SegmentSolve™: The Core Decision Support Tool that Undergirds Our Approach to Market Segmentation.
Critique of the Traditional Approach
 
We will very briefly examine the limitations of three of the traditional segmentation approaches:
 
Behavioral Segmentation is the effort to cross-tabulate or associate behaviors with known demographics. It is the oldest method of segmentation. In essence, it classifies a customer into a "bucket" based upon whether customers are similar to other customers who have performed that behavior in the past. Historically, behavioral segmentation has been generated by cross-tabulation analysis, or more recently by tree-based classification tools such as CHAID or CART. The problem with these methods is currency. That is, in many instances, they have been superceded by data mining (modeling-based) approaches to segmentation. Modeling approaches predict specifically for each customer the likelihood of performing that behavior. Thus, the newer data mining methods produce a probability score for the customer or prospect. This allows the segmentation analyst to prioritize within a "bucket" who is most likely to perform the behavior (e.g., purchase a given product.). Data mining approaches, therefore, supercede traditional segmentation approaches to customer targeting because they not only place a customer in a "bucket," but identify which customers in the bucket we should contact first. If your business objective is to best identify customers who will perform a given behavior, then it may be appropriate for you to perform data mining to predict the behavior rather than to try to use segment "buckets" to predict the behavior.
 
Demographic Segmentation or Life-Cycle Segmentation attempts to determine customer (or prospect) targets based on different combinations of demographics. This is an historically important and intuitive approach to segmentation because customers do buy different products at different stages of their life. For instance, first mortgages are highly associated with prospects in their mid- to upper-20s, and home equity loans are highly associated with customers in their 40s with children entering college. The major difficulty with life-cycle segmentation approaches is product-specificity. Since products differ by life phase, a life-cycle segmentation must be developed for each product. If we develop a demographic segmentation for each product, then, in effect, we are performing the product-specific behavioral segmentation described above, and all the limitations related to behavioral segmentation apply.
 

Attitudinal Segmentation segments the market based on how well customers perceive the product or service to be performed. The rationale behind this type of segmentation is very sound, it recognizes that attitudes drive behavior. The difficulty lies in the fact that most approaches to attitudinal segmentation utilize only performance data. While performance data does tells us how respondents think we are doing, it tells us precious little about what respondents actually will do. Another type of data, -importance or preference data- is much more suited for projecting what a respondent will do. Moreover, two other limitations of attitudinal segmentation should be mentioned. Most attitudinal segmentations are developed using cluster analysis. Cluster analysis is a very powerful approach, but it is "brittle." No single cluster method always works best, but often only one particular type of cluster analysis is run on the data. Secondly, even if the cluster is well separated from the other clusters, attitudinal segments are generally more difficult to classify by simple cross-tabulation, as compared to behavioral clusters.
 
Our approach exhaustively computes and determines millions of possible segmentation scenarios presenting you with the best solutions, visually and qualitatively.

Phone: (630) 428-1847    Fax: (630) 729-3183  Email: info@decisionsupportsciences.com