Common ground for Bass and NBD/Dirichlet: is conditional trend analysis the bridge?

Abstract

Two models are constantly referred to in the discussion of marketing science. These are Ehrenberg’s Dirichlet model of brand choice (and its foundation application of the Negative Binomial Distribution) and the Bass Model of the Diffusion of Innovation. Both models have been used to foster discussion about marketing science. Both models are at the core of market behavioral discussion. However, despite considerable research extending their individual scopes and applications they have remained as singular models. The Dirichlet provides understanding of frequently purchased products and services in stable market conditions and the Bass model providing the adoption mathematics of infrequently purchased products and services. The purpose of this paper is to suggest there is common ground between these models and to explore the implications of this observation. In this first instance an NBD based model has been shown to replicate some (but not all) of the Bass curve shapes and there are grounds to believe that, with further development an NBD/Dirichlet based penetrations model may also be able to account for brand choice behaviour in a growing category.