Big organisations often seem to spend most of their time running segmentation projects.
Projects that will unlock deep insights into the motivation and behaviour of customers, drive up sales, and deliver exceptional, differentiated, customer service.
All of which would be wonderful, if it wasn’t the third big segmentation project in five years. This one won’t be any better than the last two, and secretly you know it.
Customer experience segmentation almost never works. Why? Because businesses assume that marketing segmentation and CX segmentation are the same thing.
What makes a good marketing segmentation?
Let’s start with what doesn’t make a good segmentation—lazy stereotypes. As Mark Ritson points out, that includes silly generalisations based on gender, age, or even generation.
“Clearly millennials as a generational cohort do exist – they are the two billion people on the planet born between 1981 and 2000. But the idea that this giant army all want similar stuff or think in similar ways is clearly horseshit.”
Good marketing segments are those which reliably predict which messages will resonate and who is most likely to respond, allowing businesses to target the right customers with the right messages.
Segmentation, in practice, is usually built on statistical tools such as cluster analysis or archetypal analysis, which brings us to our next question…
What makes a good statistical segmentation?
When we segment, we look for a way to group customers together that maximises the differences between groups while minimising the differences within groups. The customers in a group are not identical, but they should be similar to each other and dissimilar from people in other groups.
If you can find groups with big differences and little overlap, that’s brilliant. If not, even small differences in average responsiveness can be useful for targeted marketing (particularly in old-school direct mail).
Why? Because the cost of making the wrong judgement (i.e. not targeting someone who would have responded) is only the missed opportunity, it doesn’t do any harm. Marketing segments can be useful, even if they’re not very good. The same isn’t true for customer experience.
What makes a good CX segmentation?
Bad CX segments have the potential to harm your customer experience.
If you can genuinely find segments which are clearly separated, that’s great. More often, in reality, segments are barely differentiated, with a lot of overlap. Unlike choosing whether or not to send someone a piece of direct mail, making the wrong judgement about which customer experience you offer can have serious negative consequences.
Rather than tailoring the experience, bad segments make feel customers that they have been slotted into clumsy, stereotypical, boxes.
So should you give up on trying to segment customers? Not at all. But stick to a few rules for safe CX segmentation:
- Segments should increase choice, not diminish it
- Segments which reinforce stereotypes are usually toxic
- Segments should clarify which needs exist, but…
- Segments should not be boxes to put people in