Are Segmentations Difficult?

Segmentations are one of the more interesting and useful solutions market researchers employ, yet almost every client side market researcher will tell you a story about the segmentation they worked on that sits in a dusty corner of the organization and provides no value. Why does this seem to happen so often?

Segmentation is one of the specialties of our firm and we are almost always working on a variety of segmentations for different clients. In what seems like at least half of those cases, the clients tell us that they have previously been involved in a segmentation that did not work. We are always interested in learning why the previous one did not work, and understanding how the organization ended up with something they felt they can't use.

It would be easy to think that the segmentations that fail are the result of suppliers that don't understand what they are doing, or brand teams that provided poor leadership. The reality is much closer to Malcolm Gladwell's observations regarding plane crashes, in that a series of small mistakes often add up into a larger problem.

Some of the more common problems we have run across:

Not knowing who the segmentation is for: It is very common to have a going in assumption that the segmentation will provide every answer and solve every problem for every department. While having a framework to view your customer base can help a variety of departments, the reality is that different people have very different needs and you can't serve everyone perfectly. A product development or innovation team will need a segmentation that is built quite differently from one designed for a traditional marketing function. A direct marketing function will need something else entirely. It is ok to try and help everyone, but in order to succeed, the client, supplier, and internal stakeholders need to agree who the primary target for the segmentation is.

Not addressing the targeting issue: This is a common problem, and one that is not well understood until too late in the process. Most of the time (there are some exceptions, but they are rare and don't apply to most categories) segmentation solutions exist on a spectrum from rich and descriptive, to easily identifiable. You can't have it both ways. You either want a segmentation that is rich and articulate about deep emotional details of your customers, or you want segments that are easy to find. It is a difficult choice, and one that is too often not discussed until far too late in the process.

Focusing on the wrong things: The techniques used in cluster analysis have become fun and interesting to lots of market researchers. You can't go to a conference without hearing someone talk about the "cutting edge" segmentation techniques they are employing. This is good for the industry, and it is important to constantly reassess our methods and come up with better approaches. But the reality is that the Segmentation quality is almost entirely driven by things other than the actual math employed. The simplest cluster techniques will produce fantastic results in a well designed segmentation approach, and the most advanced cluster techniques will be useless in a poorly designed approach.

We will go over some other common problems in part two of this post next week.