Sharp, Wright and Goodhardt (2002) (SWG) discovered that competitive repeat-purchase markets are polarised into two types: (1) “repertoire markets”, where consumers have repertoires so the typical brand is bought by its average customer less than half the time (an average share of category requirements of less than 50%) and few of its customers are 100% loyal, and (2) “subscription markets” where consumers have very small repertoires so brand loyalty metrics are all higher, e.g. the typical brand enjoys high levels of 100% loyal buyers. The time independent NBD-Dirichlet model parameter S elegantly describes these differences in loyalty metrics. Based on a very limited number of data sets Sharp, Wright and Goodhardt estimated that repertoire markets have an S value no lower than 0.6. To check this estimate we calculated S values for 468 packaged goods data sets from a TNS consumer panel. The vast majority of categories showed S values above 1.0 and only 4% of the data sets showed an S value slightly below 0.6. These categories with unusually high brand loyalty appear to be declining ‘old fashioned’ product categories with low penetration rates and a greater reliance on older buyers. This raises an interesting question for marketing theory - is high loyalty more usually a reflection of failure to acquire new consumers?
Sharp, B (2007), "Loyalty Limits for Repertoire Markets", Journal of Empirical Generalisations in Marketing Science, Vol. 11, No. 1
Brand Loyalty, Dirichlet, NBD, Repeat Purchase