Segmentation - building for business alignment
Segmentation is comparatively simple to carry out from a market research or analysis viewpoint using statistical tools to identify different groups within a market or among customers. However, making a segmentation strategy work involves more than just data classification.
In particular, the business has to be aligned to the segments to ensure each segment has appropriate products and services, and each segments needs to be clearly identified for targeting, and tracked over time to measure performance. Many pure market research segmentations fail once the research money has been spent, because the business short-circuited the implementation phase required to deliver a successful segmentation.
Market research and analytics sees segmentation as a fairly conventional statistical technique that divides a market up in order to identify customer groups with different needs and wants.
Typically this works as follows. Firstly, produce an attitudinal cluster based segmentation, or a behaviour-based classification, by conducting research to understand the key dimensions in the market or via database analysis. Then conduct a large general survey (typically 1000+ interviews) containing a large number of attitude statements ("Do you agree or disagree with..."). With these results apply a statistical technique to find distinct groups for targeting.
At this point, with the research has largely finished, the core business and implementation issues become apparent. Who is in each segment? How would I identify people on the database in each segment? How do I do follow up work to better define the product or service proposition? How do I test market to this segment? What is the sales potential? How would my sales team quickly identify customers in each segment? Do I have to re-organise my business to focus on each segment? How do I set segment based sales targets (without sales poaching)? And so on.
Unfortunately, the segments produced by cluster analysis can be difficult to replicate and identify in the real world, coming as they do straight from a statistical technique. At best, you might be able to produce new products for each cluster and mass market the products hoping self-selection will result in the right product with the right person. However, even this approach is difficult to communicate through a sales or distribution channel to show the logic of having two or more products.
The business challenge is therefore identifying and marking customers as members of segments in a way that can be captured on a database, or in an easy way in the field and aligning the business to deliver to each segment appropriately.
The route to better market research segments
Segmentation normally arises as a technique to be used during strategic analysis. The aim of the segmentation has to be to find segments that can be implemented and acted upon. Fundamentally, the process of developing the segmentation, and the amount of pre-planning, is a key factor in what you can do with the segments at the end of the analysis.
Consequently, our preferred method for segmentation is to start developing hypotheses for segmentation prior to any statistical analysis, then include hard, real criteria that we hypothesize might drive the segments, using quantitative research and statistical techniques to validate and refine the hypotheses.
If a segmentation is real and has 5-10 segments, those segments should be observable through qualitative research. This should provide the basis for questions that can be used in a quantitative large-scale study to prove the existence of the segments and to size them, with the aim of finding simple questions to place customers in different groups. This is done by using cluster analysis to check the intuition gained through the qualitative research and checking these clusters against the simple questions.
These simple questions can then be attached to data gathering exercises, or used to identify customers at the point-of-sale. In this way anyone can place a customer in a particular segment quickly and consequently know what to sell. This also makes incorporating segments into a database far easier.
Implementation is then still a question of good internal communication, getting the right people to implement, providing them with resources and getting your systems and communications right. But at least you know who you are talking about.
For help and advice on segmentation, including blending research and database data, and thinking through the business issues downstream, contact info@dobney.com
