Unbundling CRM – A RFMC Perspective

Customer lifetime value (CLV) has been gaining increasing importance in many areas but the (RFM – recency, frequency, monetary) segmentation framework is the most commonly used due to the availability of customer data as well as ease of calculation. One of the drawbacks of the RFM segmentation framework is that it is unable to identify behavior correctly in the face of customers that show hot and cold periods of visiting and buying (binge buying). This kind of phenomenon is called clumpiness (C). One of the empirical findings of this study is that such a kind of behavior is quite prevalent and clumpy customers generate (if not significantly) more revenues than customers who are non-clumpy. Even if clumpiness does not account for high revenue customers, this study shows by use of a correlation and cluster analysis that the clumpiness statistic significantly adds to the RFM framework making the establishment of an RFMC framework immanent. The clumpiness phenomenon is studied by performing a detailed empirical analysis on a representative UK based retailer's dataset. Regression and neural network analysis were implemented to identify the main drivers for clumpiness. Cluster Analysis was used to identify customers that have similar behavioral characteristics and to sort them in small groups so that targeting a group of customers is possible. Thus, clumpiness can detect customers that are high potential and high risk in nature, which were previously unseen.