Chapter–by–chapter aids: Chapter 4
Instructor’s Manual to Accompany Essentials of Marketing IV-4-3
DISCUSSION OF COMPUTER-AIDED PROBLEM 4: SEGMENTING CUSTOMERS
The questions for this problem are intended to deepen students’ understanding of ideas related to
segmenting, clustering, and related concepts. The spreadsheet for this problem is different from the ones
for most of the other problems because it is not oriented toward costs, revenue, and profits. Rather, for
this problem the spreadsheet values are the “inputs” and “results” for a (simplified) cluster analysis
technique. Because the “style” of this spreadsheet is different from some of the other spreadsheets, some
instructors may wish to wait and use it later in the course – after students have had more experience with
the more “typical” problems. This problem, could, for example, be used with the marketing research
chapter.
The approach followed in the problem is like the approach that many firms follow in using cluster analysis
techniques to aid in segmenting decisions – and to develop more information about segments. The
problem presents data about the “benefits” that a (small) sample of customers seek from a product—
voice–recognition software. Students enter the data for each customer and based on the clustering results
classify the customers into one of several segments described in the problem. Each customer is classified
into the segment with the most similar “ideal” benefits scores. Then, the students draw on the results of
their analysis to get an idea of the cluster size, characteristics (computer used) of each segment, and
other related information. A small number of customers are analyzed. The point here is for the student to
see how the ideas apply – not to try to develop a “representative” set of results. After doing this exercise,
students will have a better idea of how marketing research can be used to help with segmenting
decisions.
The technical idea underlying this exercise is similar to the notion of “positioning” segments based on
their “ideal” product features, and then seeing which are close and which are not. This point is not
developed in the student materials. But, instructors who emphasize positioning approaches in class might
want to develop this logic in discussing the exercise. The questions also show what can happen when a
company tries to develop an “average” product using the shotgun approach to satisfying everyone –
rather than an approach that targets homogeneous segments.
It will not be obvious to students how the values (the similarity scores) are computed in this spreadsheet.
The approach used in computing the similarity scores is a simplification of a “distance measure” approach
actually used in some popular cluster analysis programs. Even so, the calculations involved are not very
complicated. The overall similarity score for each customer is computed as the sum of a set of 3 similarity
scores – one for each feature. How “close” a customer is to a segment‘s typical (average) preference for
a feature is determined by subtracting that customer’s rating from the average rating for a segment, and
then squaring the resulting difference. By squaring the numbers, minus signs disappear – and bigger
differences (from a segment average) are counted more heavily. Once a similarity score is computed for
each feature, they are summed across features. Then, the same sequence is followed for the next
segment.
The key point to emphasize – perhaps before students start this exercise – is the notion of a distance
measure. The lower the computed score, the more like (closer) the potential customer is to the segment
“ideal.”
The value of this exercise will be enhanced with some in–class discussion. The key points to bring out in
the discussion are covered below. Because the “answers” for the different questions are developed from
across several different spreadsheets, summary tables (like those in the exercise) will be used here –
rather than repeating all the individual spreadsheets. The initial spreadsheet for this problem is presented
next: