A Primer on Segmentation Types
Many of our clients ask for a quantitative segmentation to be done on their brand’s target audience. This almost always leads to an interesting discussion regarding types of segmentations. (Here’s where we hope some of you say, “Hey, I didn’t realize there were varying types.”) Here’s a brief overview:
There are two types of segmentation: “a priori” and “post hoc.” “A priori” defines segments based on how the brand team views the world. “Post hoc” lets segments emerge based on how consumers view the world.
In a priori, groups are defined in advance of fielding the study. For example, frequency or amount of usage could be a variable that defines segments to be created. This might result in a heavy user segment, a medium user segment or a light user segment. Geographic and demographic attributes are also frequently used in this type of work. Sophisticated statistical techniques are not required. Basic significant difference testing can illuminate the core variances across groups.
A Priori example based on shoe purchasing:
Now this is where things get interesting. In simple terms, a post hoc segmentation would define segments based on how members view their world. This may be referred to as attitudinal, psychographic or behavioral segmentation. After data are collected, statistical techniques such as K-means, Hierarchical or TwoStep Cluster Analysis can be applied. An analyst uses these techniques to find the variables that both make sense to include and pull segments of people apart. This is certainly a no-one-size-fits-all approach and does rely on some level of trial and error to uncover results most useful to the business problem at hand. Let’s use shoe purchasing as an example again, but this time the segmentation is based on attitudes. You’ll see that the table below is different than the a prioi example, because in post hoc segmentation, multiple variables are used to define each segment.
A Post Hoc example based on shoe purchasing:
As you can see, these segments are based on different learnings than the first example. Both a priori and post hoc analysis are valuable. Let’s talk about what your team wants to learn to determine the best approach for you!