Doing the splits-Using CHAID to identify customer segments

Chi-squre automatic interaction detection, or CHAID, is a relatively new technique for use by analysts faced with market research data sets. CHAID can be used where the dependent and explanatory variables are all categorical, with the specific aim of identifying the explanatory variables (in terms of main effects or interactions) which form distinct segments with regard to responses to the dependent variable. Such segmentation, when applied to customers, supports marketing practice and theory. Two examples of the application of CHAID are described, and the technique is evaluated from the analyst’s perspective.