Business Analyst

A business analyst uses her knowledge and experience, in

combination with the insights generated by a data analyst
to make decisions that affect the business. They work within
a business or organization to identify and implement
improvements to help a company achieve its goals.

The distinction between Data Analyst and

Business Analyst
Data analysts and business analysts both help drive data-
driven decision-making in their organizations. Data analysts
tend to work more closely with the data itself, while
business analysts tend to be more involved in addressing
business needs and recommending solutions.

Data-Driven Decision Making

Data-driven decision-making means making decisions that
are supported by data rather than decisions based on
observations, gut, or even instinct. With the advancement
of technology, the data available to organizations has
become tremendous. This abundance of data provides
organizations several opportunities to understand their
performance to a greater degree, optimize their
performance and leverage it for growth.
In simple terms, the process is – collecting, extracting,
formatting, and analyzing insights.

Example

Let’s consider a leading online video streaming (OTT)

platform with several web series under its belt – some
successful and some not. The platform is trying to decide
which theme to make a new series on. A big director
approaches the platform with an out-of-the-box idea that he
knows in his gut is a blockbuster idea. The platform
assumes that the director's past success is sufficient proof
to back the project. A huge budget is allocated to produce
and market the series.
The series is made and subsequently launched, and to
everyone’s surprise, its viewership is much below expected.
The response is underwhelming, and the platform is likely to
make losses.
How could the platform have avoided losses? Could the
platform have used data points available to it to make an
informed, data-based decision that has more weight than
just a gut feeling?
Of course, yes. And this is exactly what more and more OTT
platforms are doing.
There are various factors that the platform could have relied
on, which are not limited to; but include analyzing similar
series made in the past, the director & actors’ past track
record, and currently trending themes in the OTT space
among others. The likeliness of the show underperforming
would have been predictable before the platform took it on.

Advantages of using Data in the decision-

making process
 Risk assessment - As often data-enabled insights
help us identify that carpet bombs that can be
attached to a project that we’re working on. By this,
what I mean is that often when you are working on a
project, there are certain risks that you know about and
you do whatever is within your means to mitigate
them. However, there are certain risks that you do not
know about, and often – using data insights in the
decision-making process; helps you identify these risks.
 Trend spotting and predicting outcomes - Data
insights can make it easier for you to gauge trends,
 which can help you predict outcomes. You can think
about the failed web series example that we spoke
about earlier to understand this more in detail. Taking
a more practical view of that example, since the OTT
platform commissioned the idea based on the
credibility of the director, let’s say that even data
suggests the same and back his or her credibility and
past track record. When we look at the actors’ track
records and trending themes in the OTT space, here is
where data could paint a different picture and there
could be a chance that the actors, who were part of the
series, could be more successful in genres that were
different from this particular genre. Or there could also
be a chance that even though the content theme that
the web series was based on is in trend in the OTT
space, a plethora of content with the same theme on
rival OTT platforms would have added a lot of
competition to it and hence, this could be a red flag
that could be suggested by data. Hence, as we can see
from this; trend spotting becomes much easier with
data and as a result taking decisions through it, by
using it as a tool along with your knowledge;
experience and understanding can enhance the
precision of your decisions.
 Gauge customer satisfaction - All organizations
strive for customer satisfaction. They may or may not
achieve it. But it is definitely what they strive for. When
we speak about customer satisfaction, a key challenge
is – how do you gauge customer satisfaction? This is
where data plays a key role. To quote an example, let’s
look at FMCG as a domain. In FMCG, customer loyalty is
a key measure of customer satisfaction; and often this
is defined by repeat purchases by a customer. If we
look at the online sales channel, it is often easy to
identify repeat purchases as the company’s website or
the e-commerce platform through which it is selling its
product; often maintains a record of the details of its
 customers. But what about offline channels? Often, in
India; offline channels like Kirana stores do not keep
individual customer records of their purchases and the
products they have purchased, and hence, it's more
difficult to measure customer satisfaction. As a result,
marketers have developed a parameter of ‘same-store
sales growth to gauge customer satisfaction in offline
retail channels. Through this example, we can see how
data measurement helps in gauging customer
satisfaction through both online and offline mediums.
 Enables innovation - Often as a Business Analyst,
you have to make decisions on both the market and
the product. When we speak about the product, one of
your key inputs would be in new product development.
More often than not, the success or failure of a product
depends on the market. Here is where data plays a
critical role in the decision-making process. Using data
and with the help of a data analyst, you would be able
to study the market – its tastes and preferences and
work backwards to develop the product. Often, this
reverse engineering process of product development,
when done correctly with data is something that helps
in ensuring the success of a new product.
 And lastly, data-backed decisions help improve both
external and internal processes in an organization.
External processes include interactions of an
organization with external stakeholders like its
customers, vendors, and distribution & promotion
partners among others; while internal processes
include interactions of an organization with internal
stakeholders, predominantly – its employees.

DATA
Broadly, there are two kinds of data – structured &
unstructured data.
Structured data is clearly defined data types with patterns
that make them easily searchable. In simpler terms, when a
layman looks at structured data, she can easily make out
that this is data.
Structured data typically contains data types that are
combined in a way to make them easy to search for in their
data set. This means that structured data is easily
detectable via search because it is highly organized
information. The image on your screen is an example of
structured data. It's neatly segregated into fixed fields. It’s
the data that most of us are used to and like working with to
analyze largely quantitative problems.

Unstructured data is comprised of data that is usually not

as easily searchable. Think about an image or an audio or
perhaps a video. When you think about them, they do not
essentially seem like data or data types. However, a lot of
data can be extracted from them, and hence this potential
and possibility of extracting data make them a part of
unstructured data.
Unstructured data, on the other hand, makes the capability
to search much more difficult.
Unstructured data, as evident from its very name, does not
conform neatly to a spreadsheet but may have its internal
structure. It includes everything outside the boundaries of
structured data.
While it is unruly, it is also incredibly valuable—unstructured
data has the potential to depict a complex web of
information that offers strong clues about future outcomes.

Unstructured data analysis is a crucial part of the data

analytics process. Think of customer web chats, for
example, a platform where customers commonly dish out
their complaints and resolve queries. When analyzed as a
whole, this web chat data can help guide companies on
what to prioritize resolving or what aspect of the product is
driving the most interest. Or social media data, which can
signal customer buying trends before they even start
searching for a product. If structured data could be
considered a company’s backbone, unstructured data is its
competitive edge.
Example:
One interesting modern-day example of unstructured data
is the use of it through web-scraping, which often leads to
sentiment analysis, which helps companies gather insights
on their products and services. Often, companies use web
scraping as a method to analyze social media
conversations. Through this method, a program is run on a
social media site, which searches for keywords related to
the company’s products and services. Once the keywords
bring out the relevant conversations, the program attempts
to classify the conversation to find the sentiment that is
present in it.
The sentiment can be classified as positive, negative or
neutral, or even something else – as per the requirement of
the company and the project. To lead to this sentiment, the
program has searched for certain grammatical patterns and
keywords within the conversations that it has derived from
social media sites.
By using such methods, companies are using unstructured
data to get better feedback and insights for their products
and services. Sentiment analysis helps companies narrow
down the negative conversations around their products and
then analyzes them to improve the customer experience.

Semi-Structured
Semi-structured data is a form of structured data that does
not conform to the tabular structure of data but does
contain tags or other markers to differentiate elements and
enforce hierarchies within the data.
Email messages are a good example of semi-structured
data. While the actual content of the email is unstructured,
this format does contain structured data such as the name
and email address of the sender and recipient, and the time
sent among others.
Another example is a digital photograph. The image itself is
unstructured, but if the photo was taken on a smartphone,
for example, it would be date and time stamped, geo-
tagged, and would have a device ID. Once stored, the photo
could also be given tags that would provide a structure,
such as ‘dog’ or ‘pet.’
Making Decisions
As a business analyst, your primary job would be to make
decisions based on your knowledge, understanding, and
insights generated through data analytics.
Hence, it becomes really important for you to understand
data analytics as an understanding of that domain can act
as a competitive edge in your decision-making.

Data Analytics
Data analytics is the extraction of insights from data using
statistical techniques and technologies.

A big challenge that is often faced by data analytics teams

is the availability of quality data, which they can work with.
Therefore, before moving on to data analytics, let’s see an
example that deals with data collection and data cleaning.
Example
Let’s take the help of a case study to understand this more
in detail. Let’s say that an OTT platform, XYZ is looking to
launch a Spanish series and wants to predict its response in
New York.
It forms a research team, which decides that they want to
conduct a primary study of 1000 respondents, 500 of whom
are Spanish speakers and TV show watchers; while the
other 500 do not speak Spanish but watch Spanish TV
shows.
They prepare a detailed questionnaire consisting of both
objectives as well as subjective questions that they want all
respondents to answer.
The team splits itself, with some members going to Hispanic
neighbourhoods; while others going to non-Hispanic
neighbourhoods to conduct the interviews.
Now, how many surveys do you think the team needs to
conduct to complete the data collection process for this
research?

The exact answer cannot be predicted. However, we’re

pretty sure that it would be more than 1000.

The team going to Hispanic neighbourhoods may share the

questionnaire with respondents, who speak Spanish but do
not watch TV in that language.
The team going to non-Hispanic neighbourhoods will
definitely share the questionnaire with respondents, who do
not speak Spanish and do not watch TV in that language.
To get to a defined target of 1000 respondents, with 500 of
them speaking Spanish and watching TV in that language,
while 500 of them not speaking Spanish and watching TV in
that language, there is a chance that the team may have to
conduct 2000 – 2500 surveys.

The process of going out and conducting interviews to

generate usable data is known as data collection.
The process of going from 2000 or 2500 surveys that the
team may have had to conduct as part of data collection to
1000 relevant ones is known as data cleaning. In this
process, the team is going from a large data set to a
relevant data set that fits the objective of its research.

Types of Data Analytics

There are 4 predominant types of data analytics -
Descriptive, Diagnostic, Predictive & Prescriptive.

Descriptive analytics
Descriptive analytics answers the question: What has
happened and what is happening right now?
Descriptive analytics uses historical and current data from
multiple sources to describe the present state by identifying
trends and patterns. Descriptive analytics uses basic
arithmetic like average, sum, and percentage changes
among others. Many times in data analysis, descriptive
analytics is the first step that helps organizations
understand the facts of what has already happened.
Example:

An analyst looking at a large database of account holders of

a bank can extract information about the average incomes
of account holders, their average age, their average
withdrawals, and their frequency of withdrawals among
many other such parameters that can be defined by basic
arithmetic calculations.
Another example could be a manufacturing unit, where an
analyst can get historical insights regarding the company’s
month-wise production, production per worker, the average
wage per worker, the ratio of the value of raw materials
purchased per month, and sales among other such
parameters.
As we can see, in both the examples of descriptive
analytics, we can know what has happened earlier and how
the same metrics are performing right now.

Diagnostics Analysis
Diagnostic analytics answers the question: Why did this
happen?
Diagnostic analytics uses data, which is often built on
descriptive analytics to discover the reasons or causes for
past performance. It is oftentimes referred to as root cause
analysis.
Example:
Let’s say that an unusually high number of people come to
the emergency room of a hospital complaining of similar
symptoms. After testing them, it is found that about 70% of
them have Disease A. Descriptive analytics helps us get to
the final number of people having Disease A. However, it is
Diagnostic analysis that can help to make correlations
between the symptoms and determine what symptoms
point to Disease A.

Predictive Analysis
Predictive analytics answers the question: What is likely to
happen in the future?
Predictive analytics applies techniques such as statistical
modelling and forecasting to the output of descriptive and
diagnostic analytics to make predictions about future
outcomes. Predictive analytics is often considered a type of
“advanced analytics,” and frequently depends on
technology.
Example:

Whenever an individual applies for a loan, the bank or

financial services company where she has applied, asks her
for documents like her Aadhar Card, and PAN Card among
others. The reason for this is that they want to access the
individual’s CIBIL, which is a credit score, and along with it
comes the credit history of the individual. Through the CIBIL
report, the bank can access various aspects of the
individual’s credit history, which include: The types of credit
instruments held by the individual – like active credit cards,
past credit cards, active loans, and past loans among others
1. The types of loans given to the individual – like
whether the individual has accessed an auto
loan, home loan, personal loan, consumer loan,
education loan, or say a loan against property.
2. The loans that the individual has applied for –
you would not believe it but even if you apply for
a loan that does not get approved, it still shows
up on your CIBIL report and affects your credit
score.
3. The repayment history of the individual – which
includes insights on whether the individual has
paid her EMIs on time, does the individual pay
her credit card on time, or whether has there
ever been a delay in EMI, interest, or credit card
payment.
The core idea behind accessing an individual’s CIBIL is to
predict the likelihood of the individual defaulting on loan
repayments in the future.

Prescriptive Analysis
Prescriptive analytics answers the question: What do we
need to do?
Prescriptive analytics is a type of advanced analytics that
involves the application of testing and other techniques to
recommend specific solutions that will deliver desired
outcomes. In business, prescriptive analytics uses machine
learning, business rules, and algorithms.
Example:

Travel websites use prescriptive analytics by going through

several iterations and combinations of factors such as travel
patterns, conditions, and consumer demographics to
optimize their pricing and sales.
An interesting example of this could be in the hospitality
industry. Using a prescriptive analytics model, a luxury hotel
may find that during the off-season, increasing their room
tariffs and offering free additional services such as free spa
appointments, and a couple of complimentary meals is
likely to boost bookings compared to their initial strategy of
lower tariffs with few additional services.