word of mouth at $6

trillion per year, MediaVillage, 2 December. Available at:
www.mediavillage.com/article/landmark-study-values-consumer-word-
of-mouth-at-6-trillion-per-year-brad-fay/

Fazio, R.H. (1986) How do attitudes guide behavior? In R.M. Sorrentino

and E.T. Higgins (eds), The Handbook of Motivation and Cognition:
Foundations of Social Behavior, New York: Guilford Press, 204–43.

Fazio, R.H. (1990) Multiple processes by which attitudes guide behavior:

the mode model as an integrative framework. In M.P. Zanna (ed.),
Advances in Experimental Social Psychology, 23, 75–109.

Fazio, R.H. and Zanna, M. (1981) Direct experience and attitude–behavior

consistency. In L. Berkowitz (ed.), Advances in Experimental Social
Psychology, 14, New York: Academic Press, 161–202.

Fazio, R.H., Powell, M.C. and Herr, P.M. (1983) Toward a process model
of the attitude–behavior relation: accessing one’s attitude upon mere
observation of the attitude object, Journal of Personality and Social
Psychology, 44, 723–35.

Feick, L.F. and Price, L.L. (1987) The market maven: a diffuser of
marketplace information, Journal of Marketing, 51(1), 83–97.
Feinberg, R.A. and Smith, P. (1989) Misperceptions of time in the sales
transaction. In T. Srull (ed.), Advances in Consumer Research, 16, 56–8.

Feinberg, R.A., Widdows, R., Hirsch-Wyncott, M. and Trappey, C. (1990)

Myth and reality in customer service: good and bad service sometimes
lead to repurchase, Journal of Consumer Satisfaction, Dissatisfaction and
Complaining Behavior, 3, 112–13.

Feinberg, R.A., Widdows, R. and Steidle, R. (1996) Customer

(dis)satisfaction and delays, Journal of Consumer Satisfaction,
Dissatisfaction and Complaining Behavior, 9, 81–5.

Feldwick, P. (1990) Advertising Works 5, Institute of Practitioners in

Advertising, London: Cassell, 181–95.

Festjens, A. and Janiszewski, C. (2015) The value of time, Journal of

Consumer Research, 42 (Aug.), 178–95.

Finn, D.W. and Lamb, C.W., Jr (1991) An evaluation of the SERVQUAL

scale in a retail setting. In R.H. Holman and M.R. Solomon (eds),
Advances in Consumer Research, 18, 483–90.

Fishbein, M. (1963) An investigation of the relationships between beliefs

about an object and attitudes to that object, Human Relations, 16, 233–
40.

Fishbein, M. and Ajzen, I. (1975) Belief, Attitude, Intention and Behavior,

Reading, MA: Addison-Wesley.

Fishbein, M.F. and Ajzen, I. (1981) On construct validity: a critique of

Miniard and Cohen’s paper, Journal of Experimental Social Psychology,
 17, 340–50.

Fisher, J.C. and Pry, R.H. (1971) A simple substitution model of

technological change, Technological Forecasting and Social Change, 3,
75–88.

Fiske, S.T. (1980) Attention and weight in person perception: the impact of
negative and extreme behavior, Journal of Personality and Social
Psychology, 38(6), 889–906.

Flavián, C., Martínez, E. and Polo, Y. (2001) Loyalty to grocery stores in

the Spanish market of the 1900s, Journal of Retailing and Consumer
Services, 8(2), 85–93.

Folkes, V.S. (1984) Consumer reactions to product failure: an attributional

approach, Journal of Consumer Research, 10(4), 398–409.

Folkes, V.S. (1988) The availability heuristic and perceived risk, Journal of
Consumer Research, 15(1), 13–23.

Folkes, V.S., Koletsky, S. and Graham, J.L. (1987) A field study of causal
inferences and consumer reaction: the view from the airport, Journal of
Consumer Research, 13(4), 534–9.

Fornell, C. (1992) A national customer satisfaction barometer: the Swedish

experience, Journal of Marketing, 56(1), 6–21.

Fornell, C. and Wernerfelt, B. (1987) Defensive marketing strategy by

customer complaint management: a theoretical analysis, Journal of
Marketing Research, 24(4), 337–46.
Fornell, C. and Wernerfelt, B. (1988) A model for customer complaining
management, Marketing Science, 7(3), Summer, 187–98.

Fornell, C., Johnson, M.D., Anderson, E.W., Cha, J. and Bryant, B.E.
(1996) The American Customer Satisfaction Index: nature, purpose, and
findings, Journal of Marketing, 60(4), 7–18.

Fornell, C., Mithas, S. and Morgeson, F. (2009) The statistical significance

of portfolio returns, International Journal of Research in Marketing,
26(2), 162–3.

Fornell, C., Mithas, S., Morgeson, F.V., III and Krishnan, M.S. (2006)
Customer satisfaction and stock prices: high returns, low risks, Journal of
Marketing, 70(1), 3–14.

Forward, S.E. (2009) The theory of planned behaviour: the role of

descriptive norms and past behaviour in the prediction of drivers’
intentions to violate, Transportation Research Part F: Traffic Psychology
and Behaviour, 12(3), 198–207.

Fournier, S. and Yao, J.L. (1997) Reviving brand loyalty: a

reconceptualization within the framework of consumer-brand
relationships, International Journal of Research in Marketing, 14(5), 451–
72.

Fournier, S., Dobscha, S. and Mick, D.G. (1998) Preventing the premature
death of relationship marketing, Harvard Business Review, Jan./Feb., 42–
51.

Fourt, L.A. and Woodlock, J.W. (1960) Early prediction of market success
for new grocery products, Journal of Marketing, 25(2), 31–8.
Foxall, G. and Hackett, P.M.W. (1992) Consumers’ perception of micro-
retail location: wayfinding and cognitive mapping in planned and organic
shopping environments, International Review of Retail, Distribution and
Consumer Research, 2(3), 309–27.

Fredricks, A.J. and Dossett, K.L. (1983) Attitude–behavior relations: a

comparison of the Fishbein–Ajzen and the Bentler–Speckart models,
Journal of Personality and Social Psychology, 45, 501–12.

Frisbie, G.A., Jr (1980) Ehrenberg’s negative binomial model applied to

grocery store trips, Journal of Marketing Research, 17, 385–90.

Fulgoni, G. and Morn, M. (2009) Whither the click? How online

advertising works, Journal of Advertising Research, 49(2), 134–42.

Fulgoni, G.M. (1987) The role of advertising: is there one? Admap, 262,
54–7.

Fulgoni, G.M. (2015) The rise of the digital omnivore: what it means for
advertisers, publishers and developers, Journal of Advertising Research,
55(2), 115–19.

Gabor, A. and Granger, C.W.J. (1961) On the price consciousness of

consumers, Applied Statistics, 10, 170–88.

Gabor, A. and Granger, C.W.J. (1966) Price as an indicator of quality:

report on an inquiry, Economica, 32, 43–70.

Gabor, A. and Granger, C.W.J. (1972) Ownership and acquisition of

consumer durables: report on the Nottingham consumer durables project,
European Journal of Marketing, 6(4), 234–48.
Ganesh, J., Arnold, M.J. and Reynolds, K.E. (2000) Understanding the
customer base of service providers: an examination of the differences
between switchers and stayers, Journal of Marketing, 64(3), 65–87.

Gardner, A.G. and Levy, S.J. (1955) The product and the brand, Harvard
Business Review, 33(Mar.–Apr.), 33–9.

Garretson, J.A. and Burton, S. (2003) Highly coupon and sale prone
consumers: benefits beyond price savings, Journal of Advertising
Research, 43(2), 162–72.

Gerard, H.B. (1967) Choice difficulty, dissonance and the decision

sequence, Journal of Personality and Social Psychology, 35(1), 91–108.

Giebelhausen, M., Robinson, S. and Cronin, J.J. (2011) Worth waiting for:
increasing satisfaction by making consumers wait, Journal of the
Academy of Marketing Science, 39(6), 889–905.

Gigerenzer, G. (1991) How to make cognitive illusions disappear: beyond

‘heuristics and biases’. In W. Stroebe and M. Hewstone (eds), European
Review of Social Psychology, 2, 83–115.

Gijsbrechts, E. (1993) Prices and pricing research in consumer marketing:

some recent developments, International Journal of Research in
Marketing, 10(2), 115–51.

Gilly, M. and Gelb, B. (1982) Post-purchase consumer processes and the

complaining consumer, Journal of Consumer Research, 9(3), 323–8.

Givon, M. and Horsky, D. (1990) Untangling the effects of purchase

reinforcement and advertising carryover, Marketing Science, 9(2), 171–
 87.

Godes, D. and Mayzlin, D. (2004a) Firm-created Word-of-mouth

Communication: a Field-based Quasi-Experiment, Harvard Business
School Marketing Research Papers No. 04–03.

Godes, D. and Mayzlin, D. (2004b) Using online conversations to study

word-of-mouth communication, Marketing Science, 23(4), 545–60.

Goldenberg, J., Libai, B., Moldovan, S. and Muller, E. (2007) The NPV of
bad news, International Journal of Research in Marketing, 24(3), 186–
200.

Goodey, C. and East, R. (2008) Testing the market maven concept, Journal
of Marketing Management, 24(3/4), 265–82.

Goodhardt, G.J. and Ehrenberg, A.S.C. (1967) Conditional trend analysis: a

breakdown by initial purchasing level, Journal of Marketing Research,
4(2), 155–61.

Goodhardt, G.J., Ehrenberg, A.S.C. and Chatfield, C. (1984) The Dirichlet:

a comprehensive model of buying behaviour, Journal of the Royal
Statistical Society, A, 147, 621–55.

Goodhardt, G.J., Ehrenberg, A.S.C. and Collins, M.A. (1975) The

Television Audience: Patterns of Viewing, Lexington, MA: Lexington
Books.

Goodhardt, G.J., Ehrenberg, A.S.C. and Collins, M.A. (1987) The

Television Audience: Patterns of Viewing – An Update, Aldershot:
Gower.
Goodman, J. and Newman, S. (2003) Understand customer behavior and
complaints, Quality Progress, 36(1), 51–5.

Gourville, J.T. (1998) Pennies-a-day: the effect of temporal reframing on

transaction evaluation, Journal of Consumer Research, 24(4), 395–409.

Graham, J. and Havlena, W.J. (2007) Finding the ‘missing link’:

advertising’s impact on word of mouth, Web searches, and site visits,
Journal of Advertising Research, 47(4), 427–35.

Granbois, D., Summers, J.O. and Frazier, G.L. (1977) Correlates of

consumer expectation and complaining behavior. In R.L. Day (ed.),
Consumer Satisfaction, Dissatisfaction and Complaining Behavior,
Bloomington, Indiana University, 18–25.

Granovetter, M.S. (1973) The strength of weak ties, American Journal of

Sociology, 78(6), 1360–80.

Green, L. (2007) Advertising Works 15, IPA Effectiveness Awards 2006,

Henley-on-Thames: World Advertising Research Center (Warc).

Green, P.E. and Krieger, A.M. (2002) What’s right with conjoint analysis?
Marketing Research, 14(1), 24–7.

Green, P.E., Krieger, A.M. and Wind, Y. (2001) Thirty years of conjoint
analysis: reflections and prospects, Interfaces, 31(3), S56–73.

Greenfield, S. (1997) The Human Brain: A Guided Tour, London: Phoenix.

Grewal, D., Kavanoor, S., Fern, E.F., Costley, C. and Barnes, J. (1997)
Comparative versus noncomparative advertising: a meta-analysis, Journal
of Marketing, 61(4), 1–15.

Grønhaug, K. (1977) Exploring complaining behavior: a model and some

empirical results. In W.D. Perreault, Jr (ed.), Advances in Consumer
Research, 4, 159–63.

Grønhaug, K. and Zaltman, G. (1981) Complainers and noncomplainers

revisited: another look at the data. In K.B. Monroe (ed.), Advances in
Consumer Research, 8, 83–7.

Grönroos, C. (1978) A service-oriented approach to marketing service,

European Journal of Marketing, 12(8), 588–601.

Grönroos, C. (1984) A service quality model and its marketing

implications, European Journal of Marketing, 18(4), 36–44.

Grönroos, C. (1994) From marketing mix to relationship marketing:

towards a paradigm shift in marketing, Management Decision, 32(2), 4–
20.

Gröppel, A. (1993) Store design and experience-orientated consumers in

retailing: comparison between United States and Germany. In W.F. van
Raaij and G.J. Bamossy (eds), European Advances in Consumer
Research, 1, 99–109.

Gruca, T.S. and Rego, L.L. (2005) Customer satisfaction, cash flow and
shareholder value, Journal of Marketing, 69(3), 115–30.
Guardian, The (2011) Supermarkets: changing market share, 16 August.
Available at www.guardian.co.uk/business/2011/aug/16/supermarkets-
market-share-kantar

Gupta, S. (1988) Impact of sales promotions on when, what, and how much
to buy, Journal of Marketing Research, 25(4), 342–55.

Gupta, S., Lehmann, D.R. and Stuart, J.A. (2004) Valuing customers,
Journal of Marketing Research, 41(1), 7–18.

Gupta, S., Van Heerde, H.J. and Wittink, D.R. (2003) Is 75% of the sales
promotion bump due to brand switching? No, only 33% is, Journal of
Marketing Research, 40(4), 481–91.

Habel, C. and Rungie, C. (2005) Drawing a double jeopardy line,

Marketing Bulletin, 16, Technical Note 1, 1–10.

Hallberg, G. (1996) All Consumers Are Not Created Equal, Hoboken, NJ:
John Wiley & Sons.

Halstead, D. (2002) Negative word of mouth: substitute for or supplement

to consumer complaints? Journal of Consumer Satisfaction,
Dissatisfaction and Complaining Behavior, 15, 1–12.

Hamilton, W., East, R. and Kalafatis, S. (1997) The measurement and utility
of brand price elasticities, Journal of Marketing Management, 13(4),
285–98.

Hammond, K.A. and Ehrenberg, A.S.C. (1995) Heavy buyers: how many
do you have? How important are they? In M. Bergardaà (ed.), Marketing
 Today and for the 21st Century. 24th EMAC Conference Proceedings,
Essec, Paris, 1651–6.

Hammond, K.A., Ehrenberg, A.S.C. and Goodhardt, G.J. (1996) Market

segmentation for competitive brands, European Journal of Marketing,
30(12), 39–49.

Hanna, N. and Wosniak, R. (2001) Consumer Behavior: An Applied

Approach, Englewood Cliffs, NJ: Prentice-Hall.

Hanssens, D.M. (2009) Advertising impact generalizations in a marketing

mix context, Journal of Advertising Research, 49(2), 127–9.

Hanssens, D.M. (2011) What is known about the long-term impact of

advertising. MASB Practitioner Paper No. 2011-01. Available at
www.anderson.ucla.edu/faculty/dominique.hanssens/Website/content/Lo
ng-Term%20PUP%202011%20-Hanssens.pdf

Hanssens, D.M., Parsons, L.J. and Schultz, R.L. (2001) Market Response
Models: Econometric and Time Series Analysis, 2nd edition, Dordrecht:
Kluwer Academic.

Hardie, B.G.S., Johnson, E.J. and Fader, P.S. (1993) Modeling loss aversion
and reference dependence effects on brand choice, Marketing Science,
12(4), 378–95.

Harnett, N., Kennedy, R., Sharp, B. and Greenacre, L. (2016) Creative that
sells: how advertising execution affects sales, Journal of Advertising,
45(1), 102–12.
Harrison, A.A. (1968) Response competition, frequency, exploratory
behavior and liking, Journal of Personality and Social Psychology, 9(4),
363–8.

Hartman, C.L., Price, L.L. and Duncan, C.P. (1990) Consumer evaluation of
franchise extension products. In M.E. Goldberg, G. Gorn and R. Pollay
(eds), Advances in Consumer Research, 17, 110–27.

Hartnett, M. (2006) Coupons still king, Frozen Food Age, Oct., 55(3), p. 41.

Heath, R. (2012) Seducing the Subconscious: The Psychology of Emotional

Influence in Advertising, Oxford: Wiley-Blackwell.

Heath, T.B., DelVecchio, D. and McCarthy, M.S. (2011) The asymmetric

effects of extending brands to lower and higher quality, Journal of
Marketing, 75(4), 3–20.

Heilman, C.M., Nakamoto, K. and Rao, A.G. (2002) Pleasant surprises:

consumer response to unexpected in-store coupons, Journal of Marketing
Research, 39(2), 242–52.

Helson, H. (1964) Adaptation Level Theory, New York: Harper & Row.

Hennig-Thurau, T. and Klee, A. (1997) The impact of customer satisfaction

and relationship quality on customer retention: a critical reassessment
and model development, Psychology and Marketing, 14(8), 737–64.

Hennig-Thurau, T., Gwinner, K.P., Walsh, G. and Gremler, D.D. (2004)

Electronic word-of-mouth via consumer-opinion platforms: what
motivates consumers to articulate themselves on the Internet? Journal of
Interactive Marketing, 18(1), 38–52.
Herr, P.M., Kardes, F.R. and Kim, J. (1991) Effects of word-of-mouth and
product-attribute information on persuasion: an accessibility-
diagnosticity perspective, Journal of Consumer Research, 17(Mar.), 454–
62.

Heskett, J.L., Sasser, W.E., Jr and Schlesinger, L.A. (1997) The Service
Profit Chain, New York: The Free Press.

Hirschman, A.O. (1970) Exit, Voice and Loyalty: Responses to Decline in

Firms, Organizations and States, Cambridge, MA: Harvard University
Press.

Hoad, A. (2005) Advertising Works 13, Henley-on-Thames: Institute of

Practitioners in Advertising, World Advertising Research Center, 123–
44.

Hoch, S.J., Drèze, X. and Purk, M.E. (1994) EDLP, Hi-Lo, and margin
arithmetic, Journal of Marketing, 58(4), 16–28.

Hofstede, G. (1980) Culture’s Consequences: International Differences in

Work-Related Values, Beverly Hills, CA: Sage.

Hofstede, G. (2001) Culture’s Consequences: Comparing Values,

Behaviors, Institutions, and Organizations across Nations, 2nd edition,
Thousand Oaks, CA: Sage.

Hong, W., Hess, T.J. and Hardin, A. (2013) When filling the wait makes it
feel longer: a paradigm shift perspective for managing online delay, MIS
Quarterly, 37(2), 383–A9.
Howard, J.A. and Sheth, J.N. (1969) The Theory of Buyer Behavior, New
York: Wiley.

Hubbard, R. and Armstrong, J.S. (1994) Replications and extensions in

marketing: rarely published but quite contrary, International Journal of
Research in Marketing, 11(3), 233–48.

Huff, D.L. (1962) Determination of Intra-Urban Retail Trade Areas, Los

Angeles: University of California, Real Estate Research Program.

Huff, D.L. (1981) Retail location theory. In R.W. Stampfl and E.C.
Hirschman (eds), Theory in Retailing: Traditional and Non-Traditional
Sources, Chicago: American Marketing Association, 108–21.

Hui, M.K. and Au, K. (2001) Justice perceptions of complaint-handling: a

cross-cultural comparison between PRC and Canadian customers,
Journal of Business Research, 52(2), 161–73.

Hui, M.K. and Bateson, J.E.G. (1991) Perceived control and the effects of
crowding and consumer choice on the service experience, Journal of
Consumer Research, 18(2), 174–84.

Hunt, H.K., Hunt, D. and Hunt, T. (1988) Consumer grudge holding,

Journal of Consumer Satisfaction, Dissatisfaction and Complaining
Behavior, 1, 116–18.

Iacobucci, D., Grayson, K.A. and Omstrom, A.L. (1994) The calculus of
service quality and customer satisfaction: theoretical and empirical
differentiation and integration, in T.A. Swartz, D.E. Bowen and S.W.
Brown (eds), Advances in Services Marketing and Management, Vol. 3,
Greenwich, CT: JAI Press, 1–68.
Inman, J.J., McAlister, L. and Hoyer, W.D. (1990) Promotion signal: proxy
for a price cut? Journal of Consumer Research, 17(1), 74–82.

IRI (1989) Larger sample, stronger proof of P-O-P effectiveness. Reprinted

from IRI which enlarges on a report that first appeared in P-O-P Times,
Mar./Apr., 28–32.

Jaccard, J.J. and Davidson, A.R. (1972) Toward an understanding of family

planning behaviors: an initial investigation, Journal of Applied Social
Psychology, 2(3), 228–35.

Jacoby, J. and Olson, J.C. (1970) An Attitudinal Model of Brand Loyalty:

Conceptual Underpinnings and Instrumentation Research. Purdue Papers
in Consumer Psychology, No. 159, Purdue University, West Lafayette,
IN.

Jeannes, M. and Beal, V. (2012) Planning for Synergy: Harnessing the

Power of Multi-Platform Media. A White Paper by the Ehrenberg-Bass
Institute for Marketing Science and CNBC.

Jin, L., He, Y. and Zhang, Y. (2014) How power states influence
consumers’ perceptions of price unfairness, Journal of Consumer
Research, 40(5), 818–33.

Jones, E.E. and Nisbett, R.E. (1972) The actor and observer: divergent
perceptions of the causes of behavior. In E.E. Jones, D.E. Kanouse, H.H.
Kelley, R.E. Nisbett, S. Valins and B. Weiner (eds), Attribution:
Perceiving the Causes of Behavior, Morristown, NJ: General Learning
Press, 79–94.

Jones, J.P. (1995a) When Ads Work: New Proof that Advertising Triggers
Sales, New York: Lexington Books.
Jones, J.P. (1995b) Single source research begins to fulfill its promise,
Journal of Advertising Research, 35(3), 9–16.

Jones, J.P. (1995c) Advertising exposure effects under a microscope,

Admap, Feb., 28–31.

Jones, T.O. and Sasser, W.E. (1995) Why satisfied customers defect,
Harvard Business Review, Nov.–Dec., 88–99.

Juster, F.T. (1966) Consumer buying intentions and purchase probability: an

experiment in survey design, Journal of the American Statistical
Association, 61(Sept.), 658–96.

Kahn, B.E. and McAlister, L. (1997) Grocery Revolution: The New Focus
on the Consumer, Reading, MA: Addison-Wesley.

Kahn, B.E. and Schmittlein, D.C. (1989) Shopping trip behavior: an

empirical investigation, Marketing Letters, 1(1), 55–69.

Kahn, B.E., Morrison, D.G. and Wright, G.P. (1986) Aggregating individual
purchases to the household level, Marketing Science, 5(3), 260–8.

Kahneman, D. (2002) Presentation following the award of the Nobel Prize

for Economics. Available at
http://nobelprize.org/nobel_prizes/economics/laureates/2002/kahneman-
lecture.html

Kahneman, D. (2012) Thinking Fast and Thinking Slow, London: Penguin

Books.
Kahneman, D. and Tversky, A. (1979) Prospect theory: an analysis of
decision under risk, Econometrica, 47, 263–91. Reprinted in D.
Kahneman and A. Tversky (2000) Choices, Values, and Frames, New
York: Russell Sage Foundation/Cambridge University Press, 17–43.

Kahneman, D. and Tversky, A. (1996) On the reality of cognitive illusions,

Psychological Review, 103(3), 582–91. Also available at
http://psy.ucsd.edu/~mckenzie/KahnemanTversky1996PsychRev.pdf

Kahneman, D. and Tversky, A. (2000) Choices, Values, and Frames, New

York: Russell Sage Foundation/Cambridge University Press.

Kahneman, D., Knetsch, J. and Thaler, R. (1991a) Anomalies: the

endowment effect, loss aversion, and status quo bias, Journal of
Economic Perspectives, 5(1), 193–206. Reprinted in D. Kahneman and
A. Tversky (2000) Choices, Values, and Frames, New York: Russell Sage
Foundation/Cambridge University Press, 159–70.

Kahneman, D., Knetsch, J.L. and Thaler, R.H. (1991b) Fairness as a

constraint on profit seeking: entitlements in the market. In R.H. Thaler
(ed.), Quasi Rational Economics, New York: Russell Sage Foundation,
199–219.

Kahneman, D., Slovic, P. and Tversky, A. (1982) Judgment under

Uncertainty: Heuristics and Biases, Cambridge: Cambridge University
Press, 117–28.

Kalafatis, S., Ledden, L., Riley, D. and Singh, J. (2016) The added value of
brand alliances in higher education, Journal of Business Research (e-
pub), http://dx.doi.org/10.1016/j.jbusres.2016.01.028
Kalafatis, S., Remizova, N., Riley, D. and Singh, J. (2012) The differential
impact of brand equity on B2B co-branding, Journal of Business &
Industrial Marketing, 27(8), 623–34.
Kalafatis, S., Riley, D. and Singh, J. (2014) Context effects in the
evaluation of business-to-business brand alliances, Industrial Marketing
Management, 43(2), 322–34.

Kalafatis, S., Singh, J., Crisafulli, B. and Dall’Olmo Riley, F. (2014)

Customer perceptions of service recovery fairness: an empirical
investigation in the Thai hotel sector, 43rd EMAC Conference, Valencia,
Spain.

Kalwani, M.U. and Yim, C.K. (1990) A price expectations model of

customer brand choice, Journal of Marketing Research, 27(3), 251–62.

Kalyanaram, G. and Little, J.D.C. (1994) An empirical analysis of latitude

of price acceptance in consumer package goods, Journal of Consumer
Research, 21(3), 408–19.

Kalyanaram, G. and Winer, R.S. (1995) Empirical generalizations from

reference price research, Marketing Science, 14(3), G161–70.

Kamakura, W.A. and Russell, G.J. (1991) Measuring Consumer Perceptions

of Brand Quality with Scanner Data: Implications for Brand Equity,
Report 91–122, Cambridge, MA: Marketing Science Institute.

Kardes, F.R., Cronley, M.L., Kellaris, J.J. and Posanac, S.S. (2004) The role
of selective information processing in price-quality inference, Journal of
Consumer Research, 31(2), 368–74.

Katz, E. (1961) The social itinerary of technical change: two studies on the
diffusion of innovation, Human Organization, 20(Summer), 70–82.
Katz, E. and Lazarsfeld, P.F. (1955) Personal Influence, Glencoe, IL: The
Free Press.

Katz, K., Larson, B. and Larson, R. (1991) Prescription for waiting-in-line

blues: entertain, enlighten, and engage, Sloan Management Review, 32,
44–53.

Kau, A.K. and Ehrenberg, A.S.C. (1984) Patterns of store choice, Journal of
Marketing Research, 21(4), 399–409.

Kaul, A. and Wittink, D.R. (1995) Empirical generalizations about the

impact of advertising on price sensitivity and price, Marketing Science,
14(3, Part 2 of 2), G151–60.

Keaveney, S.M. (1995) Customer switching behavior in service industries:

an exploratory study, Journal of Marketing, 59(2), 71–82.

Keiningham, T.L., Cooil, B., Andreasson, T.W. and Aksoy, L. (2007) A

longitudinal examination of ‘net promoter’ and firm revenue growth,
Journal of Marketing, 71(3), 39–51.

Keller, E. and Fay, B. (2006) Single-source WOM measurement: bringing

together senders’ and receivers’ inputs and outputs. In W.J. Carl (ed.),
Measuring Word of Mouth (Vol. 2), Chicago: Word of Mouth Marketing
Association, 31–41.

Keller, E. and Fay, B. (2009) The roles of advertising in word of mouth,

Journal of Advertising Research, 49(2), 154–8.

Keller, K.L. (1993) Conceptualizing, measuring, and managing customer-

based brand equity, Journal of Marketing, 57(1), 1–22.
Keller, K.L. (2002) Branding and brand equity. In B. Weitz and R. Wensley
(eds), Handbook of Marketing, London: Sage, 155–78.

Keller, K.L. and Aaker, D.A. (1992) The effects of sequential introduction
of brand extensions, Journal of Marketing Research, 29(1), 35–52.

Keller, K.L. and Lehmann, D.R. (2006) Brands and branding: research
findings and future priorities, Marketing Science, 25(6), 740–59.

Kendall, N. (1998) Advertising Works 10, Henley-on-Thames: NTC

Publications.

Kimmel, A. and Audrain-Pontevia, A.-F. (2007)Consumer response to

marketplace rumors: an exploratory cross-cultural analysis, Proceedings
of the 36th EMAC Conference, Reykjavik University, Iceland.

Kimmel, A.J. (2004) Rumors and the financial marketplace, Journal of

Behavioral Finance, 5, 232–9.

Kimmel, A.J. (2010) Connecting with Consumers, Oxford: Oxford

University Press.

Klein, G.A. (1989) Recognition-primed decisions. In W.B. Rouse (ed.),

Advances in Man–Machine System Research, 5, Greenwich, CT: JAI
Press, 47–92.

Knapp, A. (1944) A psychology of rumor, Public Opinion Quarterly, 8, 22–

7.
Knox, S.D. and Denison, T.J. (2000) Store loyalty: its impact on retail
revenue – an empirical study of purchasing behaviour in the UK, Journal
of Retailing and Consumer Services, 7(1), 33–45.

Koelemeijer, K. (1992) Measuring perceived service quality in retailing: a

comparison of methods. In K. Grunert (ed.), Marketing for Europe –
Marketing for the Future: Proceedings of the 21st Annual Conference of
the European Marketing Academy, Aarhus, Denmark, 729–44.

Koo, M. and Fishbach, A. (2010) A silver lining of standing in line:

queuing increases value of products, Journal of Marketing Research,
47(4), 713–24.

Koran, L.M., Faber, R.J., Aboujaoude, E., Large, M.D. and Serpe, R.T.
(2006) Estimated prevalence of compulsive buying behavior in the
United States, American Journal of Psychiatry, 163(10), 1806–12.

Kordupleski, R.E., Rust, R.T. and Zahoric, A.J. (1993) Why improving
quality does not improve retention (or whatever happened to
marketing?), California Management Review, Spring, 82–95.

Kotler, P. (1973) Atmosphere as a marketing tool, Journal of Retailing,

49(4), 48–63.

Kotzan, J.A. and Evanson, R.V. (1969) Responsiveness of drug store sales
to shelf space allocations, Journal of Marketing Research, 6(4), 465–9.

Krishnan, H.S. (1996) Characteristics of memory associations: a consumer-

based brand equity perspective, International Journal of Research in
Marketing, 13(4), 389–405.
Krueckeberg, H.F. (1969) The significance of consumer response to display
space reallocation, Proceedings of the American Marketing Association
Fall Conference, 30, 336–9.

Krugman, H.E. (1972) Why three exposures may be enough, Journal of

Advertising Research, 12(6), 11–14.

Kuehn, A.A. (1962) Consumer brand choice as a learning process, Journal

of Advertising Research, 2(Dec.), 10–17.

Kumar, N., Scheer, L.K. and Steenkamp, J.-B.E.M. (1995) The effects of
supplier fairness on vulnerable resellers, Journal of Marketing Research,
32(1), 54–65.

Kundenmonitor (2015) Home page. Available at

www.servicebarometer.net/kundenmonitor/en/home.html

Kunst-Wilson, W.R. and Zajonc, R.B. (1980) Affective discrimination of

stimuli that cannot be recognised, Science, 207, 557–8.

Labrecque, L.I., Patrick, V.M. and Milne, G.R. (2013) The marketers’
prismatic palette: a review of color research and future directions,
Psychology & Marketing, 30(2), 187–202.

Lambert-Pandraud, R. and Laurent, G. (2010) Why do older consumers buy

older brands? The role of attachment and declining innovativeness,
Journal of Marketing, 74(5), 104–21.

Lambert-Pandraud, R., Laurent, G. and Lapersonne, E. (2005) Repeat

purchasing of new automobiles by older consumers: empirical evidence
and interpretations, Journal of Marketing, 69(Aug.), 97–113.
Lapersonne, E., Laurent, G. and Le Goff, J.-J. (1995) Consideration sets of
size one: an empirical investigation of automobile purchases,
International Journal of Research in Marketing, 12(1), 55–66.

LaTour, S.A. and Peat, N.C. (1979) Conceptual and methodological issues
in consumer satisfaction research. In W.L. Wilkie (ed.), Advances in
Consumer Research, 6, 431–7.

Lazarsfeld, P.F., Berelson, B. and Gaudet, H. (1944) The People’s Choice:

How the Voter Makes Up His Mind in a Presidential Campaign, New
York: Columbia University, Bureau of Applied Social Research.

Le Boutillier, J., Le Boutillier, S.S. and Neslin, S.A. (1994) A replication

and extension of the Dickson and Sawyer price-awareness study,
Marketing Letters, 5(1), 31–42.

Leclerc, F., Schmitt, B.H. and Dubé, L. (1995) Waiting time and decision
making: is time like money? Journal of Consumer Research, 22(1), 110–
19.

Lee, A.Y. (1994) The mere exposure effect: is it a mere case of

misattribution? In C.T. Allen and D.R. John (eds), Advances in
Consumer Research, 21, 270–5.

Lee, L. and Tsai, C.I. (2014) How price promotions influence postpurchase
consumption experience over time, Journal of Consumer Research, 40(5),
943–59.

Leenheer, J., van Heerde, H.J., Bijmolt, T.H.A. and Smidts, A. (2007) Do
loyalty programs really enhance behavioral loyalty? An empirical
analysis accounting for self-selecting members, International Journal of
Research in Marketing, 24(1), 31–47.
Lees, G. and Wright, M. (2012) Does the duplication of viewing law apply
to radio listening? European Journal of Marketing, 47(3–4), 674–85.

Lees, G.J., Garland, B.R. and Wright, M.J. (2007) Switching banks: old
bank gone but not forgotten, Journal of Financial Services Marketing,
12(2), 146–56.

Lemon, K.N. and Nowlis, S.M. (2002) Developing synergies between

promotions and brands in different price-quality tiers, Journal of
Marketing Research, 39(2), 171–85.

Les Échos (2004) Étendre sa marque, un pari souvent gagnat. No. 19301, 7
December, 15. Available at
http://archives.lesechos.fr/archives/2004/LesEchos/19301-50-ECH.htm

Lewis, M. (2006) Customer acquisition promotions and customer asset

value. Journal of Marketing Research, 43(2), 195–203.

Lichtenstein, D.R. and Bearden, W.O. (1989) Contextual influences on

perceptions of merchant-supplied reference prices, Journal of Consumer
Research, 16(1), 55–67.

Lichtenstein, S., Slovic, P., Fischoff, B., Lyman, M. and Combs, B. (1978)
Judged frequency of lethal events, Journal of Experimental Psychology:
Human Learning and Memory, 4, 551–78.

Lilien, G.L. and Rangaswamy, A. (2002) Marketing Engineering:

Computer-assisted Marketing Analysis and Planning, 2nd edition, Upper
Saddle River, NJ: Prentice-Hall.
Litzenroth, H. (1991) A small town in Germany: single source data from a
controlled micromarket, Admap, 26(5), 23–7.

Liu, H.-S. (2012) How does online word-of-mouth influence revenue?

Evidence from Twitter. Preliminary draft, March. Available at
https://editorialexpress.com/cgi-bin/conference/download.cgi?
db_name=IIOC2012&paper_id=465

Liu, Y. (2006) Word of mouth for movies: its dynamics and influence on
box office revenue, Journal of Marketing, 70(3), 74–89.

Liu, Y. (2007) The long-term impact of loyalty programs on consumer

purchase behavior and loyalty, Journal of Marketing, 71(4), 19–35.

Liyin, J., Yanqun, H. and Ying, Z. (2014) How power states influence
consumers’ perceptions of price unfairness, Journal of Consumer
Research, 40(5), 818–33.

Lodish, L.M. and Lubetkin, B. (1992) How advertising works: general

truths? Nine key findings from IRI test data, Admap, February, 9–15.

Lodish, L.M., Abraham, M., Kalmansen, S., Livelsberger, J., Lubetkin, B.,
Richardson, B. and Stevens, M.E. (1995a) How TV advertising works: a
meta-analysis of 389 real-world split cable TV advertising experiments,
Journal of Marketing Research, 32(2), 125–39.

Lodish, L.M., Abraham, M., Livelsberger, J., Lubetkin, B., Richardson, B.

and Stevens, M.E. (1995b) A summary of fifty-five in-market
experiments on the long-term effect of TV advertising, Marketing
Science, 14(Part 2 of 2), G133–40.
Loken, B. (1983) The theory of reasoned action: examination of the
sufficiency assumption for a television viewing behavior. In R.P. Bagozzi
and A.M. Tybout (eds), Advances in Consumer Research, 10, Ann Arbor,
MI: Association for Consumer Research, 100–5.

Lomax, W., Hammond, K., Clemente, M. and East, R. (1996) New entrants
in a mature market: an empirical study of the detergent market, Journal of
Marketing Management, 12(4), 281–95.

Losch, A. (1939) The Economics of Location, translated by W.H. Woglom

and F. Stolper (1954), New Haven, CT: Yale University Press.

Luo, X., Homburg, C. and Wieseke, J. (2010) Customer satisfaction, analyst

stock recommendations, and firm value, Journal of Marketing Research,
47(6), 1041–58.

Lynch, J. (2015) Why TV is still the most effective advertising medium:

key findings from MarketShare study, Adweek, 9 June. Available at
www.adweek.com/news/advertising-branding/why-tv-still-most-
effective-advertising-medium-165247

Macintosh, G. and Lockshin, L.S. (1997) Retail relationships and store

loyalty: a multi-level perspective, International Journal of Research in
Marketing, 14(5), 487–97.

Madden, T.J., Ellen, P.S. and Ajzen, I. (1992) A comparison of the theory of
planned behavior and the theory of reasoned action, Personality and
Social Psychology Bulletin, 18(1), 3–9.

Madzharov, A.V., Block, L.G. and Morrin, M. (2015) The cool scent of
power: effects of ambient scent on consumer preferences and choice
behaviour, Journal of Marketing, 79(1), 83–96.
Mägi, A.W. (2003) Share of wallet in retailing: the effects of customer
satisfaction loyalty cards and shopper characteristics, Journal of
Retailing, 79(2), 97–106.

Magnini, V.P., Ford, J.B., Markowski, E.P. and Honeycut, E.D., Jr (2007)
The service recovery paradox: justifiable theory or smouldering myth,
Journal of Services Marketing, 21(3), 213–24.

Mahajan, V., Muller, E. and Bass, F.M. (1990) New product diffusion
models in marketing: a review and directions for research, Journal of
Marketing, 54(1), 1–26.

Maister, D.H. (1985) The psychology of waiting lines. In J.A. Czepiel,

M.R. Solomon and C.F. Surprenant (eds), The Service Encounter,
Lexington, MA: D.C. Heath, 113–24.

Malafi, T.N., Cini, M.A., Taub, S.L. and Bertolami, J. (1993) Social
influence and the decision to complain: investigations on the role of
advice, Journal of Consumer Satisfaction, Dissatisfaction and
Complaining Behavior, 6, 81–9.

Malec, J. (1982) Ad testing through the marriage of UPC scanning and

targetable TV, Admap, May, 273–9.

Mangold, W.G., Miller, F. and Brockway, G.R. (1999) Word-of-mouth

communication in the service marketplace, Journal of Services
Marketing, 13(1), 73–89.

Manning, K.C. and Sprott, D.E. (2009) Price endings, left-digit effects, and
choice, Journal of Consumer Research, 36(2), 328–35.
Marcel, J. (1976) Unconscious reading: experiments on people who do not
know they are reading. Paper presented at the British Association for the
Advancement of Science, Lancaster, UK, 1–8 September.

Markus, H. and Kitayama, S. (1991) Culture and self: implications for

cognition, emotion and motivation, Psychological Review, 98(2), 224–
53.

Markus, H. and Zajonc, R.B. (1985) The cognitive perspective in social

psychology. In G. Lindzey and E. Aronson (eds), Handbook of Social
Psychology, 3rd edition (Vol. 1), New York: Random House, 137–230.

Marsh, A. and Matheson, J. (1983) Smoking Attitudes and Behaviour: An

Enquiry Carried Out on Behalf of the Department of Health and Social
Security, London: HMSO.

Marsh, P., Barwise, P., Thomas, K. and Wensley, R. (1988) Managing

Strategic Investment Decisions in Large Diversified Companies, Working
Paper, London Business School, reviewed in The Economist, 9 July.

Marx, K. (1930/1867) Capital, Vols 1, 2 and 3, London: J.M. Dent & Sons.

Mason, N. (1991) An Investigation into Grocery Shopping Behaviour in

Britain, Headington, Oxford: Nielsen Consumer Research.

Mattila, A.S. and Patterson, P.G. (2004) The impact of culture on

consumers’ perceptions of service recovery efforts, Journal of Retailing,
80(3), 196–206.

Mattila, A.S. and Wirtz, J. (2001) Congruency of scent and music as a

driver of in-store evaluations and behavior, Journal of Retailing, 77(2),
 273–89.

Mazumdar, T. and Papatla, P. (2000) An investigation of reference price

segments, Journal of Marketing Research, 37(2), 246–58.

Mazumdar, T., Raj, S.P. and Sinha, I. (2005) Reference price research:
review and propositions, Journal of Marketing, 69(4), 84–102.

Mazzarol, T., Sweeney, J. and Soutar, G.N. (2007) Conceptualizing word-

of-mouth activity, triggers and conditions: an exploratory study,
European Journal of Marketing, 41(11/12), 1475–94.

McDonald, C. (1970) What is the short-term effect of advertising?

Proceedings of the ESOMAR Congress, Barcelona, 463–85.

McDonald, C. (1995) Advertising Reach and Frequency, Chicago: NTC

Business Books.

McKay, D.B. (1973) A spectral analysis of the frequency of supermarket

visits, Journal of Marketing Research, 10(Feb.), 84–90.

McPhee, W.N. (1963) Formal Theories of Mass Behavior, Glencoe, IL: The
Free Press.

McQuarrie, E.F. (1988) An alternative to purchase intentions: the role of

prior behaviour in consumer expenditure on computers, Journal of the
Market Research Society, 30(4), 407–37.

McSweeney, B. (2002) Hofstede’s model of national cultural differences

and their consequences: a triumph of faith – a failure of analysis, Human
 Relations, 55, 89–118.

McWilliam, G. (1993) The effect of brand typology on the evaluation of

brand extension fit: commercial and academic research findings. In W.F.
Van Raaij and G.J. Bamossy (eds), European Advances in Consumer
Research, 1, 485–91.

Mehrabian, A. and Russell, J.A. (1974) An Approach to Environmental

Psychology, Cambridge, MA: Massachusetts Institute of Technology
Press.

Mela, C.F., Gupta, S. and Lehmann, D.R. (1997) The long-term impact of
promotion and advertising on consumer choice, Journal of Marketing
Research, 34(2), 248–61.

Melnyk, V., van Osselaer, S.M.J. and Bijmolt, T.H.A. (2009) Are women
more loyal customers than men? Gender differences in loyalty to firms
and individual service providers, Journal of Marketing, 73(4), 82–96.

Meyers-Levy, J. and Tybout, A.M. (1989) Schema congruity as a basis for

product evaluation, Journal of Consumer Research, 16(1), 39–54.

Meyer-Waarden, L. (2007) The effects of loyalty programs on customer

lifetime duration and share of wallet, Journal of Retailing, 83(2), 223–6.

Michon, R., Chebat, J.-C. and Turley, L.W. (2005) Mall atmospherics: the
interaction effects of the mall environment on shopping behaviour,
Journal of Business Research, 58(5), 576–83.

Milgram, S. (1970) The experience of living in cities, Science, 167, 1464–8.

Milliman, R.E. (1982) Using background music to affect the behavior of
supermarket shoppers, Journal of Marketing, 46(3), 86–91.

MINTEL (2014) Online grocery retailing, UK report, March. Available at:

http://academic.mintel.com/display/679787/

Mittal, V., Ross, W.T. and Baldasare, P.M. (1998) The asymmetric impact
of negative and positive attribute-level performance on overall
satisfaction and repurchase intentions, Journal of Marketing, 62(1), 33–
47.

Mizerski, R.W. (1982) An attributional explanation of the disproportionate

influence of unfavorable information, Journal of Consumer Research,
9(1), 301–10.

Moldovan, S.E. (1984) Copy factors related to persuasion scores, Journal of

Advertising Research, 24(6), 16–22.

Mohr, L.A. and Bitner, M.J. (1995) The role of employee effort in
satisfaction with service transactions, Journal of Business Research,
32(3), 239–52.

Monroe, K.B. (1973) Buyers’ subjective perceptions of price, Journal of

Marketing Research, 10(1), 70–80.

Monroe, K.B. and Lee, A.V. (1999) Remembering versus knowing: issues
in buyers’ processing of price information, Journal of the Academy of
Marketing Science, 27(2), 207–25.

Moore, W.L. and Pessemier, E.A. (1993) Product Planning and

Management: Designing and Delivering Value, Singapore: McGraw-Hill.
Moraleda, P. and Ferrer-Vidal, J. (eds) (1991) Proceedings of the 1990
ESOMAR Conference, Monte Carlo.

Morgan, N.A. and Rego, L.L. (2006) The value of different customer
satisfaction and loyalty metrics in predicting business performance,
Marketing Science, 25(5), 426–39.

Morin, S., Dubé, L. and Chebat, J.-C. (2007) The role of pleasant music in
servicescapes: a test of the dual model of environmental perception,
Journal of Retailing, 83(1), 115–30.

Morrison, D. and Schmittlein, D.C. (1981) Predicting future random events

based on past performances, Management Science, 27(9), 1006–13.

Morrison, D. and Schmittlein, D.C. (1988) Generalizing the NBD model for
customer purchases: what are the implications and is it worth the effort?
Journal of Business and Economic Statistics, 6(2), 145–66.

Morrison, M., Gan, S., Dubelaar, C. and Oppewal, H. (2011) In-store music
and aroma influences on shopper behavior and satisfaction, Journal of
Business Research, 64(6), 558–64.

Murray-Burton, G., Dyke, M. and Harrison, T. (2007) Monopoly here and

now. In L. Green (ed.), Advertising Works 15, IPA Effectiveness Awards
2006, Henley-on-Thames: World Advertising Research Center (Warc).

Naples, M.J. (1979) Effective Frequency: The Relationship Between

Frequency and Advertising Effectiveness, New York: Association of
National Advertisers.
Narasimhan, C. (1984) A price discrimination theory of coupons,
Marketing Science, 3(2), 128–48.

Narisetti, R. (1997) Move to drop coupons puts Procter & Gamble in sticky
PR situation, Wall Street Journal, 17 April, 1, A10.

Naylor, G. and Kleiser, S.B. (2000) Negative versus positive word-of-

mouth: an exception to the rule, Journal of Satisfaction, Dissatisfaction
and Complaining Behavior, 13, 26–36.

Nelson-Field, K., Riebe, E. and Newstead, K. (2013) The emotions that

drive viral video,Australasian Marketing Journal, 21(4), 205–11.

Neslin, S.A. (1990) A market response model for coupon promotions,

Marketing Science, 9(2), 125–46.

Neslin, S.A. (2002) Sales Promotion, Cambridge, MA: Marketing Science

Institute.

Neslin, S.A. and Clarke, D.G. (1987) Relating the brand use profile of
coupon redeemers to brand and coupon characteristics, Journal of
Advertising Research, 27(1), 23–32.

Neslin, S.A. and Stone, L.G.S. (1996) Consumer inventory sensitivity and
the postpromotion dip, Marketing Letters, 7(1), 77–94.

Nielsen (2005) Retail Pocket Book, 2006. Oxford: WARC and AC Nielsen.

Nielsen (2011a) Online shopping. Available at

www.ukom.uk.net/media/Online%20Grocery%20Shopping_A4Final.pdf
Nielsen (2011b) Phone usage, women v. men. Available at
www.marketingcharts.com/direct/teens-sendreceive-3700-monthly-texts-
15579/nielsen-phone-usage-women-v-men-jan11gif/

Nijs, V.R., Dekimpe, M.G., Steenkamp, J.-B.E.M. and Hanssens, D.M.

(2001) The category–demand effects of price promotions, Marketing
Science, 20(1), 1–22.

Nitzan, I. and Libai, B. (2011) Social effects on customer retention, Journal

of Marketing, 75(6), 24–38.

North, A.C., Hargreaves, D.J. and Hargreaves, J.J. (2004) The uses of
music in everyday life, Music Perception, 22(1), 41–77.

North, A.C., Hargreaves, D.J. and McKendrick, J. (1999) The influence of

in-store music on wine selections, Journal of Applied Social Psychology,
84(2), 271–6.

Nunes, J.C. and Drèze, X. (2006) The endowed progress effect: how
artificial advancement increases effort, Journal of Consumer Research,
32(4), 504–12.

Odean, T. (1998) Are investors reluctant to realize their losses? Journal of

Finance, 53(5), 1775–98. Reprinted in D. Kahneman and A. Tversky
(2000) Choices, Values, and Frames, New York: Russell Sage
Foundation/Cambridge University Press, 371–92.

Office for National Statistics (ONS) (2014) Statistical Bulletin: Internet

Access – Households and Individuals. Available at
www.ons.gov.uk/peoplepopulationandcommunity/householdcharacteristi
cs/homeinternetandsocialmediausage/bulletins/internetaccesshouseholdsa
ndindividuals/2014-08-07
Ogilvy, D. (1987) Sound an alarm! International Journal of Advertising, 6,
81–4.

Olbrich, R. and Schultz, C.D. (2014) Multichannel advertising: does print

advertising affect search engine advertising? European Journal of
Marketing, 48(9/10), 1731–56.

Oliver, R.L. (1980) Cognitive model of the antecedents and consequences

of satisfaction decisions, Journal of Marketing Research, 17(4), 460–9.

Oliver, R.L. (1981) Measurement and evaluation of satisfaction processes

in retail settings, Journal of Retailing, 57(3), 25–48.

Oliver, R.L. (1987) An investigation of the interrelationship between

consumer (dis)satisfaction and complaint reports. In M. Wallendorf and
P. Anderson (eds), Advances in Consumer Research, 14, 218–22.

Oliver, R.L. (1989) Processing of the satisfaction response in consumption:

a suggested framework and research propositions, Journal of Consumer
Satisfaction, Dissatisfaction and Complaining Behavior, 2, 1–16.

Oliver, R.L. (1999) Whence customer loyalty? Journal of Marketing,

63(Special Issue), 33–44.

Oliver, R.L. and Swan, J.E. (1989) Consumer perceptions of interpersonal

equity and satisfaction in transactions: a field survey approach, Journal of
Marketing, 53(2), 21–35.

Olshavsky, R.W. and Granbois, D.H. (1979) Consumer decision making:

fact or fiction? Journal of Consumer Research, 6(2), 93–100.
Organization for Economic Cooperation and Development (OECD) (2000)
The Service Economy, Business and Industry Policy Forum Series, Paris:
OECD. Available at www.oecd.org/dataoecd/10/33/2090561.pdf

Organization for Economic Cooperation and Development (OECD) (2005)

Growth in services: fostering employment, productivity and innovation.
Available at www.oecd.org/general/34749412.pdf

Osgood, J.F., Suci, G.J. and Tannenbaum, P.H. (1957) The Measurement of
Meaning, Urbana, IL: University of Illinois Press.

Palmeira, M.M. and Srivastava, J. (2013) Free offer # Cheap product: a

selective accessibility account on the valuation of free offers, Journal of
Consumer Research, 40(4), 644–56.

Parasuraman, A., Berry, L.L. and Zeithaml, V.A. (1991) Refinement and
reassessment of the SERVQUAL scale, Journal of Retailing, 67(4), 420–
50.

Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1985) A conceptual

model of service quality and its implications for future research, Journal
of Marketing, 49(4), 41–50.

Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1988) SERVQUAL: a

multiple-item scale for measuring consumer perceptions of service
quality, Journal of Retailing, 64(1), 12–40.

Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1994) Reassessment of

expectations as a comparison standard in measuring service quality,
Journal of Marketing, 58(1), 111–24.
Partch, K. (1996) Still inching toward efficient promotion, Supermarket
Business, 51, 16.

Pauwels, K., Hanssens, D.M. and Siddarth, S. (2002) The long-term effects
of price promotions on category incidence, brand choice, and purchase
quantity, Journal of Marketing Research, 39(4), 421–39.

Pavlov, I.P. (1927) Conditioned Reflexes, translated by G.V. Anrep,

London: Oxford University Press.

Pere, V. and Dawes, J. (2012) The persistence of excess brand loyalty over
multiple years, Marketing Letters, 23(1), 163–75.

Perkins, D.S. (1993) An update on the CS/D&CB bibliography: revolution

and evolution, Journal of Satisfaction, Dissatisfaction and Complaining
Behavior, 6, 217–79.

Peterson, R.A. and Wilson, W.R. (1992) Measuring customer satisfaction:

fact and artefact, Journal of the Academy of Marketing Science, 20(1),
61–71.

Petty, R.E. and Cacioppo, J.Y. (1986) The elaboration likelihood model of
persuasion. In L. Berkowitz (ed.), Advances in Experimental Social
Psychology, 19, New York: Academic Press.

Pham, M., Goukens, C., Lehmann, D. and Stuart, J. (2010) Shaping

customer satisfaction through self-awareness cues, Journal of Marketing
Research, 47(5), 920–32.

Phillips, L.W. and Sternthal, B. (1977) Age differences in information

processing: a perspective on the aged consumer, Journal of Marketing
 Research, 14(Nov.), 744–57.

Pickering, J.F. (1975) Verbal explanations of consumer durable purchase

decisions, Journal of the Market Research Society, 17(2), 107–13.

Pickering, J.F. and Isherwood, B.C. (1974) Purchase probabilities and

consumer durable buying behaviour, Journal of the Market Research
Society, 16(3), 203–26.

Pieters, R. and Warlop, L. (1999) Visual attention during brand choice: the
impact of time pressure and task motivation, International Journal of
Research in Marketing, 16(1), 1–16.

Plassmann, H., O’Doherty, J., Shiv, B. and Rangel, A. (2008) Marketing

actions can modulate neural representations of experienced pleasantness,
Proceedings of the National Academy of Sciences of the United States of
America, 105(3), 1050–4.

Popper, K. (1979) Objective Knowledge: An Evolutionary Approach,

Oxford: Oxford University Press.

Pritchard, M.P., Havitz, M.E. and Howard, D.R. (1999) Analyzing the
commitment–loyalty link in service contexts, Journal of the Academy of
Marketing Science, 27(3), 333–48.

Pruyn, A.Th.H. and Smidts, A. (1993) Customers’ evaluations of queues:

three exploratory studies. In W.F. Van Raaij and G.J. Bamossy (eds),
European Advances in Consumer Research, 1, 371–82.

Putsis, W.M. and Srinivasan, V. (2000) Estimation techniques for macro

diffusion models. In V. Mahajan, E. Muller and Y. Wind (eds), New
 Product Diffusion Models, Norwell, MA: Kluwer Academic Publishers,
263–91.

Qin, L. (2012) An empirical investigation of online word-of-mouth

dynamic in different online communities, available at
www.swdsi.org/swdsi2012/proceedings_2012/papers/Papers/PA104.pdf

Raghubir, P. and Srivastava, J. (2002) Effect of face value on product

valuation in foreign currencies, Journal of Consumer Research, 29(3),
335–47.

Raj, S.P. (1982) The effects of advertising on high and low loyalty
segments, Journal of Advertising Research, 9(1), 77–89.

Raju, J.S. (1992) The effect of price promotions on variability in product

category sales, Marketing Science, 11(3), 207–20.

Raju, J. and Zhang, Z.J. (2010) Smart Pricing, Upper Saddle River, NJ:
Pearson Prentice Hall.

Randall, D.M. and Wolff, J.A. (1994) The time interval in the intention–
behaviour relationship, British Journal of Social Psychology, 33(4), 405–
18.

Rao, A.R. and Monroe, K.R. (1989) The effect of price, brand name, and
store name on buyers’ perceptions of product quality: an integrative
review, Journal of Marketing Research, 26(3), 351–58.

Rao, T.R. (1969) Consumer’s purchase decision process: stochastic models,

Journal of Marketing Research, 6(3), 321–9.
Reichheld, F. and Markey, R. (2011) The Ultimate Question 2.0, Boston:
Harvard Business Review Press.

Reichheld, F.F. (1993) Loyalty-based management, Harvard Business

Review, 71(2), 64–73.

Reichheld, F.F. (1996a) Learning from customer defections, Harvard

Business Review, 74(2), 56–69.

Reichheld, F.F. (with Teal, T.) (1996b) The Loyalty Effect, Boston: Harvard
Business School Publications.

Reichheld, F.F. (2003) The one number you need to grow, Harvard Business
Review, 81(12), 46–54.

Reichheld, F.F. and Kenny, D.W. (1990) The hidden advantages of

customer retention, Journal of Retail Banking,12(4), 19–23.

Reichheld, F.F. and Sasser, W.E. (1990) Zero defections: quality comes to
services, Harvard Business Review, 68(5), Sept.–Oct., 105–11.

Reilly, W.J. (1929) Methods for the Study of Retail Relationships, Austin,
TX: Bureau of Business Research Studies in Marketing, No. 4.

Reinartz, W. and Kumar, V. (2000) On the profitability of long-life

customers in a non-contractual setting: an empirical investigation and
implications for marketing, Journal of Marketing, 64(4), 17–36.

Reinartz, W. and Kumar, V. (2002) The mismanagement of customer

loyalty, Harvard Business Review, 80(7), 86–94.
Reinartz, W., Thomas, J.S. and Kumar, V. (2005) Balancing acquisition and
retention resources to maximize customer profitability, Journal of
Marketing, 69(1), 63–79.

Richins, M.L. (1981) An investigation of the consumer’s attitudes towards

complaining. In A. Mitchell (ed.), Advances in Consumer Research, 9,
502–6.

Richins, M.L. (1983) Negative word of mouth by dissatisfied consumers,

Journal of Marketing, 47(1), 68–78.

Richins, M.L. (1985) The role of product importance in complaint

initiation, Proceedings of the Eighth and Ninth Conferences on
Consumer Satisfaction and Complaining Behavior, Baton Rouge, LA and
Phoenix, AZ, 50–3.

Richins, M.L. (1987) A multivariate analysis of responses to dissatisfaction,

Journal of the Academy of Marketing Science, 15(3), 24–31.

Riebe, E., Sharp, B. and Stern, P. (2002) An empirical investigation of

customer defection and acquisition rates for declining and growing
pharmaceutical brands,Australian and New Zealand Marketing Academy
(ANZMAC) 2002 Conference Proceedings.Available at
http://members.byronsharp.com/7716.pdf

Riebe, E., Wright, M., Stern, P. and Sharp, B. (2014) How to grow a brand:
retain or acquire customers? Journal of Business Research, 67(5), 990–7.

Riskey, D.R. (1997) How TV advertising works: an industry response,

Journal of Market Research, 34(2), 292–3.
Rivis, A. and Sheeran, P. (2003) Descriptor norms as an additional predictor
in the theory of planned behaviour: a meta-analysis, Current Psychology:
Developmental, Learning, Personality, Social, 22(3), 218–33.

Roberts, A. (1996) What do we know about advertising’s short-term

effects? Admap, February, 42–5.

Roberts, A. (1999) Recency, frequency and the sales effects of TV

advertising, Admap, February, 40–4.

Roberts, A. (2000) tvSpan: the medium-term effects of TV advertising,

Admap, November, 12–14.

Roedder-John, D. and Cole, C.A. (1986) Age differences in information

processing: understanding deficits in young and elderly consumers,
Journal of Consumer Research, 13(3), 297–315.

Rogers, E.M. (1962) Diffusion of Innovations, 1st edition, New York: The
Free Press.

Rogers, E.M. (2003) Diffusion of Innovations, 5th edition, New York: The
Free Press.

Rokeach, M.J. (1973) The Nature of Human Values, New York: The Free
Press.

Romaniuk, J. (2003) Brand attributes: ‘distribution outlets’ in the mind,

Journal of Marketing Communications, 9(2), 73–92.
Romaniuk, J. (2007) Word of mouth and the viewing of television
programs, Journal of Advertising Research, 47(4), 462–71.

Romaniuk, J. (2013) Modeling mental market share, Journal of Business

Research, 66(2), 188–95.

Romaniuk, J. and Dawes, J. (2005) Loyalty to price tiers in purchases of

bottled wine, Journal of Product and Brand Management, 14(1), 57–64.

Romaniuk, J. and Gaillard, E. (2007) The relationship between unique

brand associations, brand usage and brand performance: analysis across
eight categories, Journal of Marketing Management, 23(3–4), 267–84.

Romaniuk, J. and Sharp, B. (2003) Brand salience and customer defection

in subscription markets, Journal of Marketing Management, 19(1–2), 25–
44.

Romaniuk, J. and Sharp, B. (2016) How Brands Grow: Part 2, South

Melbourne: Oxford University Press.

Romaniuk, J., Bogomolova, S. and Dall’Olmo Riley, F. (2012) Brand image

and brand usage: is a forty-year-old empirical generalization still useful?
Journal of Advertising Research, 52(2), 243–51.

Romaniuk, J., Nguyen, C. and East, R. (2011) The accuracy of self-reported

probabilities of giving recommendations, International Journal of Market
Research, 53(4), 507–31.

Rosenberg, L.J. and Czepiel, J.A. (1984) A marketing approach to customer

retention, Journal of Consumer Marketing, 1(2), 45–51.
Rosenberg, M. (2011) India’s population. Available at
geography.about.com/od/obtainpopulationdata/a/indiapopulation.htm

Rosenberg, M.J. (1956) Cognitive structure and attitudinal affect, Journal of

Abnormal and Social Psychology, 53, 367–72.

Rosnow, R.L. (2001) Rumor and gossip in interpersonal interaction and

beyond: a social exchange perspective. In R.M. Kowalski (ed.), Behaving
Badly: Aversive Behaviors in Interpersonal Relationships, Washington,
DC: American Psychological Association, 203–32.

Rossiter, J. and Bellman, S. (2005) Marketing Communications, Sydney:

Pearson Education.

Rubinson, J. (2009) Empirical evidence of TV advertising effectiveness,

Journal of Advertising Research, 49(2), 220–6.

Russo, J.E. and Leclerc, F. (1994) An eye-fixation analysis of choice

processes for consumer non-durables, Journal of Consumer Research,
21(2), 274–90.

Ryan, B. and Gross, N.C. (1943) The diffusion of hybrid seed corn in two
Iowa communities, Rural Sociology, 8, 15–24, as cited in E.M. Rogers
(2003) Diffusion of Innovations, 5th edition, New York: The Free Press.

Saegert, S.C. and Jellison, J.M. (1970) Effects of initial level of response
competition and frequency of exposure on liking and exploratory
behavior, Journal of Personality and Social Psychology, 16(3), 553–8.

Sandell, R. (1981) The dynamic relationship between attitudes and choice

behaviour in the light of cross-lagged panel correlations, Dept of
 Psychology, University of Stockholm, Report no. 581.

Sasser, W.E. (1976) Match supply and demand in the service industry,
Harvard Business Review, 54(6), 133–8.

Sattler, H., Völckner, F., Riediger, C. and Ringle, C.M. (2010) The impact
of brand extension success drivers on brand extension price premiums,
International Journal of Research in Marketing, 27(4), 319–28.

Scherhorn, G., Reisch, L.A. and Raab, G. (1990) Addictive buying in West
Germany: an empirical study, Journal of Consumer Policy, 13, 355–87.

Schindler, R.M. (2006) The 99 price ending as a signal of a low-price

appeal, Journal of Retailing, 82(1), 71–7.

Schindler, R.M. and Kirby, P.N. (1997) Patterns of rightmost digits used in
advertised prices: implications for nine-ending effects, Journal of
Consumer Research, 24(2), 192–201.

Schmidt, S. and Eisend, M. (2015) Advertising repetition: a meta-analysis

on effective frequency in advertising, Journal of Advertising, 44(4), 415–
28.

Schmitt, P., Skiera, B. and Van den Bulte, C. (2011) Referral programs and
customer value, Journal of Marketing, 75(1), 46–59.

Schmittlein, D.C., Bemmaor, A.C. and Morrison, D.G. (1985) Why does
the NBD model work? Robustness in representing product purchases,
brand purchases and imperfectly recorded purchases, Marketing Science,
4(3), 255–66.
Schmittlein, D.C., Cooper, L.G. and Morrison, D.G. (1993) Truth in
concentration in the land of (80/20) laws, Marketing Science, 12(2), 167–
83.

Schuman, H. and Johnson, M.P. (1976) Attitudes and behavior, Annual

Review of Sociology, 2, 161–207.

Schwartz, S.H. (1992) Universals in value content and structure, Advances

in Experimental Social Psychology, 25.

Schwartz, S.H. and Bilsky, W. (1987) Toward a universal psychological

structure of human values, Journal of Personality and Social Psychology,
53(3), 550–62.

Schwartz, S.H. and Bilsky, W. (1990) Toward a theory of the universal

content and structure of values: extensions and cross-cultural
replications, Journal of Personality and Social Psychology, 58(5), 878–
91.

Schwartz, S.H. and Sagiv, L. (1995) Identifying culture-specifics in the

content and structure of values, Journal of Cross-Cultural Psychology,
26(1), 92–116.

Seawright, K.K. and Sampson, S.E. (2007) A video method for empirically
studying wait-perception bias, Journal of Operations Management, 25(5),
1055–66.

Sharma, A. and Stafford, T.F. (2000) The effect of retail atmospherics on

customers’ perceptions of salespeople and customer persuasion: an
empirical investigation, Journal of Business Research, 49(2), 183–91.
Sharp, B. (2010) How Brands Grow: What Marketers Don’t Know, Oxford:
Oxford University Press.

Sharp, B. and Dawes, J. (2001) What is differentiation and how does it

work? Journal of Marketing Management, 17(7–8), 739–59.

Sharp, B. and Sharp, A. (1997) Loyalty programmes and their impact on

repeat purchase loyalty patterns, International Journal of Research in
Marketing, 14(5), 473–86.

Sharp, B., Beal, V. and Collins, M. (2009) Television: back to the future,
Journal of Advertising Research, 49(2), 211–19.

Sharp, B., Wright, M.J. and Goodhardt, G.J. (2002) Purchase loyalty is
polarised into either repertoire or subscription patterns, Australasian
Marketing Journal, 10(3), 7–20.

Sheppard, B.H., Hartwick, J. and Warshaw, P.R. (1988) The theory of

reasoned action: a meta-analysis of past research with recommendations
for modifications and future research, Journal of Consumer Research,
15(3), 325–43.

Shiv, B., Carmon, Z. and Ariely, D. (2005) Placebo effects of marketing

actions: consumers may get what they pay for, Journal of Marketing
Research, 42(4), 383–93.

Shogren, J.F., Shin, S.Y., Hayes, D.J. and Kliebenstein, J.B. (1994)
Resolving differences in willingness to pay and willingness to accept,
American Economic Review, 84(1), 255–70.
Shukla, P., Singh, J. and Banerjee, M.(2015) They are not all same:
variations in Asian consumers’ value perceptions of luxury brands,
Marketing Letters, 26(3), 265–78.

Silverman, G. (2001) The Secrets of Word-of-Mouth Marketing, New York:

AMACOM.

Simcock, P., Sudbury, L. and Wright, G. (2006) Age, perceived risk and
satisfaction in consumer decision making: a review and extension,
Journal of Marketing Management, 22(3), 355–77.

Simmel, G. (1908) The Sociology of Georg Simmel, translated by Kurt H.

Wolf, New York: The Free Press, 1964, as cited in E.M. Rogers (2003)
Diffusion of Innovations, 5th edition, New York: The Free Press.

Simon, C.J. and Sullivan, M.W. (1993) The measurement and determinants
of brand equity: a financial approach, Marketing Science, 12(1), 28–52.

Simon, H.A. (1957) Administrative Behavior, New York: Macmillan.

Simon, J.L. (1979) What do Zielske’s real data show about pulsing? Journal
of Marketing Research, 16(3), 415–20.

Simon, J.L. and Arndt, J. (1980) The shape of the advertising response
function, Journal of Advertising Research, 20(4), 11–28.

Singh, J. (1990) Voice, exit, and negative word-of-mouth behaviors: an

investigation across three categories, Journal of the Academy of
Marketing Science, 18, 1–15.
Singh, J. (2016) The influence of CSR and ethical self-identity in consumer
evaluation of co-brands, Journal of Business Ethics (e-pub), DOI:
10.1007/s10551-015-2594-4.

Singh, J. and Crisafulli, B. (2015) Customer experiences in service failure

and recovery. In S. Sahdev, K. Purani and N. Malhotra (eds), Boundary
Spanning Elements and the Marketing Function in Organizations:
Concepts and Empirical Studies, Dordrecht: Springer, 117–36.

Singh, J. and Crisafulli, B. (2016) Managing online service recovery:

procedures, justice and customer satisfaction, Journal of Service Theory
and Practice.

Singh, J. and Howell, R. (1985) Consumer complaining behaviour: a review

and prospectus. In H.K. Hunt and R.L. Day (eds), Consumer Satisfaction,
Dissatisfaction and Complaining Behavior, Bloomington, IN: Indiana
University Press, 41–9.

Singh, J. and Uncles, M. (2016) Measuring the market performance of

brands: applications in brand management. In F. Dall’Olmo Riley, J.
Singh and C. Blankson (eds), The Routledge Companion to Brand
Management, London: Routledge.

Singh, J., Dall’Olmo Riley, F., Hand, C. and Maeda, M. (2012) Measuring
brand choice in older customers in Japan, International Journal of Market
Research, 54(3), 347–68.

Singh, J., Ehrenberg, A. and Goodhardt, G. (2004) Loyalty to product

variants: a pilot, Journal of Customer Behaviour, 3(2), 123–32.

Singh, J., Ehrenberg, A. and Goodhardt, G. (2008) Measuring loyalty to

product variants, International Journal of Market Research, 50(4), 513–
 23.

Singh, J., Kalafatis, S. and Ledden, L. (2014) Consumer perceptions of co-

brands: the role of brand positioning strategies, Marketing Intelligence &
Planning, 32(2), 145–59.

Singh, J., Scriven, J., Clemente, M., Lomax, W. and Wright, M. (2012) New
brand extensions: patterns of success and failure, Journal of Advertising
Research, 52(2), 234–42.

Skinner, B.F. (1938) The Behaviour of Organisms, New York: Appleton

Century Crofts.

Skinner, B.F. (1953) Scientific and Human Behavior, New York:

Macmillan.

Skowronski, J.J. and Carlston, D.E. (1989) Negativity and extremity biases
in impression formation: a review of explanations, Psychological
Bulletin, 105(1), 131–42.

Smith, A.M. (1995) Measuring service quality: is SERVQUAL now

redundant? Journal of Marketing Management, 11, 257–76.

Smith, D.C. and Park, C.W. (1992) The effects of brand extensions on
market share and advertising efficiency, Journal of Marketing Research,
29(3), 296–313.

Smith, E.R. and Queller, S. (2001) Mental representations. In A. Tesser and

N. Schwarz (eds), Intra-individual Processes, Oxford: Blackwell.
Smith, R.B. and Sherman, E. (1992) Effects of store image and mood on
consumer behavior: a theoretical and empirical analysis. In L. McAlister
and M.L. Rothschild (eds), Advances in Consumer Research, 20, 631.

Smith, W. and Higgins, M. (2000) Reconsidering the relationship analogy,

Journal of Marketing Management, 16(1–3), 81–94.

Snow, C. (2011) Advertising Works 20: Proving the Payback on Marketing

Investment – Case Studies from the IPA Effectiveness Awards 2011,
London: Institute of Advertising Practitioners.

Solnick, S.J. and Hemenway, D. (1992) Complaints and disenrollment at a

health maintenance organization, Journal of Consumer Affairs, 26(1),
90–103.

Spangenberg, E.R., Grohmann, B. and Sprott, D.E. (2005) ‘It’s beginning to

smell (and sound) a lot like Christmas’: the interactive effects of ambient
scent and music in a retail setting, Journal of Business Research, 58(11),
1583–9.

Spence, C., Puccinelli, N.M., Grewal, D. and Roggeveen, A.L. (2014) Store
atmospherics: a multisensory perspective, Psychology & Marketing,
31(7), 472–88.

Srinivasan, S., Vanhuele, M. and Pauwels, K. (2010) Mind-set metrics in

market response models: an integrative approach, Journal of Marketing
Research, 47(4), 672–84.

Steenkamp, J.M. and Geyskens, I. (2006) How country characteristics affect

the perceived value of web sites, Journal of Marketing, 70(3), 136–50.
Steenkamp, J.M., Ter Hofstede, F. and Wedel, M. (1999) A cross-national
investigation into the individual and national cultural antecedents of
consumer innovativeness, Journal of Marketing, 63(2), 55–69.

Steiner, R.L. (1973) Does advertising lower consumer prices? Journal of

Marketing, 37(4), 19–27.

Steiner, R.L. (1993) The inverse association between the margins of

manufacturers and retailers, Review of Industrial Organisation, 8, 717–
40.

Stern, P. and Hammond, K. (2004) The relationship between customer

loyalty and purchase incidence, Marketing Letters, 15(1), 5–19.

Stern, P. and Wright, M. (2007) Predicting the Innovator, European

Marketing Academy Conference (EMAC).

Stern, P. and Wright, M. (2016) The adoption of new prescription drugs is

strongly associated with prior category prescribing rate, International
Journal of Research in Marketing, 33(1), 220–4.

Stiving, M. and Winer, R.S. (1997) An empirical analysis of price endings

with scanner data, Journal of Consumer Research, 24(1), 57–67.

Stø, E. and Glefjell, S. (1990) The complaining process in Norway: five

steps to justice, Journal of Consumer Satisfaction, Dissatisfaction and
Complaining Behavior, 3, 92–9.

Sultan, F., Farley, J.U. and Lehmann, D.R. (1990) A meta-analysis of

applications of diffusion models, Journal of Marketing Research, 27(1),
70–7.
Sundaram, D.S., Mitra, K. and Webster, C. (1998) Word-of-mouth
communications: a motivational analysis, Advances in Consumer
Research, 25, 527–31.

Sunde, L. and Brodie, R.J. (1993) Consumer evaluation of brand

extensions: further empirical results, International Journal of Research in
Marketing, 10(1), 47–53.

Sutton, S., Marsh, A. and Matheson, J. (1990) Microanalysis of smokers’

beliefs about the consequences of quitting: results from a large
population sample, Journal of Applied Social Psychology, 20(22), 1847–
62.

Swan, J.E. and Trawick, I.F. (1980) Satisfaction related to predicted versus
desired expectations. In H.K. Hunt and R.L. Day (eds), Refining
Concepts and Measures of Consumer Satisfaction and Complaining
Behavior, Bloomington, IN: School of Business, Indiana University, 7–
12.

Swan, J.E. and Trawick, I.F. (1981) Disconfirmation of expectations and

satisfaction with a retail service, Journal of Retailing, 57(3), 49–67.

Sweeney, J., Soutar, G. and Mazzarol, T. (2014) Factors enhancing word-

of-mouth influence: positive and negative service-related messages,
European Journal of Marketing, 48(1/2), 336–59.

Swinyard, W.R. (1993) The effects of mood, involvement and quality of

store experience on shopping intention, Journal of Consumer Research,
20(2), 271–80.

Tarde, G. (1903) The Laws of Imitation, translated by Elsie Clews Parson,

New York: Holt (reprinted University of Chicago Press, 1969), first
 published in 1890 as Les Lois de l’Imitation and available at
http://archive.org/details/lesloisdelimita00tarduoft

TARP (1979) Consumer Complaint Handling in America: A Summary of

Findings and a Recommendation, Washington, DC: US Office of
Consumer Affairs.

Tate, R.S. (1961) The supermarket battle for store loyalty, Journal of
Marketing, 25(6), 8–13.

Tauber, E.M. (1981) Brand franchise extension: new product benefits from
existing brand names, Business Horizons, 24, 36–41.

Tauber, E.M. (1988) Brand leverage: strategy for growth in a cost-

conscious world, Journal of Advertising Research, 28(4), 26–30.

Taylor, J.W. (1977) A striking characteristic of innovators, Journal of

Marketing Research, 14(1), 104–7.

Taylor, S. (1994) Waiting for service: the relationship between delays and
evaluations of service, Journal of Marketing, 58(2), 56–69.

Taylor, S. (1995) The effects of filled waiting time and service provider
control over the delay on evaluations of service, Journal of the Academy
of Marketing Science, 23(1), 38–48.

Taylor, S.E. (1982) The availability bias in social perception and

interaction. In D. Kahneman, P. Slovic and A. Tversky (eds), Judgment
under Uncertainty: Heuristics and Biases, Cambridge: Cambridge
University Press, 190–200.
Tellis, G.J. (1988a) Advertising exposure, loyalty and brand purchase: a
two-stage model of choice, Journal of Marketing Research, 25(2), 134–
44.

Tellis, G.J. (1988b) The price elasticity of selective demand: a meta-

analysis of economic models of sales, Journal of Marketing Research,
25(4), 331–41.

Tellis, G.J. and Wernerfelt, B. (1987) Competitive price and quality under
asymmetric information, Marketing Science, 6(3), 240–54.

Ter Hofstede, F., Steenkamp, J.-B. and Wedel, M. (1999) International

market segmentation based on consumer-product relations, Journal of
Marketing Research, 36(2), 1–17.

Thaler, R. (1980) Toward a positive theory of consumer choice, Journal of

Economic Behavior, 1, 39–60. Reprinted in D. Kahneman and A.
Tversky (2000), Choices, Values, and Frames, New York: Russell Sage
Foundation, Cambridge University Press, 269–287.

Thaler, R. (1985) Mental accounting and consumer choice, Marketing

Science, 4(3), 199–214.

Thaler, R. (1999) Mental accounting matters, Journal of Behavioral

Decision Making, 12, 183–206. Reprinted in D. Kahneman and A.
Tversky (2000) Choices, Values, and Frames, New York: Russell Sage
Foundation/Cambridge University Press, 241–68.

Thaler, R. and Sunstein, C. (2008) Nudge, Cambridge, MA: Yale University

Press.
Theil, H. and Kosobud, R.F. (1968) How informative are consumer buying
intentions surveys? Review of Economics and Statistics, 50, 50–9.

Thøgersen, J. and Zhou, Y. (2012) Chinese consumers’ adoption of a

‘green’ innovation: the case of organic food, Journal of Marketing
Management, 8(3–4), 313–33.

Thomas, M. and Morwitz, V. (2005) Penny wise and pound foolish: the left-
digit effect in price cognition, Journal of Consumer Research, 32(1), 54–
64.

Thorndike, E.L. (1911) Animal Intelligence, New York: Macmillan.

Tobin, J. (1969) A general equilibrium approach to monetary theory,

Journal of Money, Credit, and Banking, 1(1), 15–29.

Tom, G. and Lucey, S. (1995) Waiting time delays and customer

satisfaction in supermarkets, Journal of Services Marketing, 9(5), 20–9.

Totten, J.C. and Block, M.P. (1987) Analyzing Sales Promotion: Text and
Cases, Chicago: Commerce Communications Inc.

Treasure, J. (1975) How advertising works. In M. Barnes (ed.), The Three

Faces of Advertising, London: The Advertising Association, 48, 52.

Triandis, H.C. (1977) Interpersonal Behavior, Monterey, CA: Brooks Cole.

Triandis, H.C. (1989) Self and social behavior in differing cultural contexts,
Psychological Review, 96, 506–20.
Triandis, H.C. (2004) The many dimensions of culture, Academy of
Management Executive, 18(1), 88–93.

Trinh, G., Rungie, C., Wright, M., Driesener, C. and Dawes, J. (2014)
Predicting future purchases with the Poisson log-normal model,
Marketing Letters, 25(2), 219–34.

Trusov, M., Bucklin, R.E. and Pauwels, K. (2009) Effects of word-of-

mouth versus traditional marketing: findings from an Internet social
networking site, Journal of Marketing, 73(5), 90–102.

Tse, D.K. and Wilton, P.C. (1988) Models of consumer satisfaction

formation: an extension, Journal of Marketing Research, 25(2), 204–12.

Tuttle, B. (2013) The 5 big mistakes that led to Ron Johnson’s ouster at JC
Penney,Time Magazine, April. Available at
http://business.time.com/2013/04/09/the-5-big-mistakes-that-led-to-ron-
johnsons-ouster-at-jc-penney/

Tversky, A. and Kahneman, D. (1980) Causal schemas in judgments under

uncertainty, Progress in Social Psychology, 1, 49–72.

Tversky, A. and Kahneman, D. (1981) Causal schemas in judgements under

uncertainty. In ‘The framing of decisions and the psychology of choice’,
Science, 211, 453–8.

Tversky, A. and Kahneman, D. (1992) Advances in prospect theory:

cumulative representation of uncertainty, Journal of Risk and
Uncertainty, 5, 297–323. Reprinted in D. Kahneman and A. Tversky
(2000) Choices, Values, and Frames, New York: Russell Sage
Foundation/Cambridge University Press, 44–65.
Uncles, M. and Ehrenberg, A. (1990) Brand choice among older consumers,
Journal of Advertising Research, 30(4), 19–22.

Uncles, M. and Lee, D. (2006) Brand purchasing by older customers: an

investigation using the Juster scale and the Dirichlet model, Marketing
Letters, 17(1), 17–29.

Uncles, M., East, R. and Lomax, W. (2010a) Market share is correlated with
word-of-mouth volume, Australasian Marketing Journal, 18(3), 145–50.

Uncles, M., East, R. and Lomax, W. (2013) Good customers: the value of
customers by mode of acquisition, Australasian Marketing Journal, 21(2),
119–25.

Uncles, M., Kennedy, R., Nenyz-Thiel, M., Singh, J. and Kwok, S. (2012)
User profiles for competing brands seldom differ: re-examining the
evidence, Journal of Advertising Research, 52(2), 252–61.

Uncles, M.D. (2010) Retail change in China: retrospect and prospects,

International Review of Retail, Distribution and Consumer Research,
20(1), 69–84.

Uncles, M.D. (ed.) (2011) Perspectives on Brand Management, Prahran,

VIC: Tilde University Press.

Uncles, M.D. and Ehrenberg, A.S.C. (1990) The buying of packaged goods
at US retail chains, Journal of Retailing, 66(3), 278–96.

Uncles, M.D. and Hammond, K.A. (1995) Grocery store patronage,

International Journal of Retail, Distribution and Consumer Research,
5(3), 287–302.
Uncles, M.D. and Kwok, S. (2008) Generalizing patterns of store-type
patronage, International Review of Retail, Distribution and Consumer
Research, 18(5), 473–93.

Uncles, M.D. and Kwok, S. (2009) Patterns of store patronage in urban

China, Journal of Business Research, 62(1), 68–81.

Uncles, M.D., Dowling, G.R. and Hammond, K. (2003) Customer loyalty

and customer loyalty programs, Journal of Consumer Marketing, 20(4),
294–316.

Uncles, M.D., Wang, C. and Kwok, S. (2010b) A temporal analysis of

behavioural brand loyalty among urban Chinese consumers, Journal of
Marketing Management, 26(9–10), 921–42.

United Nations (2002) World Population Ageing: 1950–2050, New York:

United Nations, Department of Economic and Social Affairs, Population
Division.

Usunier, J.C. and Lee, J. (2011) Marketing Across Cultures, 5th edition,
Harlow: Pearson.

Van der Plight, J. and van Schie, E.C.M. (1990) Frames of reference,
judgement and preference. In W. Stroebe and M. Hewstone (eds),
European Review of Social Psychology, 1, Chichester: Wiley & Sons,
61–80.

van Heerde, H.J., Leeflang, P.S.H. and Wittink, D.R. (2004) Decomposing
the sales promotion bump with store data, Marketing Science, 23(3),
317–34.
Vanhuele, M. (1994) Mere exposure and the cognitive-affective debate
revisited. In C.T. Allen and D.R. John (eds), Advances in Consumer
Research, 21, 264–9.

Vanhuele, M. and Drèze, X. (2002) Measuring the price knowledge

shoppers bring to the store, Journal of Marketing, 66(4), 72–85.

Vanhuele, M., Laurent, G. and Drèze, X. (2006) Consumers’ immediate

memory for prices, Journal of Consumer Research, 33(2), 163–72.

Varan, D., Lang, A., Barwise, P., Weber, R. and Bellman, S. (2015) How
reliable are neuromarketers’ measures of advertising effectiveness? Data
from ongoing research holds no common truth among vendors, Journal of
Advertising Research, 55(June), 176–91.

Vargo, S.L. and Lusch, R.F. (2004) Evolving to a new dominant logic for
marketing, Journal of Marketing, 68(1), 1–17.

Venkatesh, V. and Davis, F.D. (2000) A theoretical extension of the

technology acceptance model: four longitudinal field studies,
Management Science, 46(2), 186–204.

Venkatraman, M.P. and Price, L.L. (1990) Differentiating between cognitive

and sensory innovativeness: concepts, measurement, and implications,
Journal of Business Research, 20(4), 293–315.

Venkatraman, V., Dimoka, A., Pavlou, P.A., Vo, K., Hampton, W.,
Bollinger, B.B., et al. (2015) Predicting advertising success beyond
traditional measures: new insights from neurophysiological methods and
market response modeling, Journal of Marketing Research, 52(4), 436–
52.
Verhoef, P.C. (2003) Understanding the effect of customer relationship
management efforts on customer retention and customer share
development, Journal of Marketing, 67(Oct.), 30–45.

Verhoef, P.C., Franses, P.H. and Hoekstra, J.C. (2002) The effect of
relational constructs on customer referrals and number of services
purchased from a multiservice provider: does age of relationship matter?
Journal of the Academy of Marketing Science, 30(3), 202–16.

Villanueva, J., Yoo, S. and Hanssens, D.M. (2008) The impact of

marketing-induced versus word-of-mouth customer acquisition on
customer equity growth, Journal of Marketing Research, 45(1), 48–59.

Viscusi, W.K. (1984) The lulling effect: the impact of child resistant
packaging on aspirin and analgesic ingestions, American Economic
Review, 74(2), 324–7.

Völckner, F. and Sattler, H. (2006) Drivers of brand extension success,

Journal of Marketing, 70(2), 18–34.

Wadhwa, M. and Kuangjie, Z. (2015) This number just feels right: the
impact of roundedness of price numbers on product evaluations, Journal
of Consumer Research, 41(5), 1172–85.

Wakefield, K.L. and Inman, J.J. (1993) Who are the price vigilantes? An
investigation of differentiating characteristics influencing price
information processing, Journal of Retailing, 69(2), 216–34.

Wallach, M.A. and Kogan, N. (1961) Aspects of judgment and decision

making: interrelationships and changes with age, Behavioral Science,
6(1), 23–36.
Wangenheim, F.v. (2005) Postswitching negative word of mouth, Journal of
Service Research, 8(1), 67–78.

Wangenheim, F. v. and Bayón, T. (2004) Satisfaction, loyalty and word of

mouth within the customer base of a utility provider: differences between
stayers, switchers and referral switchers, Journal of Consumer Behaviour,
3(1), 211–20.

Wangenheim, F. v. and Bayón, T. (2007) Behavioral consequences of

overbooking service capacity, Journal of Marketing, 71(4), 36–47.

Warland, R.H., Herrmann, R.O. and Willits, J. (1975) Dissatisfied

customers: who gets upset and what they do about it, Journal of
Consumer Affairs, 9(Winter), 152–62.

Watts, D.J. and Dodds, P.S. (2007) Influentials, networks, and public
opinion formation, Journal of Consumer Research, 34(Dec.), 441–58.

Webb, T.L. and Sheeran, P. (2006) Does changing behavioral intentions

engender behavior change? A meta-analysis of the experimental
evidence, Psychological bulletin, 132(2), 249–68.

Weber, M. and Camerer, C.F. (1998) The disposition effect in securities

trading: an experimental analysis, Journal of Economic Behavior and
Organization, 33, 167–84.

Wedel, M. and Leeflang, P.S.H. (1998) A model for the effects of

psychological pricing in Gabor–Granger price studies, Journal of
Economic Psychology, 19(2), 237–61.
Wee, C.H. and Pearce, M.R. (1985) Patronage behavior toward shopping
areas: a proposed model based on Huff’s model of retail gravitation. In
E.C. Hirschman and M.B. Holbrook (eds), Advances in Consumer
Research, 12, 592–7.

Weigel, R.H. and Newman, L.S. (1976) Increasing attitude–behavior

correspondence by broadening the scope of the behavioral measure,
Journal of Personality and Social Psychology, 33, 793–802.

Weiner, B. (1980) Human Motivation, New York: Holt, Rinehart and

Winston.

Weiner, B. (1990) Searching for the roots of applied attribution theory. In S.

Graham and V.S. Folkes (eds), Attribution Theory: Application to
Achievement, Mental Health and Interpersonal Conflict, Hillsdale, NJ:
Lawrence Erlbaum Associates, 1–16.

Wellan, D.M. and Ehrenberg, A.S.C. (1988) A successful new brand:

Shield, Journal of the Market Research Society, 30(1), 35–44.

Wellan, D.M. and Ehrenberg, A.S.C. (1990) A case of seasonal

segmentation, Marketing Research, 1, 11–13.

Westbrook, R.A. (1980) Intrapersonal affective influences upon consumer

satisfaction, Journal of Consumer Research, 7(1), 49–54.

Wetzer, I., Zeelenberg, M. and Pieters, R. (2007) ‘Never eat in that

restaurant, I did’: exploring why people engage in negative word-of-
mouth communication, Psychology and Marketing, 24(8), 661–680.
White, K.M., Smith, J.K., Terry, D.J., Greenslade, J.H. and McKimmie,
B.M. (2009) Social influence in the theory of planned behaviour: the role
of descriptive, injunctive, and ingroup norms, British Journal of Social
Psychology, 48(1), 135–58.

Whyte, W.H. (1954) The web of word of mouth, Fortune, 50(Nov.), 140.

Wicker, A.W. (1969) Attitude vs actions: the relationship of verbal and

overt behavioral responses to attitude objects, Journal of Social Issues,
25, 41–78.

Wicker, A.W. (1984) An Introduction to Ecological Psychology, Monterey,

CA: Brooks/Cole.

Wilbur, K.C. and Farris, P.W. (2014) Distribution and market share,Journal
of Retailing, 90(2), 154–67.

Wilkie, W.L. and Dickson, P.R. (1985) Shopping for Appliances:

Consumers’ Strategies and Patterns of Information Search, Cambridge,
MA: Marketing Science Institute Research Report No. 85–108.

Williams, L.G. (1966) The effect of target specification on objects fixed

during visual search, Perception and Psychophysics, 1, 315–18.

Winchester, M. and Romaniuk, J. (2008) Negative brand beliefs and brand

usage, International Journal of Market Research, 50(3), 1–20.

Winchester, M., Romaniuk, J. and Bogomolova, S. (2008) Positive and

negative brand beliefs and brand defection/uptake, European Journal of
Marketing, 42(5/6), 553–70.
Winer, R.S. (1986) A reference price model of brand choice for frequently
purchased products, Journal of Consumer Research, 13(2), 250–6.

Winer, R.S. (1988) Behavioral perspectives on pricing: buyers’ subjective

perceptions of price revisited. In T.M. Divinney (ed.), Issues in Pricing,
Lexington, MA: Lexington Books, 35–57.

Wittink, D.R., Vriens, M. and Burhenne, W. (1994) Commercial use of

conjoint analysis in Europe: results and critical reflections, International
Journal of Research in Marketing, 11(1), 41–52.

Wood, L.A. and Poltrack, D.F. (2015) Measuring the long-term effects of
advertising: Nielsen-CBS study uses single-source data to reassess the
‘two-times’multiplier, Journal of Advertising Research, 55(2), 123–31.

World Bank (2015) Services, etc., value added (% of GDP). Available at

http://data.worldbank.org/indicator/NV.SRV.TETC.ZS

Wright, M. (2016) Predicting what? The strengths and limitations of a test

of persuasive advertising principles, European Journal of Marketing,
50(1/2), 312–16.

Wright, M. and Riebe, E. (2010) Double jeopardy in brand defection,

European Journal of Marketing, 44(6), 860–73.

Wright, M. and Sharp, A. (2001) The effects of a new brand entrant on a

market, Journal of Empirical Generalisations in Marketing Science, 6,
15–29.

Wright, M. and Stern, P. (2015) Forecasting new product trial with

analogous series, Journal of Business Research, 68(8), 1732–8.
Wright, M., Sharp, A. and Sharp, B. (1998) Are Australasian brands
different? Journal of Product and Brand Management, 7(6), 465–80.

Wright, M.J. and Charlett, D. (1995) New product diffusion models in

marketing: an assessment of two approaches, Marketing Bulletin, 6, 32–
41.

Wright, M.J. and MacRae, M. (2007) Bias and variability in purchase

intention scales, Journal of the Academy of Marketing Science, 35(4),
617–24.

Wright, M.J. and Stern, P. (2006) Extending consumer trial models to

national panel data, Working Paper, Victoria University of Wellington,
New Zealand.

Wright, M.J., Upritchard, C. and Lewis, A. (1997) A validation of the Bass

new product diffusion model in New Zealand, Marketing Bulletin, 8, 15–
29.

Xia, L., Monroe, K.B. and Cox, J.L. (2004) The price is unfair! A
conceptual framework of price fairness perceptions, Journal of
Marketing, 68(4), 1–15.

Yalch, R. and Spangenberg, E. (2000) Using store music for retail zoning: a
field experiment. In L. McAlister and M.L. Rothschild (eds), Advances
in Consumer Research, 20, 632–6.

Yi, Y. (1990) A critical review of consumer satisfaction. In V.A. Zeithaml

(ed.), Review of Marketing, Chicago: American Marketing Association,
68–113.
Zajonc, R.B. (1968) Attitudinal effects of mere exposure, Journal of
Personality and Social Psychology Monograph Supplement, 9(2, Part 2),
1–27.

Zajonc, R.B. (1980) Feeling and thinking: preferences need no inferences,

American Psychologist, 35, 151–75.

Zajonc, R.B. and Rajecki, D.W. (1969) Exposure and affect: a field
experiment, Psychonomic Science, 17, 216–17.

Zeithaml, V.A. (1988) Consumer perceptions of price, quality, and value: a

means–end model and synthesis of evidence, Journal of Marketing,
52(3), 2–21.

Zeithaml, V.A. (2000) Service quality, profitability, and the economic worth
of customers: what we know and what we need to learn, Journal of the
Academy of Marketing Science, 28(1), 67–85.

Zettelmeyer, F., Morton, F.S. and Silva-Risso, J. (2006) How the Internet
lowers prices: evidence from matched survey and auto transaction data,
Journal of Marketing Research, 43(2), 168–81.

Zhang, H., Sung, J., Liu, F. and Knight, J.G. (2014) Be rational or be
emotional: advertising appeals, services types and consumer responses,
European Journal of Marketing, 48(11/12), 2105–26.

Zhang, J. (2006) An integrated choice model incorporating alternative

mechanisms for consumers’ reactions to in-store display and feature
advertising, Marketing Science, 25(3), 278–90.
Zhang, J. and Wedel, M. (2009) The effectiveness of customized
promotions in online and offline stores, Journal of Marketing Research,
46(2), 190–206.

Zhang, J., Beatty, S.E. and Walsh, G. (2008) Review and future directions
of cross-cultural consumer services research, Journal of Business
Research, 61(3), 211–24.

Zielske, H. (1959) The remembering and forgetting of advertising, Journal

of Marketing, 23(3), 239–43.

Zufryden, F.S. (1996) Multibrand transition probabilities as a function of

explanatory variables: estimation by a least squares approach, Journal of
Marketing Research, 23(2), 177–83.
Index

Aaker, D.A., 56, 57–59, 60

Aaker, J.L., 121
Abraham, M.M., 287
actions, 135–136, 139–140, 140
Adams, H.F., 283
adopter categories, 100–101
advertising effect
biometrics and, 297–298
brands and, 294–295
breakthrough and, 285–286
concentration and, 284–285, 285
consumer neuroscience and, 297–298
consumer segments and, 292–294
decay of, 288–290, 289, 290
mechanisms in, 276–282
model of, 286–288, 287
new approaches to, 290–292
new media and, 295–297
schedules and, 282–284, 283
advertising elasticity, 294–295
Advertising Research Foundation, 297
AGB, 236
age, 114, 125–128, 126, 128, 265
Ahluwalia, R., 266
Ailawadi, K.L, 217
Ailon, G., 117
Ajzen, I., 137, 139, 141, 145, 150–151, 153, 193
Aksoy, L., 192
Albion, M.S., 278
Alexander, V., 222
Allais, M., 169
Allport, G.W., 139, 250
Ambler, T., 276
American Customer Satisfaction Index (ACSI), 187, 191, 259, 272
Anderson, E.I., 214
Anderson, E.W., 191–192, 259, 260
Anderson, K., 126
Anderson, P.F., 57
Andreasen, A.R., 42, 192
Anglin, L.K., 247
annuities, 8
Araujo, T., 295–296
Arce, M., 237
Areni, C.S., 244, 246
ARIMA (autoregressive integrated moving average), 92
Armitage, C.J., 146, 152, 153, 154
Armstrong, J.S., 106, 282
Arndt, J., 266, 283
Ataman, M.B., 92, 98, 109, 281–282
atmospherics, 243–246, 245
attention, 161–163, 162
attitudes
actions and, 139–140, 140
definition of, 135–136
expected-value theory of, 136–137
measures of, 140–144, 141
modal salient beliefs and, 137–139
word of mouth and, 265
See also theory of planned behaviour (TPB)
attribution, 183–184
Au, K., 122
Audrain-Pontevia, A.-F., 250
availability heuristic, 163–164, 184, 185

Babacus, E., 190

Babin, B.J., 244
back-translation, 122–123
Bagozzi, R.P., 150, 151, 153
Baker, J., 247
Baldinger, A.L., 293
Baldwin, M.W., 153
Balter, D., 272
Bargh, J.A., 153
Bartlett, F.C., 157
Bass, F.M., 69, 82–83, 100
Bass model, 103–107, 105
Basu, K., 41
Bateson, J.E.G., 246
Batra, R., 59
Bayón, T., 192, 265
Bayus, B.L., 7
Beales, H.J., 7
Beatty, S.E., 7
BehaviorScan, 280–281
behaviourism, 22
Bellizzi, J.A., 244
Bellman, S., 53
Benartzi, S., 174
Benterud, T., 192
Bentler, P.M., 151
Berger, J., 261, 262
Bernoulli, D., 166–167
Berry, L.L., 33
Berthon, P., 120
Biel, A.L., 53
Bijmolt, T.H.A., 211
Bilsky, W., 118–119
Binet, L., 277
Binks, M.R., 42
biometrics, 297–298
Bird, M., 23
Bitner, M.J., 186
Black, D.W., 238
Blattberg, R.C., 221
Bogomolova, S., 45, 96
Boller, G.W., 190
Bolton, L.E., 213
Bolton, R.N., 42, 185, 221
bonds (fixed interest investments), 173–174
bonus offers, 215
Bornemann, T., 207
Bosmans, A., 244
Bottomley, P.A., 59, 60
brand alliance (co-branding), 56, 60–61
brand attitude, 50, 54, 54
brand awareness, 49, 49, 52–53
brand commitment, 267–268
brand equity, 53–55, 54
brand extension, 12, 55–63, 58, 62, 179
brand image, 49–53, 49, 80
brand insistence, 28
brand knowledge, 49–53, 49
brand names, 56–57
brand strength, 50–52, 54, 54
brands
advertising effect and, 294–295
definition of, 17–18
mental representation of, 47–49, 48
breakthrough, 285–286
Briesch, R.A., 204
Brislin, R.W., 122–123
Broadbent, S., 221, 276, 284, 288
Broadbent, T., 277
Brodie, R.J., 59
Brown, G.H., 29
Brown, J.R., 246
Buday, T., 62
Budd, R., 137
Burberry, 81
Burke, R.R., 285–286
Burnkrant, R.E., 283
Burton, S., 222
Butman, J., 272
Buzzell, R.D., 179
BzzAgent, 250–251

Cacioppo, J.Y., 286

Camerer, C.F., 172–173
Campbell, M.C., 212
cannibalization, 61–63
Cardozo, R.N., 183
Carl, W.J., 250–251
Carlston, D.E., 266
Carman, J.M., 188, 239–240
cashback (rebates), 215
categories, 17
category extension, 55, 58–59
causal inferences, 184, 185–186
causal relationships, 22–23
Cebollada, J., 237
central place theory, 227
Chan, H., 122
Chan, T., 217
Chandon, P., 144, 223
change aversion, 127
Charlett, D., 100
Charlton, P., 239, 240
Chatfield, C., 82–83
Chawdhary, R., 262
Cheng, Y., 196
Chevalier, J.A., 262
Chicago Tribune (newspaper), 29
China, 124–125, 131n2
Chitpanorak, P., 42
Christiansen, T., 253
Churchill, G.A., Jr, 183
Churchill, H., 28
Clare, J.E., 22
Clarke, D.G., 288
classical conditioning, 11–12, 12
Clemmer, E.C., 195
co-branding (brand alliance), 56, 60–61
cognitive decline, 127
cognitive model, 6–9
cognitivism, 22
Coleman, J., 251
Collins, M., 79, 82
colour, 244
comparison, 16, 17
compatibility, 140–142
complaining, 192–195, 193
compulsive shoppers, 237–238, 239
concentration, 284–285, 285
conditional trend analysis (CTA), 93–94
confirmation model of consumer satisfaction, 180–181, 181
conjoint analysis, 209
Conner, M., 146, 152, 153, 154
consumer behaviour, scope of, 3–5
consumer confusion, 50–51
consumer decision models, 5–15, 6
consumer group differences
age and, 114, 125–128, 126, 128
gender and, 114, 128–129
importance of, 113–114
national cultural differences, 115–125, 115, 118, 119
segmentation and, 114–115
consumer neuroscience, 297–298
consumer panel studies, 28–29, 65
consumer satisfaction and quality
loyalty and, 178–179
measures of, 186–190
outcomes of, 190–198, 191, 193, 196, 198
service quality and, 179–180
theories of, 180–186, 181, 183
word of mouth and, 259–261, 260
consumer segmentation, 21–22, 292–294
contentment, 180–181
contests, 215
controllability, 185
convergence, 114, 124–125
Copeland, M.T., 28
Coulter, K.S., 214–215
Coulter, R.A., 214–215
couponing, 215–216, 222–223
Critical Incident Technique (CIT), 44
Crocker, J., 157
Cronin, J.J., Jr, 188, 190
Crosby, L.A., 42
cross-purchase, 78–82, 79
crowding, 246–247, 247
culture, 115–125, 115, 119
Cunningham, R.M., 30
customer acquisition, 38, 96
customer churn, 93, 95–96
customer loyalty. See loyalty
Customer Rage Study (CCMC, 2013), 179, 192–193, 195
customer relationship management (CRM), 33
customer surveys, 186–187, 209, 253–256
customer tenure, 34–37, 34, 37, 265
customer typologies
compulsive shoppers, 237–238, 239
heavy shoppers, 237, 293–294
store-loyal customers, 235–237, 239–240
studies on, 234–235
customized promotions, 223–224
Czepiel, J.A., 33–34

Dabholkar, P.A., 188

Dacin, P.A., 57
Dall’Olmo Riley, F.D., 23, 57, 262
Danaher, P.J., 286
Dant, R.P., 246
D’Astous, A., 238
Davidson, A.R., 147
Davis, F.D., 154
Dawar, N., 57
Dawes, J., 33, 43
Day, D., 143
Day, G.S., 40–41
Day, R.L., 193
De Matos, C.A., 194
de Mooij, M., 117
decline, 107, 108–109
defection, 37, 38, 42–45, 44–45, 95–97
Degeratu, A.M., 237
Dekimpe, M.G., 221–222
delays, 195–198, 196, 198
deliberate actions, 152
Denison, T.J., 237
desensitization, 11
Dichter, E., 257, 262
Dick, A.S., 41
Dickson, P.R., 7, 205–206
diffusion theory, 108
Dimson, E., 174
Dinner, I.M., 296
direct price reduction, 215–216, 217–219, 218
direct product profitability (DPP), 243
Dirichlet model, 66–67, 82–86, 83, 85, 235, 241–242
disconfirmation model of consumer satisfaction, 182–183, 183
discontent, 180–181
discretionary loyalty, 240
displays, 215–216, 218–219, 218
Dittmar, H., 238
divided brand loyalty, 29
Dodds, P.S., 268–270, 272
dominant logic, 19
Donovan, R.J., 244–245
Dossett, K.L., 151
double jeopardy (DJ), 75–78, 76
Dowling, G.R., 35
Doyle, P., 57
Drèze, X., 39–40, 206, 243
Driesener, C., 52
Driver, B.L., 153
Dubé-Rioux, L., 196
Dunn, R.S., 230, 237, 240
Dunnhumby, 231, 234–235, 241
duplication of purchase law, 79–80

early adopters, 100

early majority, 100
East, R.
on complaining, 194
on decision-making, 144, 147, 152, 161–162
on delays, 196–197
on loyalty, 30–32, 35–36, 37, 38, 41, 43, 44–45, 70, 94–95, 235,
239, 240
on market share, 21
on promotions, 219
on shopper choice, 233
on word of mouth, 127, 253, 256–257, 260–261, 263, 266–267,
272
Les Échos (newspaper), 56
econometric estimation, 209, 210–211
Edwards, W., 137
Efficient Consumer Response movement, 56
Ehrenberg, A.S.C.
on brand awareness, 52
on market dynamics, 110
on market share, 20
on price experiments, 210
on prior behaviour, 23
on promotions, 219–220
on repertoire markets, 93
on stationary markets, 66, 74, 76, 78–79, 80, 82, 83, 84
Eisend, M., 284
Eisingerich, A.B., 262
electroencephalograms (EEG), 297
elicitation, 137–139
Elliott, R., 137, 238
empirical generalizations, 4
England, L.R., 210
Enis, B.M., 239–240
Ennew, C.T., 42
environmental determinism, 124–125
equities, 173–174
Eroglu, S.A., 245
Eskin, G., 111
Euromonitor, 125
everyday low price (EDLP) strategy, 217
exit, 179
expected-value theory of attitude, 136–137
exponential decline, 107, 108–109
external reference prices (ERP), 204–205
extra quantity, 215

Faber, R.J., 238

Fader, P.S., 111
Farris, P.W., 278
Fay, B., 54, 252, 271, 280
Fazio, R.H., 152–153, 153, 286
feature advertising, 215–216, 218–219, 218
Feick, L.F., 101
Feinberg, R.A., 183–184, 195
femininity, 115–116, 115, 121
Festjens, A., 175
Field, P., 277
Finn, D.W., 188
first purchases, 109–111, 110
first-store loyalty, 235
Fishbein, M., 137, 139, 141, 144–145, 147, 150–151
Fisher, J.C., 101–103, 102
Fiske, S.T., 176
fixed interest investments (bonds), 173–174
Flavián, C., 32, 240
Floyd, J., 212
Folkes, V.S., 185, 196
Fornell, C., 183–184, 187, 192, 194, 272
Fortune (magazine), 212
Forward, S.E., 146
Fournier, S., 35
Fourt, L.A., 111
Fox, S., 278–279
Foxall, G., 227
framing, 213–214, 287
Fredricks, A.J., 151
Frisbie, G.A., Jr, 230
Fry, T., 288
Fulgoni, G.M., 296
functional magnetic resonance imaging (fMRI), 297

Gabor, A., 142, 143

Gaillard, E., 50
Gale, B.T., 179
games, 215
Gamma distribution, 70, 71
Ganesh, J., 179
Gardner, A.G., 49
Garretson, J.A., 222
Gelb, B., 194
gender, 114, 128–129
geographical extension, 55
Gerard, H.B., 161–162
Geyskens, I., 120
Giebelhausen, M., 198
Gigerenzer, G., 165–166
Gilbert & George, 32
Gilly, M., 194
Givon, M., 288
Glefjell, S., 192
globalization, 123–124
Godes, D., 251, 253, 262, 265
Goodey, C., 272
Goodhardt, G.J., 67, 79, 82–83, 93, 110
Goodman, J., 260
goods, 19
Gourville, J.T., 213
Granbois, D.H., 7
Granger, C.W.J., 142, 143
Granovetter, M.S., 98
gravity models, 226–228
Grønhaug, K., 193–194
Gröppel, A., 245
Gross, N.C., 98
Gruca, T.S., 192
Gupta, S., 95, 219

Habel, C., 76
habit model, 6–7, 12–13
habits, 32–33, 240
Hackett, P.M.W., 227
Hamilton, W., 277
Hammond, K., 37
Hanssens, D.M., 281–282, 289
Hardie, B.G.S., 204
Hargreaves Lansdown, 8
Harnett, N., 282
Harrison, A.A., 159, 160
Hartman, C.L., 59
Heath, R., 291
Heath, T.B., 57
heavy-half principle, 71–73
heavy shoppers, 237, 293–294
Heilman, C.M., 223
Hellman’s, 51–52
Helson, H., 204
Hemenway, D., 194
Hennig-Thurau, T., 42, 262
heuristics, 163–166
Hi-Lo strategy, 217
Higgins, M., 35
Highley, J., 235
Hirschman, A.O., 179, 193
Hoch, S.J, 217
Hofstede, G., 113–114, 115–117, 115, 117, 121
Hogg, A., 38
Holden, S.J.S., 59, 60
Holmes, J.G., 153
Homburg, C., 207
Horsky, D., 288
Howard, J.A., 183
Howell, R., 193
Huff, D.L., 227
Hui, M.K., 122, 246
Hunt, H.K., 32–33

imitation, 104
individualism–collectivism, 115–116, 115, 120–122
influentials, 22
Information Resources Inc. (IRI), 28–29, 218–219, 280–281, 284, 289
Inman, J.J., 206, 219
innovation
models of, 98–107, 99, 102, 105, 107
national cultural differences and, 121
word of mouth and, 249–250
innovators, 100
intention, 146–150
Interbrand Group, 64n1
internal reference prices (IRP), 204–205
Internet
advertising effect and, 295–297
complaining and, 192–193, 193
customized promotions and, 223–224
decision-making and, 14–15, 15
delays and, 195–196
loyalty and, 236–237
 price experiments and, 210
word of mouth and, 251–253, 295
investment, 173–174
involvement, 6
iPhone, 48, 48
Isherwood, B.C., 142, 143
Iyengar, R., 262

Jaccard, J.J., 147

Jacoby, J., 40
Janiszewski, C., 175
Jellison, J.M., 159
Jobber, D., 137
Johnson, M.P., 139
Johnson, R., 222–223
Jones, J.P., 287
Juster, F.T., 143, 266

Kahn, B.E., 69, 70, 230

Kahneman, D., 163, 164–166, 168, 170, 175, 211–212
Kalafatis, S., 61, 122
Kalyanaram, G., 205
Kamakura, W.A., 54
Kantar Worldpanel, 229
Kardes, F.R., 207
Katz, E., 251, 257, 268–269
Keaveney, S.M., 43, 44, 257
Keller, E., 54, 252, 271, 280
Keller Fay Group, 252, 262–263
Keller, K.L., 57–59, 60
Kendall, N., 281
key account management, 72
Kim, D., 244, 246
Kimmel, A., 250
Kimmel, S.K., 153
Kiser, C.V., 22
Kitayama, S., 118
Klee, A., 42
Klein, G.A., 8
Kleiser, S.B., 262
Knapp, A., 250
Knox, S.D., 237
Koelemeijer, K., 188
Koran, L.M., 237–238
Kordupleski, R.E., 42
Kosobud, R.F., 143
Kotler, P., 244
Krishnan, H.S., 54
Krugman, H.E., 284
Kuangjie, Z., 214
Kuehn, A.A., 69
Kumar, V., 35, 37
Kunst-Wilson, W.R., 160

Labrecque, L.I., 244

laggards, 100
Lamb, C.W., Jr, 188
Lambert-Pandraud, R, 126, 127
Lapersonne, E., 43
late majority, 100
Lazarsfeld, P.F., 257, 268–269
Le Boutillier, J., 206
Leaky Bucket Theory, 93
learning theory, 10–13, 10, 12
Leclerc, F., 162, 175
Lee, A.Y., 160
Lee, D., 114
Lee, L., 223
Leenheer, J., 242
Lees, G.J., 82, 95, 96
Lehmann, D.R., 59
Lemon, K.N., 219
Levy, S.J., 49
Libai, B., 43
Libération (newspaper), 212
Lichtenstein, S., 163
line extension, 55, 56
Little, J.D.C., 205
Liu, Y., 39, 242, 251
locus of causality, 185
Lodish, L.M., 284, 287, 289, 294–295
Loken, B., 151
Lomax, W., 62
long-shot bias, 174
long-term decay, 289–290, 289
long-term orientation, 115, 116
loss aversion, 168, 173–174
low-innovation products, 107, 108–109
loyalty
advertising effect and, 292–293
age and, 126
combination definitions of, 40–43, 41–42
consumer satisfaction and, 178–179
Dirichlet model and, 67
gender and, 128–129
Internet and, 236–237
levels of, 18
market dynamics and, 92–98, 94
reasons for defection and, 43–45, 44–45
in repertoire markets, 28–33, 31, 94–95
as retention, 33–40, 34, 36, 37
store-loyal customers and, 235–237, 239–240
in subscription markets, 95–96
television viewing and, 82
types of, 27–28
See also share-of-category requirement (SCR)
loyalty programmes, 39–40, 240–242, 242
loyalty proneness, 30–32
Lubetkin, B., 284, 289
Luo, X., 192
Lusch, R.F., 19
MacRae, M., 143
Madden, T.J., 146–147
Madzharov, A.V., 244
Magnini, V.P., 194
Mangold, W.G., 260, 262
Manning, K.C., 214
Mansi, S.A., 192
Marcel, J., 160–161
market concentration, 20
market mavens, 101
market partitioning, 79
market potential, 104
market share, 20–21, 21, 91
marketing, 4–5
Markus, H., 118, 163
Marsh, A., 151
Marsh, P., 8
Marx, K., 19, 23
masculinity, 115–116, 115, 121
Mason, N., 236, 237, 240
Matheson, J., 151
Mattila, A.S., 122, 244
Mayzlin, D., 251, 253, 262, 265
Mazumdar, T., 204–205
Mazzarol, T., 267
McDonald, C., 283
McKay, D.B., 230
McPhee, W.N., 75–76
McQuarrie, E.F., 144
McSweeney, B., 116–117
McWilliam, G., 57, 59–60
mean population purchase frequency, 68–69
Mehrabian, A., 244, 245
Mela, C.F., 278
Melnyk, V., 128–129
memory, 48
mental accounting, 172
mental representations theory, 48–49, 48, 50, 62–63, 136
mere exposure, 159–160
Meyer-Waarden, L., 241–242
Meyers-Levy, J., 59
Michon, R., 247
Milgram, S., 247
Milliman, R.E., 244, 245
MINTEL, 236
modal salient beliefs, 137–139
Mohr, L.A., 186
Monroe, K.B., 204
Monroe, K.R., 207
Morin, S., 246
Morn, M., 296
Morrison, M., 244
Morwitz, V., 214
multi-brand purchase, 78–82, 79
multibuy, 215
Murray-Burton, G., 277
music, 244

Naples, M.J., 284

Narasimhan, C., 222
Narisetti, R., 223
national cultural differences, 115–125, 115, 118, 119
Naylor, G., 262
NDB-Dirichlet model of purchase incidence, 66–67
negative binomial distribution (NBD), 66–67, 70, 83, 93, 220
negativity bias, 176
Nelson-Field, K., 296
Neslin, S.A., 220–221
Net Promoter Score (NPS), 96, 271–272, 271
new media, 295–297
Newman, L.S., 141
Newman, S., 260
niche brands, 81
Nielsen, 129, 229, 236
Nijs, V.R., 222
Nitzan, I., 43
normal distribution, 98–99, 99
North, A.C., 244, 246
Nowlis, S.M., 219
Nunes J.C., 39–40

odd pricing, 214

Odean, T., 173
Office for National Statistics (ONS), 15
Ogilvy, D., 221
O’Guinn, T.C., 238
Olbrich, R., 296
Oliver, R.L., 180–181, 183–184, 192
Olshavsky, R.W., 7
Olson, J.C., 40
one-child policy, 131n2
optimal scaling, 150

Palmeira, M.M., 221

panel research, 28–29, 65
Papatla, P., 204–205
Parasuraman, A., 188–190
Pareto, V., 72
Park, C.W., 57
Patterson, P.G., 122
Paul, G.W., 239–240
Pauwels, K., 217, 222
Pavlov, I.P., 11–12, 12
Pearce, M.R., 227
penetration
product category and, 75–78, 76, 77
single brands and, 68–69, 73, 74, 74
pension funds, 8
perceived control (PC), 145, 146–150, 152, 154
personality traits, 114–115
Peterson, R.A., 263
Petty, R.E., 286
Pham, M., 184
Pickering, J.F., 142, 143, 144
Pieters, R., 162
Plassmann, H., 208
Poisson distribution, 70
Poltrack, D.F., 289–290, 293
Popper, K., 120
positioning, 80–82
Postman, L., 250
power distance, 115–116, 115
price
consumer response to, 204–208, 208
psychological reactions to, 211–215
See also sales promotions
price changes, 211–213
price elasticity, 208–209
price endings, 214
price experiments, 209–210
price memory, 205–207
price–quality relationship, 207–208, 208
price sensitivity, 208–211, 277–278
price tolerance, 36–37
Price, L.L., 101, 114–115
probability, 169–170, 170
products, 17
propensity, 12
Prospect Theory, 170–176
Pruyn, A.Th.H., 197
Pry, R.H., 101–103, 102
purchase frequency
advertising effect and, 293–294
product category and, 75–78, 76, 77
single brands and, 68–75, 71, 71, 74
purchase intentions, 142–144
purchase occasions, 68–69
purchase patterns
 product categories and, 75–86, 76, 77, 79, 83, 85
single brands and, 69–75, 71, 71, 74
purchase reinforcement, 287–288
Putsis, W.M., 105

Qin, L., 251

quality. See consumer satisfaction and quality
Queller, S., 48

Raj, S.P., 292–293

Rajecki, D.W., 160
Raju, J.S., 221
Randall, D.M., 144
Rao, A.R., 207
Rao, T.R., 32
rebates (cashback), 215
recall, 52–53
recall bias, 256–257
recent purchase, 69–70
recognition, 52–53
reference prices, 204–205
referral programmes, 272–273
reflection effect, 168
Rego, L.L., 192
regression analysis, 92
Reichheld, F.F., 34, 36–39, 42, 95, 96, 179, 272
Reilly, W.J., 227
Reinartz, W., 35, 37
reinforcement model, 6–7, 10–13, 10
relationship marketing (RM), 33–40, 34, 36, 37
repeat purchase, 73–74, 74, 93–94, 111
repertoire markets, 20, 28–33, 31, 94–95
representativeness heuristic, 164–165
resource constraint, 239
response competition, 158–160
retention, 33–40, 34, 36, 37, 235
retrieval bias, 44, 157, 166
reverse causality, 92
Richins, M.L., 193
Riebe, E., 38, 95, 96
risk and risk aversion, 127, 163–164, 168, 175
Riskey, D.R., 280, 287, 294
Roberts, A., 283, 284–285, 288, 289, 293, 294
Rogers, E.M., 98–101, 106, 108
Rokeach, M.J., 113–114, 117–119, 118
Rokeach Value Survey, 117–119, 118
Romaniuk, J.
on brand knowledge, 50, 52
on defection, 96
on loyalty, 33, 38, 45
on stationary markets, 73
on word of mouth, 262, 263, 271
Rosenberg, L.J., 33–34
Rosenberg, M.J., 137
Rosnow, R.L., 250
Rossiter, J.R., 53, 244–245
Rubinson, J., 293
rumours, 250
Rungie, C., 76
Russell, G.J., 54
Russell, J.A., 244, 245
Russo, J.E., 162
Ryan, B., 98

Saegert, S.C., 159

Sagiv, L., 121
sales equation, 68–69
sales fluctuations, 90–92, 91
sales promotions
couponing, 215–216, 222–223
customized promotions, 223–224
effects of, 90–92, 91, 98, 216–222, 218, 220–221, 220
types of, 215–216
sales rate, 68–69
Sampson, S.E., 196
Sasser, W.E., 230
satisfaction, 42–43, 140
satisficing model, 8–9
Sattler, H., 59, 60
Sawyer, A.G., 205–206
scents, 244
schedules, 282–283, 283
schemas, 157–161
Scherhorn, G., 238
Schindler, R.M., 214
Schmidt, S., 284
Schmittlein, D.C., 70, 72, 230
Schneider, B., 195
Schultz, C.D., 296
Schuman, H., 139
Schwartz, E.M., 261
Schwartz, S.H., 118–119, 121
seasonality, 90–92, 91
Seawright, K.K., 196
segmentation, 21–22, 114–115
service recovery paradox (SRP), 194
services, 19
SERVQUAL, 188–190
shaping, 11
share-of-category requirement (SCR)
advertising effect and, 292–293
definition of, 20, 30, 31, 68–69
loyalty programmes and, 241–242
multi-brand purchase and, 78–79
store loyalty and, 235–236
word of mouth and, 265
Sharma, A., 245
Sharp, A., 110, 241–242
Sharp, B., 20, 38, 50, 73, 126, 241–242
Sheeran, P., 146
Sheppard, B.H., 143
Sherman, E., 245
Sheth, J.N., 183
Shiv, B., 208
shopper choice
gravity models and, 226–228
shopping times and, 233–234, 234
shopping trip patterns and, 230–233, 231–232, 232–233
store preferences and, 228–229, 229, 230
short-term decay, 288–289
short-term orientation, 115, 116
Shukla, P., 5, 120
Silverman, G., 260
Simcock, P., 127
Simester, D., 214
Simmel, G., 98
Simon, C.J., 64n1
Simon, H.A., 8
Simon, J.L., 283
Singh, J.
on brands, 60–61, 63, 75
on complaining, 193
on loyalty, 33
on market dynamics, 111
on segmentation, 114
on stationary markets, 80
on stock keeping units, 17
on variants, 18
Skinner, B.F., 10–11, 22
Skowronski, J.J., 266
SKUs (stock keeping units), 17
Smidts, A., 197
Smith, A.M., 190
Smith, D.C., 57
Smith, E.R., 48
Smith, P., 195
Smith, R.B., 245
Smith, S.M., 7
Smith, W., 35
smoking, 147–148
social advertising, 281–282
social influence, 288
socio-demographic characteristics, 114
socio-emotional selectivity, 127
sole-brand loyalty, 28, 29
Solnick, S.J., 194
Spangenberg, E., 244
Speckart, G., 151
spontaneous actions, 152
Sprott, D.E., 214
Srinivasan, S., 92, 98
Srinivasan, V., 105
Srivastava, J., 221
Srull, T.K., 285–286
stability, 185
Stafford, T.F., 245
stationary markets
loyalty and, 92–98, 94
mathematical models and, 66–69
role of panel research in, 65
sales fluctuations in, 90–92
single brand purchase patterns in, 69–75, 71, 71, 74
whole category purchase patterns in, 75–86, 76, 77, 79, 83, 85
Steenkamp, J.M., 120, 121
Steiner, R.L., 278
Stephens, N., 42
Stern, P., 109, 111
stimulus–organism–response (SOR) model, 244–245
stimulus–response (SR) model, 245–246
Stø, E., 192
stock keeping units (SKUs), 17
stock-outs, 288
stockpiling, 217
Stone, L.G.S., 220–221
store environment
 atmospherics, 243–246, 245
crowding, 246–247, 247
store layout and locations, 242–243
store loyalty, 32, 235–237, 239–240
Stroop test, 158
sub-brands, 17
subjective expected utility (SEU) model of decision, 137, 146
subjective norm (SN), 145, 146, 152, 154
subscription markets, 20, 95–96
sufficiency, 150–151
Sullivan, M.W., 64n1
Sultan, F., 106
Sundaram, D.S, 262
Sunde, L., 59
surface effects, 214–215
Surprenant, C., 183
surveys, 186–187, 209, 253–256
Sutton, S., 147–148
Swan, J.E., 183–184
Swedish Customer Satisfaction Barometer, 191, 259
Sweeney, J., 267
Swinyard, W.R., 245

TACT (target, action, context and time), 140

Tarde, G., 98
TARP, 194
Tauber, E.M., 57
Tax, S.S., 253
Taylor, J.W., 100, 111
Taylor, S., 195–196
Taylor, S.A., 188, 190
Taylor, S.E., 44, 157
Taylor Nelson Sofres, 241
technology acceptance model (TAM), 154
technology substitution model, 101–103, 102
television, 82, 294, 296–297
Tellis, G.J., 207, 210, 293
Thaler, R.H., 168, 171–172, 174
Theil, H., 143
theory of planned behaviour (TPB)
adjustments and alternatives to, 153–154
complaining and, 193–194
overview, 144–146, 145
problems with, 150–154, 151, 153
use of, 146–150
Thøgersen, J., 125
Thomas, M., 214
Thorndike, E.L., 10
three-hit theory, 284
translation, 122–123
Treasure, J., 78
trial and error learning, 10
Trusov, M., 258
Tsai, C.I., 223
Tsai, Y., 196
Tversky, A., 164–165, 168, 170, 175
Tybout, A.M., 59

uncertainty avoidance, 115–116, 115, 121

Uncles, M.
on Chinese consumers, 124–125
on loyalty, 35
on segmentation, 114
on stationary markets, 80, 82
on word of mouth, 258–259, 264, 265
unique selling proposition (USP), 80
Unnava, H.R., 283
unstable loyalty, 29

value, 166–168, 167

Van der Plight, J., 175
van Heerde, H.J., 217
van Schie, E.C.M., 175
Vanhuele, M., 161, 206
Varan, D., 297
Vargo, S.L., 19
variants, 17, 18
Venkatesh, V., 154
Venkatraman, M.P., 114–115
Venkatraman, V., 298
vertical extension, 55
Villanueva, J., 258
voice, 179
Völckner, F., 59

Wadhwa, M., 214

Wakefield, K.L., 206, 247
Wangenheim, F.v., 192, 263, 265
Warlop, L., 162
Watts, D.J., 268–270, 272
Webb, T.L., 146
Weber, M., 173
Wedel, M., 223
Wee, C.H., 227
Weigel, R.H., 141
Weiner, B., 185–186, 196
Wellan, D.M., 78
Wernerfelt, B., 194, 207
Westbrook, R.A., 188
White, K.M., 146
Whyte, W.H., 251
Wicker, A.W., 139, 141, 246–247
Wilkie, W.L., 7
Williams, L.G., 12
Williams, P., 121
Wilson, W.R., 263
Winchester, M., 42–43, 263
Winer, R.S., 205
Wirtz, J., 244
Wolff, J.A., 144
woman tax, 212
Wood, L.A., 289–290, 293
Woodlock, J.W., 111
word of mouth (WOM)
age and, 127–128, 128
applications of reserch on, 271–273, 271
complaining and, 33, 194–195
impact of, 266–268, 268, 279–280
Internet and, 251–253, 295
measures of, 251–257
nature of, 249–251
occurrence of, 21, 22, 257–265, 258, 260, 261, 264, 269, 269
social influence and, 288
social networks and, 268–271, 270
Wright, M.J.
on effective advertising, 290–292, 297–298
on market dynamics, 95, 96, 100, 104, 109, 110, 111
on purchase intentions, 143
on stationary markets, 82
Wrigley, N., 237, 240

Yalch, R., 244

Yi, Y., 194

Zajonc, R.B., 159–160, 163

Zaltman, G., 193–194
Zeithaml, V.A., 191, 207
Zettelmeyer, F., 15
Zhang, H., 276
Zhang, J., 219, 223
Zhou, Y., 125
Zielske, H., 283
Zufryden, F.S., 95