Evaluating the Appropriateness of Market Segmentation Solutions Using Artificial Neural Networks and the Membership Clustering Criterion

Abstract

The “appropriateness” of a given segmentation solution is a key consideration in all marketing segmentation studies. By appropriate, it is meant that not only has the optimal segmentation solution been identified, but also that the proper number of segments to market to has been correctly specified. This research focuses on the second, and more fundamental, issue of determining the appropriate number of segments in a marketplace. If the appropriate number of segments is over-specified, marketers may over-segment the market and treat audience segments separately that could effectively be treated inclusively. Conversely, if the appropriate number of segments is under-specified, marketers may under-segment the market and fail to identify distinct, viable segments that should be marketed to separately. The issue of market under- and over-segmentation may be addressed with the membership clustering criterion (MCC), an analytical technique based on fuzzy sets derived from artificial neural networks (mathematical models of animal nervous systems). Using artificial and real world data sets, we empirically test the MCC, compare it to existing methods for determining the number of segments in a market, and demonstrate its advantages in evaluating the appropriateness of marketing to different numbers of market segments.