Proposal of Loyal Customer Discriminant Model Based on RFM Concept

Abstract

Customer loyalty measurements attract more attention from a viewpoint of customer relationship management; hence, there are some studies that have explored this field. However, none of these studies have considered the potential customer features. In this study, the customer behaviors out of and in the site were considered. This helped determine a wide variety of customer behaviors by adding the behavior data out of the site and has also helped identify a loyal customer more appropriately. We designed a model that can determine if a customer has high loyalty to the electronic commerce (EC) site. To achieve this design, we used the behavioral data out of the EC site in addition to recency, frequency, and monetary (RFM) data from the purchasing data. Furthermore, we adopted a logistic regression from previous studies to accomplish this model. Moreover, we analyzed the purchase history of the loyal customer identified in the discriminant model in order to understand the purchasing tendency. Three customer classifications were used in this study: Customer A, B, and C, and the precision of the RFM and our RFMO models was compared. The RFMO model proved to be more precise in classifying loyal customers and the other customers with high precision compared with the ordinal RFM model about