Simple Models for Customer-Based Analysis: Linking RFM with CLV

Peter S. Fader, University of Pennsylvania

Bruce G.S. Hardie, London Business School

A “hot topic” today is the quest to use a firm's transaction data to predict its customers' future purchasing patterns. While many computationally intensive models have been developed to undertake such analyses, the “costs” of the required modelling infrastructure have served as a barrier to widespread implementation. To address this concern, we recommend the use of a “simple” model; one that makes use of carefully defined data summaries and that can often be built and implemented using readily available software (e.g., spreadsheets).

As a detailed example, we present a highly parsimonious model that links the well-known RFM (recency, frequency, monetary value) paradigm with customer lifetime value (CLV). We utilize a Pareto/NBD framework to capture the flow of transactions over time and a gamma-gamma sub-model for dollars per transaction. Much of our focus is on a variety of holdout tests to demonstrate the validity of the model as a whole as well as its underlying components. We summarize a number of substantive insights and point out a set of broader issues and opportunities in applying this type of model in practice.