Steps in Segmentation

1
Steps in Segmentation

• 1. Articulate a strategic rationale for
  segmentation (i.e., why are we segmenting this
  market?).
• 2. Select a set of base segmentation variables
  most useful for achieving the strategic goals.
• 3. Select a method (criteria) for aggregating (or
  disaggregating) customers into segments.
• 4. Group customers into a particular segment.
• 5. Choose the segments that will best serve the
  firms strategy, given its capabilities and the
  likely reactions of competitors.

2
Need-Based Segmentation

• Factor analysis to reduce data before cluster
  analysis
• Cluster analysis to form segments
• Discriminant analysis to describe/reach the
  segments

3
Factor Analysis

• Segmentation studies rely on individual-level
  measurements on a number of relevant
  attributes/variables.
• Correlated attributes may make it difficult to
  analyze true segment structure.
• Factor analysis can be used to reduce the data
  from a large number of correlated variables to a
  much smaller set of independent underlying
  factors.
• Ex cleanness
• checkout ? Quality
• greeting

4
Cluster Analysis Procedures

• 1. Define a measure to assess the similarity of
  customers on the basis of their needs
• 2. Group customers with similar needs
• Hierarchical clustering (e.g., Single linkage,
  average linkage, and minimum variance) Wards
  method
• Partitioning methods (e.g., K-Means)
• 3. Select the number of segments using numeric
  criteria, and managerial judgment
• 4. Describe the needs of the selected segments
  (e.g., using cluster means which are
  significantly above or below overall means)

5
1. Similarity Measures

• Dependent upon the type of input data
• Nominal data (yes/no, male/female)
• Matching coefficient the similarity between two
  individuals is captured by the ratio of the
  number of matches to total possible matches.
• Ex Respondents are asked whether it is an
  essential feature for a cell-phone on 8 features
  If two respondents agree on the essentiality of 3
  features (they disagree on the other 5 features),
  then their matching coefficient is 3/8.
• Scaled data (perceived quality) distance-type
• Measures of similarity correlation coefficient
• Measures of dissimilarity Euclidean distance,
  absolute distance

6
2. Clustering Methods

• Hierarchical clustering build up or break down
  the data customer by customer
• e.g., Wards method
• How to join clusters single linkage, average
  linkage, and minimum variance methods
• Partitioning methods (e.g., K-means) break the
  data into a pre-specified number of segments and
  then reallocate or swap customers to improve some
  measure of effectiveness.

7
Graphical Illustration
a distance from member to cluster
center b distance from cluster I to cluster III
8
Single Linkage

• Distance Matrix
• Co1 Co2 Co3 Co4 Co5
• Company 1 0.00Company 2 1.49 0.00Company
  3 3.42 2.29 0.00Company 4 1.81 1.99 1.48 0.00C
  ompany 5 5.05 4.82 4.94 4.83 0.00

ResultingDendogram
1
2
3
Company
4
5
1
2
3
4
5
Distance
9
Minimum Variance

• First Stage A 2 B 5 C 9 D 10 E 15
• Second Stage AB 4.5 BD 12.5
• AC 24.5 BE 50.0
• AD 32.0 CD 0.5
• AE 84.5 CE 18.0
• BC 8.0 DE 12.5
• Third Stage CDA 38.0 CDB 14.0 CDE 20.66 AB
  5.0
• AE 85.0 BE 50.5
• Fourth Stage ABCD 41.0 ABE 93.17 CDE
  25.18
• Fifth Stage ABCDE 98.8

10
Minimum Variance (contd)
98.80
25.18
5.00
0.50
A
B
C
D
E
11
Interpreting Cluster Analysis

• 3. Select the appropriate number of clusters
• Are the clusters separated well from each other?
• Should we combine or separate the clusters?
• Can you come up with descriptive names for each
  cluster (e.g., professionals, techno-savvy,
  etc.)?
• 4. Segment the market independently of your
  ability to reach the segments (separately
  evaluate segmentation and discriminant analysis
  results).

12
Describing Clusters
Two Cluster Solution for PC Data Need-Based
Variables
1
Designer
Means of Variables
0
Business
1
size
power
office use
LAN
storage needs
color
periph.
wide connect.
budget
13
Discriminant Analysis

• Identify a set of observable variables that
  help understand how to reach and serve the needs
  of selected clusters.
• Use discriminant analysis to identify the
  underlying dimensions (functions) that maximally
  differentiate between the clusters.

14
Two-Group Discriminant Analysis
XXOXOOO XXXOXXOOOO
XXXXOOOXOOO XXOXXOXOOOO XXOXOOOOOOO
Price Sensitivity
X-segment
Need for Data Storage
O-segment
x high propensity to buy o low propensity
to buy
15
Interpreting Discriminant Analysis

• What proportion of the total variance in the
  descriptor data is explained by the statistically
  significant discriminant functions?
• Does the model have good predictability in each
  cluster?
• Hit rate the proportion of customers who are
  correctly assigned cluster membership by the
  discriminant analysis, according to the cluster
  analysis.
• Can you identify good descriptors to find
  differences between clusters? (i.e. examine
  correlations between discriminant functions and
  each descriptor variable).

16
Which Segments to Serve?
17
Segments Selection (GE Matrix)
E
Strong
Firms Competitive Position
B
Medium
D
A
C
Weak
Low
Average
High
Segment Attractiveness