MARKET RESEARCH

IN PRACTICE

TRAN LIEN PHUONG, DEC. 2019

Marketing Framework

Strategi Creative Managin

Potentia
c Excellen g&
l of
Marketi ce Steering
Position
ng
ing
Plan

U&A Concept/ Ad Tracking

Studies Volume Testing
Testing
 Implementation Decisions
Business Objectives
Consumer action objectives

Positioning strategy
Advertising strategy
Creative execution
Media strategy and
placement
Role of research

Obtain

information
Mine the

information
Obtain

understanding
Elicit Insights
Empower
clients to take
decisions
Research Map
Product CommercialiSation Roadmap – WHEN
RESEARCH
Discovery / Phase I Phase II Phase III Lau Post launch
preclinical Demand Forecast & production planning nch management

Clinical trials Planning & Assessment Launch

strategy
Commercialisati Positioning & Com. Operation
on Planning & Strategy strategy
Assessment Communication
-Needs segmentation testing
- Attributes &
Opport. Assessment
& Revenue tradeoffs U&A, trends assessment
forecasting - Channels

Risk/profitability Assessment

Competitive Benchmarking Distribution assessment

Tracking /
monitoring
Price assessment Pricing study
Sales force
tracking/
Portfolio analysis & monitoring
Planning

Life cycle planning (New opportunity, gap analysis) and Implementation

RESEARCH PROCESS
Guidelines & Procedure
Marketing research process flowchart
Selection of
Problem sample
Problem design
discovery and
discovery Sampling
definition

Selection of Probability Non-probability

exploratory research
technique

Secondary Data Collection of

Experience Pilot gathering data
(historical) Case Study
survey study (FIELD WORK)
data
Data
processing Editing and
Problem definition & analysis Coding
(research objectives) data

Data
Selection of processing
Research Design
basic research
/Proposal
method
Conclusions
Interpretation
and Report of Findings

EXPERIMENT SURVEY SECONDARY

OBSERVATION
Laboratory/ FieldInterview/questionnaire DATA STUDY
Report
Problem Definition Stage
• Identify the Problem is the first step towards its solution.
• Problem Definition Stage:
• the stage in which management seeks to identify a clear-cut
statement of the problem or opportunities.
• a problem definition error or omission is likely to be a costly
mistake that cannot be corrected in later stages of the
process.
Research Design Stage
• Research Design stage:
• the stage in which the researcher determines a framework
for the research plan of action by selecting a basic
research method.
• Research Design:
• A master plan that specifies the methods and procedures for
collecting and analyzing needed information.
• The objectives of the study are included to ensure
that the information collected is appropriate for
solving the problem.
• Sources of information, design technique, sampling
methodology, schedule and cost of the research are
also determined.
Data Gathering Stage
• The stage in which the data is collected – by humans or
machines (programming)
• Data # Information
• Gathering data process: 2 phases
1 - pre-testing / pilot
2 - main study / actual field work
The ACTUAL PROCESSES
ACTUAL WORKINGS OF RESEARCH
• THE FIELDWORK PROCESS

• QUALITY CONTROL & ITS VALUE

• SELECTING YOUR TARGET AUDIENCE

• A SAMPLING BLOCK

• THE CONTACT SHEET

Fieldwork Procedure
1. TRANSLATION OF QUESTIONNAIRE INTO VIETNAMESE (AGENCY)
2. BACK TRANSLATION OF QUESTIONNAIRE INTO ENGLISH (AGENCY – NON-PROJECT
TEAM)
3. CONFIRMATION OF FINAL VERSION OF QUESTIONNAIRE (WITH CLIENT)
4. SUPERVISORS BRIEFING (AGENCY)
5. INTERVIEWER TRAINING (AGENCY)
6. DRY RUN (INTERVIEWER INTERVIEWS SUPERVISOR TO CHECK WORDING,
QUESTIONNING, STYLE, TIMING, LOGIC)
7. PILOT TEST (INTERVIEWER INTERVIEWS SUPERVISOR RANDOM RESPONDENT F-TO-F,
TO TEST TIMING, STYLE, LOGIC, COMPREHENSION)
8. INTERVIEWER SELECTION (SELECT BEST INTERVIEWERS)
9. START INTERVIEWS FOR 1 DAY
10. STOP INTERVIEWS, CARRY- OUT 100% QUALITY CONTROL
11. PROBLEM?  STOP FIELDWORK AND RE-BRIEF
Quality Control Procedure
1. INTERVIEWER SCREENING VIA PILOT TEST & DRY RUN
2. SELECTS MOST QUALIFIED INTERVIEWERS
3. 100% QUALITY CONTROL AFTER 1 St. DAY OF FIELDWORK
4. 100% CHECK FOR LOGIC, COMPLETION & TIMING
5. 33% PHYSICAL BACK-CHECK SELECTED RANDOMLY FROM COMPLETED INTERVIEWS ON SAMPLING METHOD
6. IF INHERENT PROBLEM DISCOVERED, 100% BACK-CHECK ON ALL INTERVIEWERS COMPLETED
7. CANCELLATION OF ALL DISQUALIFIED INTERVIEWS
8. BACK CHECK INCLUDES :
• Logic check
• Interview time
• Screening criteria
• Starting point
• Sampling interval
• Sampling direction
• Questionnaire verification
• Respondent demographics
• Questionnaire completion
• Delivery of incentive

REDO PROJECT IF CLIENT IS UNSATISFIED

SAMPLE SELECTION PROCEDURE
1. Define target respondents 8. Left shoulder to houses, always turning
left and entering all motorcycle access
2. Determine number of starting points to streets ( 2 meters or wider)
increase representation
9. Marking contact sheet for refusal and
3. Divide city into population size by wards reason
4. Computer generated random selection of 10.Drawing map of interviewed area for
representative wards quality control
5. Allocation of interviews by representative 11.Making return interview date if
interviews by population size average respondent not home
6. Allocation of number of interviews by 12.Total of 3 attempted contacts per
block (# of starting points - number of respondent who is not home
interviews = # of interviews per block)
1 block = 100 to 120 living quarters 13.Selecting sample a home on either side
of the unable to contact home
7. Selection of interval (# of interviews per
block x block size)
SAMPLING BLOCK IMAGE
Shop with living place Gas Station
1st SP

SUPER MARKET
SCHOOL

HOTEL
Next SP Office Building

Apartment
Sampling Selection Stage
Random Sampling Procedure:
• Sampling stage: the researcher determine • Define and determine the number of Primary
• who is to be sampled: specifying the target population Sampling Units (PSU) to cover. Prepare the
is a crucial aspect of the sampling plan sampling frame. For e.g. in Jakarta, the PSU will
be Kelurahan (village). In each village, 10
• how large a sample is needed  Sample size households will be covered.
• how sampling units will be selected
(e.g. For N=300, we need to cover 30 villages)
Number of sample size is determined by Yamane’s approach below. • List out specific Sampling units to cover.
N = total target population • For example, there are 28,957 villagese in
e = precision level Jakarta. In determining the specific village to
cover, we go by interval.
n = sample size

(e.g. For N=300, interval = 28,957/30 = 965)

• there are two basic sampling techniques:
• Define a random start then skip by interval
• probability & non-probability
(e.g. i=965).
• simple random sample
e.g. Random start is 200th
• stratified samples village, other village to cover are: 1165 th, …28th,
• quota samples
• cluster samples 185th in the sampling frame
• judgmental samples • For each village, household to knock should go in a
clockwise direction, skipping 3 houses after
• Stratified random sampling which is a random sample of
every successful or unsuccessful interviews until
a population in which the population is first divided into the 10th household is reached.
district subpopulations, or strata, and random samples are
then taken separately from each stratum.
• Quota is often set by demographics, psychographics,
DO’ES & DON’TS IN INTERVIEWING
• Never cross a major street or intersection
• Never enter a different district or ward to interview
• Never interview a friend, colleague or relative
• If more than 1 family lives in one living quarter, interview according to interval (top
floor to bottom floor)
MARKETING RESEARCH DEFINITIONS
 Apartments: begin interviews on top floor on far left corner maintaining interval and working your away down to the ground floor
 Non - residential structure: Any building not used as private residence (i.e. hospitals, hotels, gas stations, offices, both government
and private, schools.
 Non - residential structures: If interval includes NRS, indicate type of structure on contact sheet and continue with interval
 Living quarter: Any structure with its own separate entrance, bath room, cooking facilities
 Resident: Any person/persons who sleep & eat at that residence 4 or more days a week.
 Family unit: Any persons, regardless of relations, who eat, sleep and live together 4 o more days per week and share the
costs of food and rent (does not include dependents)
 Sampling interval: A sequence of houses selected to maintain non - bias contact with respondents. the sampling interval
largely depends on the number of interviews allocated per block and the representation desired.
DATA COLLECTION METHOD
• OBSERVATION
• FOCUS GROUP DISCUSSION / IN-LAB INTERVIEW
• CENTRAL LOCATION TEST (CLT), HALL TEST
• INTERCEPT INTERVIEW, STREET INTERCEPT
• MYSTERY SHOPPING
• ETHNOGRAPHY / CONSUMERS SAFARI / IMMERSION
• Face-to face interview
• CAPI (Computer-Aided Personal Interview)
• CATI (Compute-Aided Interview)
• PAPI (Paper-and-Pen Interview)
• Internet/online surveys
• Home-Use test / Product Placement Test
• Telephone Interview
Operations processes: Typical ad-
hoc studiesClient
requirements
By
Coding research
Materials By research processes team
preparation team
Utilize code frame
Finalize questionnaire, from similar study
showcard, product concept,
etc. Brainstorm Check verbatim to design
Communicat and customize code frame
Internal briefing between with internal
e with client
OP managers sup. and source
research team
Briefing between
supervisors and
interviewers, monitored
by client, researchers and
OP manager
Pilot test On-going FW Quality
processes assurance
Weekly progress Senior Involvement in all
Soft launch report projects in every
Partial data department
Full launch processing, if it is 100% Logic Check of
required by client all questionnaires
Data entry Supervisor accompanied
interviews of each 30% call back for each
interviewer interviewer by external
Data QC
process
Data collection & processing: typical ad-hoc studies

Softwar Data Entry and DP Quality

e processing assurance
QPSMR Design data entry Parallel data entry to
Desktop template check error of punchers
Reporter
100% “sense check” which
SPSS &
Data processing is the solution to check
Other
consistency of data of each
advance
interviewer
d
software

Examples of statistical
Factors techniques Cluster
analysis Analysis
Multiple
regressio Max
Test
n Double Diff
signification of
differentiation Jeopardy
& correlation
Data Collection and Quality Control
• Training of Interviewers
• Field Briefing
• Mock interviews done immediately after the field briefing in office
• Pilot interviews observed by Field Supervisor and Researcher
• Monitoring
• Individual quota sheet are provided and will be used by interviewers to keep track of their
quota achievements. This will be submitted during progress reporting.
• Regular progress reporting from interviewers
• Log sheets for call status on each respondent are also filled-in and submitted (includes time of
visit, status for each callbacks made, etc.)
• Quality Assurance
• 100% Editing and walk-through of hardcopy questionnaires
• 30% Back-Check of each interviewer’s work (either through re-visit or callback)
• Field observation by researchers within the first 3 days of fieldwork to monitor if the survey is
being administered properly and if the protocols are being followed (for new types of studies as
well as on a random basis for repeat studies).
DATA PROCESSING & QUALITY
CONTROL
• 100% data cleaning
• 100% final raw data checking (by both DP and Research
Executives)
• 10% double-punch for data entry
• Testing of data entry format. Pre-tested questionnaires will be
used for testing. Testing is done to ensure that data entry
program is accurate, complete, and all logic and skipping checks
are in place.
• Briefing of punchers on the data entry program
• Strict checks to minimize data entry errors using QPSMR analysis
software
Conclusions and Report Preparation
stage
• Researcher interprets information and draws conclusions so
they can be communicated to the clients/the decision
makers/the project owners.
• Research Report
• communicate the research findings
• other purposes of research report
• providing historical documents that will be a source of
record for later use, such as repeating the survey, or
• providing a basis for building on the previous survey
findings.
 Data Processing and Analysis Stage
• Data processing and analyzing stage:
• researcher performs several interrelated procedures to convert the data into a
Editing format that will answer management’s questions.
• Editing and Coding: this is part of the stage
• Editing involves checking the data collection forms for omissions, legibility, and
consistency in classification.
Coding • The editing process corrects problems such as interviewers errors before
the data are key punched onto the computer.
• Coding
• interpreting, categorizing, recording and transferring the data storage
media into numeric CODES.
Data Entry • Data Processing:
• Responses which are transferred into numeric codes are to be key-punched
onto computer for processing.

Analysis • Analysis: is the application of logic to understand the data that have been
gathered about a subject.
• Determining consistent patterns
• summarizing the relevant details revealed in the investigation
DATA ANALYSIS - Quantitative
studies
Analysis plan, scrutiny instruction

Questionnaires from field

Open ended coding, scrutiny

Data entry (if off-line)

Data validation and cleaning

Tabulation Tables

Visualize output for report writing

Logical Checks Basic Scrutiny
• If there is internal • Skip instructions have been followed.
consistency in the data. • If a section has to be administered only to
• Awareness should be there triallists of a particular brand – then whether
the instruction has been followed.
for a brand that has been
tried. • Brands have been coded correctly.
• A brand that has high
preference should not have • Totals add up to the specified number.
• Percentages add up to 100.
a low overall score.
• In ranking questions, ranks are not
repeated.
• Multiple responses are not coded in single
response questions.
• The range of response is within the
specified scale.
• The values should be between 0 and 10 for a
marks on 10 question.
CODING – Code list Development
 For additional brands (for example:
Automotive manufacturer brands such as
Brand code list Mercedez-Benz, Fords, Mazda, BMW, Audi,
Toyota, Nissan, Mitsubishi, Subaru, Vinfast,
etc)
 For brand related response captured
verbatim.
 Verbatim
o Salience,responses  clubbing similar
Open-ended unaided awareness.
responses together, then putting related
responses descriptions of the benefit under one group
(net).
 For example: Perfume (net)
 Good perfume
 Long lasting perfume
 Mild perfume
Open Ended Codelist - An Example (2) Codelist
Reasons for preferring one motorcycle brand vis-à-vis
100 POWER another
101 Good power
• Fuel tank is big 102 Faster/speed
103 It is 110 cc
• Seat is big The Verbatim responses 110 MILEAGE
• Good mileage 111 Good mileage
• 120 FUEL TANK
Fuel tank shape is good
121 Fuel tank is big/has more fuel capacity
• Headlight is good 122 Fuel tank is bigger than other bikes
• Headlight is round 130 SAFETY
131 Good brakes
• Bigger fuel tank than other bikes 132 Will not slip/good road grip
• Looks like a racing bike 140 RUGGESNESS
• 141 Strong bike
It is stylish
150 RIDING COMFORT
• Has good power 151 Is easier to ride/handle
• Will not slip 152 Seat is big
160 MANUFACTURER REPUTATION
• Has good brakes
161 Good company/Honda is a well known brand
• Has more fuel capacity 200 LOOKS AND APPEARANCE
• Strong bike 201 GENERAL LOOKS AND APPEARANCE
202 Is stylish
• Broad shape of the rear is good 203 Looks from the rear is good
• Good company 204 Looks like a racing bike
• 210 HEADLAMP
Honda is a well known brand
211 Headlamp is good
• It is 110cc 212 Headlamp is round
• It is fast, more speed 220 FUEL TANK
221 Fuel tank shape is good
Analysis plan - The Key
Determinants
• Step 1: Specify the decisions to be made.

• Step 2: List of out & prioritise the information

needed to make each decision. They should be:
The decisions • Stated in marketing terms
The information • Grouped logically, e.g. objectives followed by sub-
to be taken objectives, if any
obtained • Phrased in question form
from research • Limited in number: 5 ± 2
• Step 3: Specify the data items (e.g. questions,
ratings, etc.) from the questionnaire or other
sources which will be used to provide each piece
of information stated in Step 2.

• Step 4: Develop an analysis plan, clearly stating

the statistical/analytical procedures, e.g. cross
tabs, summary statistics, significance tests,
Analysis plan multivariate procedures, etc. which will be used
to link/transofrm the data

• Step 5: Anticipate / visualise the output from

the planned analysis.
Key points to cover in an analysis
plan
The point to be covered The determinant
• The segments that will be
compared and contrasted,
• The segment by which • Segments of interest to
tables are required. client.
• Segments that could
behave/respond differently.
• The tables required • The decision that will be
taken from each table.
• If there are representation
• Weights to be used
skews to be corrected.

• The additional tools and • The inferences to be drawn.

• The amenability of the data
techniques to be used. to the technique.
• Reporting deadlines
• The time lines.
Some Commonly Examined
Segments
• Income segments
• Gender segments
• Centre segments
• Usage segments (heavy, medium, light)
• Brand usage segments
• sample size of the segments need to be significant (at least n=30+)
enough to draw meaning inferences.
The Tables request & tabulation
Questio Legend Base Addition
The table request n Table Table for the (are there any al
number number title row filters for
respondents) remarks
• Example:
• The question: could you tell me if there was anything in particular that you liked about this
product sample? If yes, what did you like about it? PROBE: anything else? Anything else?
RECORD VERBATIM.

Questio Legen
The table form Table d for Base
n Table title
number number the
row
Q1.a 1.1 Spontaneous likes about Aspec all
the product t liked
Q1.a Spontaneous likes about the product %
The table Q1.a Aspect liked …%
Base : All respondents
Some Examples
The • COUNTs:
 Cups of coffee consumed in a day.
question Number of
respondents
The  To find out how many cups of coffee are in a specific
purpose consumed in a day. category.
The table
• Or a simple
that will be  Frequency distribution and average. frequency
asked for distribution
The  To contrast coffee consumption between
purpose men and women.
 Both average and distribution.
The tables  But average will be used far more than
that will be distribution as it will be easier to
asked for
compare.
Data Collected
• Question : Number of cups • Basic count
of coffee drunk in a day.
Resp No No of cups Gender Number of cups
Count %
1 1 F drunk in a day
2 5 M  1 5 30
3 4 M  2 4 26
4 2 M  3 2 -
5 10 M  4 2 13
6 5 M  5 2 13
7 7 M  6 - -
8 2 F  7 1 7
9 2 F  8 - -
10 1 M
 9 - -
11 4 M
 10 1 7
12 1 F
Base : All respondents 15
13 1 F
14 1 F
The inference : Most drink less than
5 cups of coffee a day. In fact,
15 2 F
majority drink only one or two cups.
Crosstab
• Cross analysis of the data by segments.
• Income x age
• Gender x income
%
• An example Cups of coffee Male Female
consumed per day
• 1 13 57
• 2 13 43
• 3 - -
• 4 26 -
• 5 26 -
• 6 - -
• 7 13 -
• 8 - -
• 9 - -
• 10 13 -
Base : All respondents in the 8 7
respective segment

The inference : Women consume fewer cups of coffee in a day

compared to men.
Summarising The Data Weighting
• Weighted mean – the average
The  To correct for skews in
weighted by another variable.
purpose representation.
• Weighting by consumption of the
category.
When to  When aggregating data
• Weighting by value contribution.
weight across segments, leads to a
 A study has
distortion in been done across
inference.
• Median – The central point of the two centres – A and B.
distribution.  The sample size covered in
both centres is same.
• Mode – The value with the highest  However, the size of the
universe in city A is larger
frequency in the distribution. Example
than in city B.
The mean is the most commonly Will not weight if A and B are
•used
Variance – The
measure forextent of variability
comparisons. The in equally important markets for the
client. Or if the data is presented
modal values are also used, to
the data. separately by centres.
describe the table. Will weight if the aggregate data
needs to reflect the size of the
universe in the different markets.
The Types Of Weighting
 A weight is attached to each respondent so that
the total across respondents adds up to the size of
the universe.
Extrapolation  The users of brand A in the universe is 1000,00 people.
 And 45 were covered in the sample.
 The weight assigned to users of brand A will be 100,000/45
= 2222.

 The weight attached to each respondent is equal

to the proportion in the universe, of the segment
to which the respondent belongs.
Proportional  A study was conducted among 200 people, 100 from the
upper income group and 100 in the lower income group.
weighting  Lower income group people represent 70% of the
population then
 The weight to be assigned to every respondent in the lower
income group is 0.7.
Additional tools and techniques for data analysis

The technique When used

Significance tests • While comparing and contrasting
Regression and correlation • Relationship building
• Assessing importance of attributes
in choice
Factor analysis, • To summarise data
correspondence analysis
Cluster analysis • For segmentation
Analysis Plan

Should we go ahead with the new

 The key issue
formulation.
 The question Rating of current and new formulation
asked overall.
Differences in the (normalised) mean
 The analysis score for new Vs. current formulation.
required Tested for significance, using a one
tailed test.
The Key Diagnostic Tables
• Spontaneous likes

• Spontaneous dislikes

• Normalised rating of sample (on specific attributes)

• Distribution of distance from ideal (for ideal attributes)

• Distribution of ideal
To check position of sample
vis-à-vis ideal.
• Is the perfume stronger or
milder, to what extent.
To check what is it they like.
Do they want a strong perfume
or mild perfume.
An Example : Post Launch Evaluation Of a
Deodorant Brand
• The study perspective
• To assess the status of the brand six months post launch.
• And to obtain all relevant diagnosis.
• To guide future marketing strategy.

 Target group
 Females aged 18-35
 Currently using deodorants - have a pack open and in use in
the household.
 The information collected
 Status of brands
- Salience, awareness, trial, current usage.
 Importance of attributes in choice.
 Knowledge of distribution of our brand.
 Awareness and trial buildup – when got to know, when first
tried
 Reasons for trial, non trial, retention
 Image of our deodorant brand vis-à-vis competition
 Reach of media on which advertised
The Analysis Plan (1)

 The key question What is the status of our brand.

 The answer to the
From the questions asked on status of the brand.
question
 The tables to ask for Salience, awareness, trial, current usage.
Of those aware , how many tried the brand
 The additional (awareness - trial conversion).
analysis possible Of those who tried, how many are retained with the
brand (retention ratio)
A summary of status of our brand.
 Additional table To obtain at a glance, the status, specifically, across
segments.
The Analysis Plan (2)

The next question Why this status

Why some are not aware
 The information to look Why some are aware but have not tried
at Why some have tried but dropped out
Why some are retained with the brand
Those aware vs not aware
 The comparisons to
Triallists Vs aware not tried
make
Those retained Vs drop outs
 The relevant segments
Not aware, aware, aware not tried, triallists,
to ask for tables by (for
dropouts, those retained
the diagnostic section)
Q&A?