“So this seems on the face of it quite an easy thing to get right, when in reality, it’s an easy thing to get wrong.” Leigh Morris
When conducting segmentation programmes, there is often a trade-off between breadth of audience and depth of understanding.
As an example, you might start with a simple premise: let’s engage with consumers who have been making purchases in our category or likely to have been thinking about this.
On paper this sounds like a straight-forward idea but raises the issue around engagement. Many consumers who buy products in your category may not give the matter much thought and have low engagement. Doubtless these low engaged people contribute a reasonable share of financial value to your category, but there are many circumstances under which a segmentation will have greater operational impact by ignoring them.
There are many circumstances in which a segmentation should ignore this low-engaged customer.
Real-life example
We have been working for several years with a global fashion retailer. In 2017 they worked with a global agency partner to carry out a segmentation programme based on consumer’s needs and motivations in relation to fashion. The ultimate goal of the programme was to help with brand portfolio strategy, giving a clearer definition of target consumers for their different brands.
What the agency found was that there were a large amount of consumers who had little interest in fashion, people who buy clothes for purely practical reasons, to keep warm for work etc. The segments that such people fell into did not represent an appealing target for any of the retailer’s portfolio of fashion brands, and so the brands all converged towards the more engaged segments.
We stepped up and delivered a style-based segmentation that actively is designed to help avoid this engagement issue through a combination of needs-based data and brand perception. We were able to accomplish this whilst excluding the low engaged consumers from the sample and now the client is happy and able to use their segmentation for actual brand portfolio management.
Including people not engaged in your category in your segmentation can have other negative effects. First – they use up a lot of sample, and so when you want to deep dive into key target segments your ability to do so is limited by sample size. Second – they constrain the segmentation model itself, such that less obvious dimensions that can meaningfully differentiate you from your competitors are missed by the algorithms.
