Since starting boobook, I’ve been privileged to run a large number of segmentation projects. Some are a delight to work on, with segments that pop out and immediately make intuitive sense, but many take much more ‘data finessing’ to get to that point. Although each project presents its unique considerations, there are challenges shared by all of these projects, and we can all benefit from addressing them earlier in the process. Asking the following four questions can help you on the path to segmentation success.
1. Who are you really trying to segment?
We often analyze samples that are generally directed towards a specific target market but not tightly so. Imagine that your client is planning on launching a new digital radio station and wants to know more about the different groups of listeners it will appeal to. Do you want to include people who don’t listen to the radio? No, that’s easy enough to answer. What about people who don’t like the type of music the station is going to play or people who only listen to non-commercial stations because they hate advertisements? To answer these questions, we need to think about what the station wants to do with its segments and define its objectives.
2. Will consumers engage with your subject matter in the same way as your client?
As researchers, we spend our days talking and thinking about our clients’ products and services. By the time we get to work on a data set, we’ve spent weeks, if not months (and in some cases years!), focussing on it. This frequency bias leads us to feel that consumers should engage with what we’re asking them about. Remember to step back for a minute and ask yourself, “If I hadn’t invested all this time with this client, would I have an opinion here?” If the answer is no (and it won’t always be), should you be surprised when a group of respondents doesn’t feel strongly either?
3. What do we need to know about our segments?
Ok, so this one isn’t just about segmentation. We ask so many questions in surveys, and the temptation is to ask every question that could be important and put it into the segmentation, just in case. It’s much more important to focus instead on things we know are important and will influence a choice a consumer will make. Clients usually know this stuff already, and it is a very rare occurrence for us to run a segmentation as an entirely exploratory exercise. Certainly, from an analyst’s point of view, I’d much rather have ten well-answered questions pinpointing a concise area than a hundred that cover every possibility.
4. Should we use the entire sample?
Sometimes sample sizes restrict what we can do with segmentation, and if that’s the case for you, you can skip this paragraph. For most of you, though, you’re likely going to be working with decent base sizes. Why worry about kicking out some respondents who don’t provide interesting answers? I’m not just talking about speeders or respondents with the same answers to every attitudinal question you throw at them. I’m talking about respondents who simply don’t show a substantial deviation from the mean. Maybe they slightly agree with a few things here and somewhat disagree with a few others there, but are they then contributing to well-differentiated and useful segments? Probably not. If you’ve got a sample of 1,000 respondents, will it matter if we only use 900 to base the segments on?
One size never fits all
Now, this is by no means an exhaustive list. One thing we can say for sure is that every segmentation that Boobook executes presents its own set of considerations, and they’ll often override one or more of these tips. My best bit of advice: always focus on the objective of the segmentation. If you think about why the client needs segments, what they’ll do once they’ve got them, and keep these two answers in mind throughout the process, you will come to a better, more useful solution.
Want to know more? Contact us and discover how we can help your business!