Case Studies

Segmentation: The Typical Dilemma for the CPM Marketer

The Challenge

The marketing problem was a familiar one. The marketplace in which this consumer packaged goods brand operated was far from homogeneous and the brand team knew this well. But how to proceed? Could the market actually be segmented successfully? The question was contentious, and management wanted to know how to get the whole brand team to 'buy into' this new system and use it.

The Solution

Online research is, of course, ideally suited to segmentation work since images and other projective techniques can be employed. As well, large quantities of information can be gathered economically, including extensive assessments of needs, purchase habits, generalized lifestyle information, and extensive media profiles.

Working with our strategic partner Catherine Dine, our analytical process involved over two dozen clustering runs (K-means, Hierarchical, Hybrid and Fixed Start Point K-means). We cut the data with these candidate solutions in many ways to see which made most sense from a marketing standpoint. Reliability tests were then run on the top candidates to ensure the segments would stand up over time.

Reporting Is Key

The richness afforded by over 200 variables and 1500 interviews always produces reporting challenges. One secret is to cater to an audience's capacity for "visual recognition" – that is, presenting the information in a way that people immediately connect or relate to, through the recognition of a person in their lives in the descriptions. In this way, the information starts to 'make sense'. The use of imagery helps greatly. In this case we reported our results in both summary and expanded forms. The summary results presented the cluster descriptions as single paragraphs or single pages in autobiographical terms ("I am this, I like to do that") while the full richness of the segment information was presented in detailed tables and figures.

How to help the brand team 'buy into' the results? We had the marketing team themselves guess which segment they belonged to and then had them do the survey themselves to see if they were right. (It worked well.)


<  Back to Case Studies Overview