MKT5610 Chap.5 Market Segmentation and Targeting
Market Segmentation and Targeting
The right basis is the one you cannot see
The course reading states the position immediately: a good way to segment customers is on their differences in needs or preferences, for instance between price-sensitive and price-insensitive buyers, and yet it is not always easy to identify those segments. Both halves matter.
The first says what the correct basis is; the second says that the correct basis is frequently invisible at the counter, and that the whole practical difficulty lives in that gap.
A proxy is what closes the gap, and abandoning the basis is the wrong repair
A proxy is something observable that correlates with the unobservable need.
Its quality has two dimensions that trade against each other: how tightly it tracks the need, and how cheaply it can be read at the moment of the offer.
The standard error is to solve the problem by segmenting on whatever is easy to see, which produces reachable groups that want different things internally.
A distribution is not a weight until you multiply by the counts
Survey ratings arrive as a distribution over rating levels.
Averaging the labels returns a number in the right range and discards the shape, which is where the segmentation signal usually is: a split distribution is not noise around an average, it is two groups added together.
What this chapter covers
- 01
Segmenting on needs or preferences, and the identification problem it creates
- 02
Segmentation, targeting and positioning as three separate decisions
- 03
Proxy variables, and the trade between accuracy and cost of observation
- 04
Micromarket detection, and why a coarse cut hides variation
- 05
Four questions that reject a candidate segment
- 06
Converting a rating distribution into an importance weight
- 07
Bimodal distributions as a segmentation signal rather than as noise
Find a proxy that makes a need-based segment addressable
- 2Say why replacing the basis with something visible is the wrong repair.
- 4Propose three candidate proxies and rank them on accuracy and cost.
- 3Choose one, state the action it supports, and accept its error rate.
Key terms
- Segmentation Basis
- The variable on which a market is divided. Need or preference predicts behaviour best and is hardest to observe, which is the trade the whole topic is about.
- Proxy Variable
- An observable quantity that correlates with an unobservable need closely enough to act on. Every workable segmentation in practice runs on one, and an imperfect proxy beats an unaddressable segment.
- Targeting
- Choosing which identified segments to serve. It is a resource decision rather than an analytical one, and a segment can be real, reachable and still wrong because someone else already serves it better.
- Micromarket
- A pocket of demand visible only at a fine level of detail and averaged away by a conventional regional or demographic cut. Finding one is useful only where you can also act at that level.
- Importance Weight
- How much a segment cares about a benefit, computed by multiplying each rating by the number of respondents who gave it and dividing by the total. It belongs to the segment rather than to any brand.
- Bimodal Distribution
- A pattern in which responses cluster at two separated points rather than around one. It is the clearest available signal that an analysis should be run twice.
Market Segmentation and Targeting FAQ
Why not just segment on something I can actually see, like year of study?
Because the groups you get will be reachable and internally mixed, so the offer aimed at each has to be broad enough to suit everybody in it, and broad offers are what segmentation was meant to escape. Keep the basis that predicts behaviour and hunt for something observable that tracks it. That hunt is most of the practical work and it is the part a plan usually leaves out entirely.
How do I know whether a segment is worth targeting?
Four questions dispose of most candidates. Is it large enough to justify a dedicated offer. Can you reach it at a cost proportionate to what it is worth. Does it want something you are genuinely better at supplying than the alternatives it currently uses. And will the difference persist. A segment that fails the third is the expensive one, because it passes the first two convincingly and loses money slowly.
What exactly goes wrong if I average the rating labels?
You discard the shape of the distribution and keep a number in the right range, so nothing looks broken. It fails whenever responses are skewed, which is most of the time, and it fails hardest on the benefits that matter.
A benefit almost everyone rates at the top and one where opinion splits evenly can return the same label average while meaning entirely different things: the first is a shared requirement and the second is a segmentation variable.
Is a proxy that misclassifies people acceptable?
Usually yes, and refusing one is how a correct segmentation ends up changing nothing. The question is not whether the proxy is accurate but whether acting on it beats acting on nothing, and for a business currently making one undifferentiated offer the answer is almost always that it does. The threshold to watch is reversibility: use a noisy proxy for what is displayed, not for what is charged.
Assessment move
Take the ratings your group collects during validation and, before computing anything, plot each benefit as a distribution. Look for the ones that split. Those are where your two segments are, and finding them at validation is far cheaper than discovering in month two that your single offer suits neither group.
Working through Market Segmentation and Targeting in MKT5610? Sia is AskSia’s AI Marketing tutor — ask any MKT5610 Market Segmentation and Targeting question and get a clear, step-by-step explanation grounded in how MKT5610 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
