The purpose of this paper is to test whether brand defection shows double‐jeopardy effects, and whether stochastic models provide useful benchmarks of expected brand defection rates.
The approach takes the form of an empirical study of brand defection in four markets using panel data, comparing the performance of simple OLS models with a stochastic model.
Brand defection shows double jeopardy. Almost all brand defection in the markets studied could be explained by the category examined and the market share of the focal brand. A stochastic model of choice fits these data well, and provides many further practical and theoretical applications.
The study provides improved benchmarks for brand defection, allowing managers to spot whether their brand is performing better or worse than expected. It also allows better analysis of market structure for subscription services, especially under dynamic conditions, and better estimation of customer lifetime value when defection rates are not known for all brands.
The paper closes a gap in the brand defection literature by showing that a given level of defection is a natural characteristic of any market, subject to within‐market variations dominated by market share. Recent work has overlooked these points, resulting in unrealistic goals for customer defection programmes and inefficient estimates of customer lifetime value. The paper also provides a method of defection analysis that can be deployed by numerate managers and market researchers, and used as a basis for further academic work.
