A Loyalty Group Segmentation Model for Brand Purchasing Simulation

Abstract

A model for consumer package goods is described by which consumers can be segmented into loyalty groups. A survey is used to obtain measures of consumer willingness to switch from a regular brand, if there is one. Each loyalty group is associated with a unique purchase probability vector. The probabilities are derived empirically from purchase behavior. By use of the purchase probability vectors, brand shares and repeat rates can be simulated by the loyalty group segmentation (LGS) model, which yields good fits. Empirical data were available which reflected the effect of price changes on brand switching behavior by loyalty groups. At first price elasticities based on deviations from average price were used with poor results. By a revised version of the model, average price was recomputed as the sum of the prices of competitive brands weighted by the percentage of the brand's total switching with each other brand. By use of such loyalty group cross-price elasticity measures, high correlations were obtained with the empirical observations of share, repeat rate, and switching behavior.