If you’ve ever heard someone say, “you can’t please all of the people all of the time,” and you understand what they mean, you understand the essence of segmentation. Your client population is not homogeneous. It is comprised of subgroups of people who have similar interests or needs. These subgroups are your customer segments.
Many astute and experienced marketers will have a pretty good feel for the customer segments they are dealing with already. In fact, it’s human nature to “stereotype” or to “discriminate” and “classify” people into various categories intuitively.
Credo will help you formalize a process that classifies your clients into different segments.
What processes do we use to accomplish this? How do we segment? We use well tested statistical methods and we combine these in our own, unique Credo way.
- CART analysis = Classification and Regression Tree analysis is a technique that assigns people to a segment (one of a number of established segments, that is) based on a set of simple, sequential categorizations
- Discriminant analysis is another tool Credo uses in segmentation exercises. It also categorizes people into one of a number of defined segments based on their circumstance. This approach to segmentation is a little more flexible than CART, as it doesn’t apply the constraint of sequenced categorizations
- kMeans Clustering is another statistical method used in segmentation analysis. Effectively, the analyst tells the model how many different segments are expected. Then, through an iterative process, a multidimensional data set is explored until the many data points are classified into one of the segments such that the Euclidean distance between the randomly chosen segment centroids (or kernels) and all of the data points from the data set is minimized. A grand optimization exercise, really.
- Latent Class analysis is yet another statistical technique that looks for patterns within data sets. And, on the basis of the patterns that are uncovered by the analysis, people are assigned to
Really, there are lots of statistical techniques for segmenting your client base. In fact, the statistical side of things is the easy part. The more challenging parts of segmentation include:
- The artistry of understanding the business circumstances in a way that enables the creation of an effective segmentation model at the conceptual level
- The collection and organization of data that permits the application of the segmentation model
- The roll-out and implementation of a new approach to thinking about the customer…as this generally requires challenging cultural shifts within the organization.