March has been a month of segmentations for CS Analytics. There are many different ways to segment consumers or products but we typically get involved when some type of statistical clustering is required to try and discover underlying segments from multiple questions in a survey.
We say ‘try’ because when you go in, you can’t always be sure you’ll find any meaningful, and perhaps more crucially, actionable, segments. So there’s always some risk involved.
In some cases, researchers and/or their clients will have hypothesised about the nature of the segments. This was the case in one of our very recent segmentations where the qualitative sessions revealed a small set of archetypes in attitudes towards a particular type of financial product among a specific age group (age groups being a pre-determined segment of course).
Fortunately the segmentation in this case both confirmed and extended the ‘qual’ findings.
In our experience segmentations can take anything from a few hours to a few weeks depending on the scope of the analysis … and how strategic the results will be when they are deployed within the business.
Whatever the size it requires a team effort. For the smaller segmentations that usually means the subject matter expert (typically from the MR agency) and the statistical analyst need to put their heads together. For the larger projects there are often multiple business stakeholders involved.
Despite the risk that the data doesn’t speak to us the whole exercise can be great fun – as clearly describable, and sometimes surprising, segments emerge. And the more strategically they end up getting used, the more satisfying they are to those of us who collaborate in their discovery.