Modeling Customer Growth

Abstract

This article uses the multinomial modeling technique (Batchelder 1998) to decompose the brand-switching matrix of the RLZ model and produces conditional probabilities for alternative decision outcomes (Rust, Lemon, and Zeithaml 2004; Rust, Lemon, and Narayandas 2005). It then links the probabilities with Bayes’ theorem and creates a new customer growth function on consumer choice (i.e., the customer retention rate). The growth function offers novel insights on customer growth from behavioral choice modeling perspective comparing to mainstream time-based adoption models (Bass 1969, 2004; Gupta, Lehmann, and Stuart 2004). These insights include three predictions that are corroborated with a peak analysis. The market share (k) of a brand determines the shape of the growth function. For small and medium brands, the relationship between customer growth and consumer choice follows an inverted-U curve. The peak of the growth curve may not exist if the market share of the brand is too large.