Segmentation and

Selections
Goyal said a separate fleet for “pure veg”
deliveries was launched because sometimes,
food gets spilt into the delivery boxes and
thus, the smell of the food is inevitably carried
to the next order. “For this reason, we had to
separate the fleet for veg orders,” he said.
Learning Objectives
• To understand the strategic aspects of segmentation
• To understand selection, Who to be approached in each segment
About Nykaa

• Nykaa is an Indian e-commerce company, founded by Falguni Nayar in

2012 and headquartered in Mumbai. Falguni Nayar, a former managing
director at Kotak Mahindra Capital Company, founded Nykaa
• An omnichannel model and began selling products apart from beauty.
• It retails over 2,000 brands and 200,000 products across its platforms.
About Nykaa

• It follows an inventory-based model with warehouses in Mumbai, New Delhi,

Pune, Haryana, Kolkata and Bangalore
• it has an offline presence via 76 brick-and-mortar stores across the country.
• It has three offline store formats called Nykaa Luxe, Nykaa On Trend and
Nykaa beauty Kiosks.
• Nykaa has a series of in-house brands within beauty and fashion. Some of
them include:
Nykaa House of Brands – Nykaa Naturals, Nykaa Cosmetics, Kay Beauty
Nykaa Fashion – Nykd by Nykaa, 20 Dresses, RSVP, Mondano, Likha, Pipa
Bella
Apple Customer Segmentation
• Apple customer demographics include people aged 18 to 45.
• They are either single, married with no kids, or married with young
children or teens.
• The Apple target audience skews strongly female, with around 66%
female to 34% male customers.
• Among Apple primary customers, you'll find less older or retired
individuals
Define segmentation?

• Segmentation is the division of customers into various, distinct,

homogeneous groups who it can be desirable to approach in a specific way
• Differences exist between the groups.
• Customers within a segment have communication or purchasing behaviour
and/or needs and wants in common
• Segmenting not only has strategic objectives, but also plays a role in the
selection of people and organisations we would like to approach with a
particular marketing campaign
• The topic of selection will also be examined: Who will be approached in
which segment during a marketing campaign?
Chapter addresses the following questions
• How can we define segmentation?
• What are segmentation criteria?
• Segmentation for the benefit of strategic marketing issues
• The selection of people and organisations for marketing campaigns
• What are the research techniques?
 Segmentation Study as Input for the
Formulation of the Marketing Strategy
• Segmentation criteria
• Segmentation criteria are the indicators used as the basis to identify or assign
the groups.
• In consumer markets(B2C), the following are used as criteria:
• Geographical characteristics, Postcode
• Demographic characteristics, age or gender
• Socio-economic data, income, social class or education;
• Behaviour, purchase or communication behavious
• Psychographic characteristics, lifestyle, norms or values
• Buying motives and purchase considerations.
 Segmentation Study as Input for the
Formulation of the Marketing Strategy (Cont.)
• Bonoma and Shapiro distinguish five categories of segmentation criteria for B2B:
1. Demographic factors,
2. Operating variables,
3. Purchasing approaches,
4. Situational factors,
5. Personal characteristics.
• Bruhn makes a distinction between the normal segmentation criteria which he
characterises as exogenous, and endogenous criteria such as customer value,
customer satisfaction, recommendation behaviour, and the likelihood of
retaining the customer.
Exterior(exogenous) and Interior(endogenous)
of the market: B2B Factors
• Connecting the criteria within the framework of relationship marketing
• At the top of this list(5 Criteria) are the criteria that describe the exterior
of the market; this involves data which can be easily obtained.
• The further down the list we travel, the more we zoom in on the interior
of organisations, until finally we reach the last level and the people with
whom we come into contact in the organisation.

The organisation cannot use policy to influence the first category, but
can do so with the second category, for instance, it can influence
customer satisfaction.
Segmentation Criteria for B2B
 Segmentation Study as Input for the
Formulation of the Marketing Strategy (Cont.)
• Segmentation technique
• Markets may be segmented in a large number of ways. Choices must be
made regarding the number and the specific segmentation criteria to be
used.
• The five guidelines to go on in this segmentation process are:
1. Measurability
2. Substantial
3. Accessibility
4. Differentiable
5. Actionable
Five guidelines segmentation process: Requirements
Cluster and Discriminant Analyses: Segmentation
• Cluster and discriminant analyses are often used in market research for
segmenting customer.
• Cluster analysis is a data analysis technique that explores the naturally
occurring groups within a data set known as clusters.
• K-means cluster is a method to quickly cluster large data sets.
• Individual subjects are assigned to the group which is ‘closest’ to them in
terms of distance
• The researcher define the number of clusters in advance. This is useful to
test different models with a different assumed number of clusters
A group of data points would comprise together to form a cluster
in which all the objects would belong to the same group.
Dendrogram Method: Hierarchical Clustering
Why Cluster Analysis?
• Segmenting the customers
• Product positioning
• New product development
• Test marketing
• Customer groups into A, and B
• Tailoring marketing strategies to each cluster
Discriminant Analysis
• Discriminant Analysis refers to a statistical technique that may determine group
membership based on independent variables.
• The primary function of this technique is to assign each observation to a
particular group or category according to the data’s independent characteristics.
Discriminant Analysis
• Discriminant analysis is a technique that is used by the researcher to
analyze the research data when the dependent variable is categorical and
the independent variable is interval in nature.
• The term categorical variable means that the dependent variable is
divided into a number of categories.
• For example, if you have data on the age, income, gender, and purchase
history of your customers, you can use discriminant analysis to find out
which variables are the most important in differentiating between loyal
and occasional customers, or between customers who prefer different
product categories.
• SBI: House Loan to be given or rejected. Credit score is one factor, but
many other factors can be considered to give or reject loan
 Segmentation Study as Input for the
Formulation of the Marketing Strategy (Cont.)
• Cluster and discriminant analyses are often used in market research
for segmenting customer.
• The goal of cluster analysis is to classify subjects as customers in relatively
homogeneous groups, or clusters.
• The first and probably most important step in a cluster analysis is the determination
of the segmentation variables.
• K-means cluster analysis researcher specifies the number of groups to be
distinguished beforehand.
• The determination of the number of segments to be distinguished is arbitrary.
• In general, the recommendation is to condense the analyses for a different number
of clusters. The solution that appeals most intuitively is usually the best one.
The first, and probably most important, step in a cluster analysis is to
determine the segmentation variables.
 Segmentation Study as Input for the
Formulation of the Marketing Strategy (Cont.)
• Performing a K-means cluster analysis requires the researcher to make (arbitrary)
decisions. For this reason, outcomes should be treated with caution.
• A discriminant analysis is a more solid technique. Executing the technique
requires advance knowledge of the groups to which the individual subjects belong.
• Discriminant analysis can also offer a good means of testing the cluster
solution(Next Slide)
• Profiling a segment
• External/Internal sources can supply data to enrich the profile.
• Qualitative research can also result in enrichment, considering the fact that it
provides a great deal of insight in general terms.
 Group Membership: Cluster Analysis &
Discriminant Analysis

Discriminant analysis is then used to develop a model used to allocate the observations to the classes. If, using the model,
the customers are assigned to the groups in the same way as occurred with the cluster analysis, the solution can be
considered reliable.
Developing marketing service concepts for each segment

• Segmentation is a research process that aids in the formulation of

marketing strategy.
• The goal of segmentation is to approach customer groups in a differentiated
manner so that they will become more satisfied and loyal and will spend more
with the supplier.
• A service concept can gain in quality if the segmentation research provides
clear insight into the customer profile.
• The product range, service, price and communication are determined
for each of the distinct (combinations of) customer groups.
• Who is the person or organisation behind the buyer and what motivates
them?
Segmentation Research used in Compiling the List: Selection
• The objective of tactical segmentation research is to provide input for the selection
question.
• Whom do we want to approach for a certain marketing campaign activity?

• Three Methods for Selection: RFM, CHAID and CART

• Recency Frequency Monetary (RFM)
• Behaviour observed in the past appears to be a better predictor of future
behaviour than pronouncements made by customers regarding their purchasing
intentions and attitudes.
• the RFM value model was developed to identify the most attractive prospects. By
focusing on the frequency and the most recent transaction date in addition to the
annual amount spent, better selections may be made and higher response
percentages can be achieved.
 Three Methods for Selection: RFM, CHAID and CART
• The determination of the score model is left more or less open:
• R: The last purchase date,
• F: The purchase frequency,
• M: The amount spent.
• Behaviour from the past is an accurate predictor of the response to marketing
campaigns. Technique employs data that are maintained in the most basic of
databases, namely transaction data.
• CLV is calculated for each customer

In the RFM model, an index is calculated on the

basis of scores customers receive for three aspects.
 RFM Analysis
• Recency: A high recency rate implies that the customer already has a good
enough impression of your brand to have paid a visit to you recently.
• You can figure out Recency in the customer base section by sorting
customers accordingly to the date of their last purchase.
• Frequency, meaning how often your customers buy from you. A high
frequency indicates that the customer likes your brand, your products and
services, so he or she often come back to you.
• To calculate the frequency of purchase, you need to divide the total
number of purchases by the number of months/days/years, etc.
• Monetary value, meaning total amount of purchases. A high level of this
indicator implies that the client likes to spend money at your shop.
RFM model, an index is calculated on the basis of scores for three aspects
RFM model index
• In order to calculate the index value, the separate scores must be weighted
for their importance.
• Recency is usually the most important and monetary value the least.
• The weights assigned might be 5 for last purchase date, 3 for frequency
and 2 for amount spent.
• Example of RFM on next slide
• The cumulative RFM score is presented, which is calculated from all the
purchases the customer makes in a year
• Cumulative Score: A-202, and B-79 score
 RFM Analysis: Benefits
RFM analysis will answer the following questions:
• Who are my best clients?
• Which of my clients are on the verge of being lost for my business?
• Who out of my clients can potentially be transferred to a more paying
segment?
• Which customers should to be retained by all means?
• Which of my clients is the most loyal to my brand?
• Which customer segment is responding best to current ad campaigns?
First, behaviour from the past is an accurate predictor of the response to marketing campaigns
Three Methods for Selection: RFM, CHAID and CART
• CHAID and CART
• Chi-squared Automated Interaction Detection (CHAID)
• A CHAID analysis produces a tree diagram, such as that shown in Figure 8.2.
• A CHAID analysis creates a categorisation within each group until further
significant division is no longer possible.
• CHAID can produce better results than RFM analysis.
• The response to a marketing campaign that is obtained from the best 25–50%
of the respondents from the database, according to CHAID, may be higher
than the response from the best 25–50% indicated by RFM.
• The CHAID software is normally unable to process more than 40 variables with a
maximum of 15 classes per variable
 The response to the marketing campaign

the travel time from the customer to the service provider

1.28 times better to marketing campaigns

which portion of the products purchased

length of membership

An index of 105 was reached

 Chi-squared Automated Interaction Detection (CHAID)
• At the top of the diagram, the response to the marketing campaign(s) is shown for the entire
customer database.
• The organisation here has 240,000 customers of which an average of 4.36 per cent responds to a
marketing campaign activity
• On the level below this, these customers are split according to the most discriminating significant
segmentation criterion.
• The travel time from the customer to the service provider; the better customers – 80,000 in
number – live closer than those who tend to produce less of a response
• Customers who do not have to travel more than 30 minutes respond 1.28 times better to
marketing campaigns
• Customers with a short travel time may best be divided up according to length of membership, and
• Customers with a longer travel time according to ‘folder share’ (which indicates which portion of
the products purchased had appeared in the folder).
• In the less attractive segments on the first level, there are still identifiable sub-segments which
none the less produce an above-average response to marketing campaign(s). An index of 105 was
reached.
 Three Methods for Selection: RFM, CHAID and CART
• Classification and Regression Trees (CART)
• Is a technique that is often compared with CHAID. CART is not characterised by
limitations for the number of variables and classes to be included.
• With CART, a binary decision tree is constructed (repeatedly with two branches).
• CART first splits the data set into two groups for each variable; segment A is separated
from segment B
• CART is capable of working with a ratio scale. 5, 11, any numbers
• The advantage offered by CHAID is that more than two segments may be
identified for each level, whereas CART always splits the database in two.
• Response gain chart and composing the list
• In order to produce the gain chart, the customer database is split into equal parts
with respect to size.
 Response gain chart and composing the list
• It is the marketer’s task to determine the cut-off point: Which portion of the database will
continue to be approached with campaigns and which will not?
• The objective of segmentation is to divide up the customer database into
homogeneous groups, yet it stops short of selecting the prospects in a
campaign.
• Data on the response to past campaigns may be used, but also the outcomes
from a test may prove to be useful.
• The decision on the composition of the list targeted for a campaign is not an
isolated or independent one, but instead will have to be made within the
context of the relationship policy.
• The effect of the campaign is not profitable on a transaction basis, whereas its
impact on the lifetime value is profitable.
Questions?