Centre for Marketing MODELING LONG-RUN LOYALTY

Very little is known empirically about customer loyalty to frequently-bought goods over the long term (e.g. over five years or more, or over hundreds of purchases). In this paper we first explore patterns of long-run loyalty in two very different markets; a consumer market – laundry detergents, and a more frequently-used quasi-industrial market – the prescribing of antihypertensive drugs. Second, we determine how well long-run loyalty can be predicted from a stationary-market model – the Dirichlet. We use two measures of loyalty; portfolio size and the share of category requirements accounted for by the buyer’s favorite brand. We overcome problems of purchase incidence heterogeneity by repeatedly examining our loyalty measures for different usage rates rather than varying the analysis period as in traditional panel studies. We find that: • Portfolio size increases with increasing category usage and then reaches a plateau largely independent of usage. This plateau for portfolio size is around 50% of available brands for laundry detergent, 77% for the prescribing of antihypertensive drugs. • Share of category requirements given to the favorite brand declines as usage increases, but again reaches a minimum level (64% for laundry detergent, 30% for anti-hypertensive drugs) that is then largely independent of usage • The patterns of loyalty (if not the loyalty levels) are similar in both markets and both broadly predictable from an extension of the Dirichlet model. However, over the longer term there are small but consistent deviations from the model predictions Both the model fit and managerial implications are discussed.