Brand Loyalty. Plus ça change...? Using the NBD-Dirichlet parameters to interpret long-term purchase incidence and brand choice.

he NBD-Dirichlet model (Goodhardt, Ehrenberg & Chatfield, 1984) has recently been described as one of marketing’s “true scientific theories” (Sharp, 2010) and yet its two assumptions of stationarity and non-partitioning cannot allow for customer or brand equity effects. The aim of this paper is first to describe the effects of brand loyalty on evolving longter market structure, and then to extend uses of the model’s parameters A, K and S as performance benchmarks. Using a six-year panel of continuous reporters we show that the normal heterogeneity in category purchase incidence remains approximately stable between six months and a year, but that brand switching greatly increases with time. One interpretation of a higher S parameter value might be that as loyalty decreases, brand shares become more dynamic, leading to sustained changes in market structure. By fitting the model we show that this is not so. The Dirichlet output fits brand performance measures well in both extended and short-term data, and although there is evidence of emerging triple jeopardy for the largest brands, this does not lead to sustained brand-share growth. Instead, it is a deviation from the fit of the NBD-Dirichlet, but a general characteristic of extended repertoire purchase. For managers, the main implication of such fixed and long-term propensities is that brand growth remains quite unlikely, but also that brand-share maintenance requires constant effort in the face of increased switching propensities over time.