Deriving and exploring behavior segments within a retail loyalty card program

Despite the proliferation of retail loyalty programs, little is known about differences in the behavior patterns of the consumers within them. There may be several unique segments within a loyalty program, and significant managerial implications may accrue from identification of these segments and analysis of differences among them. This paper aims to investigate the potential for deriving meaningful, managerially relevant customer segments within a retail loyalty‐type program.

Data from the one‐year test of a retail loyalty card program by a major US retailer were used to derive patronage‐based segments and explore determinants of segment differences and strategy implications. A numerical taxonomy process was used to group 57,650 loyalty‐card program members into distinct segments. Cluster analysis was used to generate a range of potential market structures on a set of managerially relevant variables. The most appropriate market structure was selected using scree testing and discriminant analysis. Each of the resulting six segments was named profiled. Finally, a set of patronage‐related variables was regressed on to the clusters using multinomial logistic regression.

The results indicate that, at least for this card program, clearly defined segments with unique patronage profiles do exist both descriptively and statistically; only a small percentage of loyalty card program members demonstrate behaviors that can be considered truly loyal; and different marketing strategies appear appropriate to try and increase patronage among the segments based on their profile characteristics.

This study should prove valuable to academic researchers as well as managers. It is the first effort to generate meaningful, rigorously derived segments within a large loyalty program. It shows that behavioral variables can yield managerially relevant segments, and that the segments appear to call for individualized strategy initiatives.