“Counting Your Customers” One by One: A Hierarchical Bayes Extension to the Pareto/NBD Model

Abstract

This research extends a Pareto/NBD model of customer-base analysis using a hierarchical Bayesian (HB) framework to suit today's customized marketing. The proposed HB model presumes three tried and tested assumptions of Pareto/NBD models:(1) a Poisson purchase process, (2) a memoryless dropout process (i.e., constant hazard rate), and (3) heterogeneity across customers, while relaxing the independence assumption of the purchase and dropout rates and incorporating customer characteristics as covariates. The model also provides useful output for CRM, such as a customer-specific lifetime and survival rate, as by-products of the MCMC estimation.Using three different types of databases ---music CD for e-commerce, FSP data for a department store and a music CD chain, the HB model is compared against the benchmark Pareto/NBD model. The study demonstrates that recency-frequency data, in conjunction with customer behavior and characteristics, can provide important insights into direct marketing issues, such as the demographic profile of best customers and whether long-life customers spend more.

Identifying future defaulters: A hierarchical Bayesian method

“…We estimate it in Section 3 by MCMC integration method. As is pointed by Abe (2009), closed-form expressions would provide a better understanding of the effect from each variable. Similar to the experiment conducted in Section 4, we can complement our model by applying logistic regression on the augmented data with default labels predicted in the MCMC iteration.…”

“…Our method is similar to that in Abe (2009) which used a hierarchical Bayesian extension of the Prato/NBD model to forecast customers future purchases. Through an empirical analysis on three datasets, they conclude that the customer purchasing behavior is better tracked and, most importantly, the correlation estimated provides significant insights in customer relationship management.…”

“…Through an empirical analysis on three datasets, they conclude that the customer purchasing behavior is better tracked and, most importantly, the correlation estimated provides significant insights in customer relationship management. Unlike the consumer behavior discussed in Abe (2009), the event of default on a fixed-term loan cannot occur repeatedly, so the Poisson assumption of customer behavior is replaced by a Bernoulli distribution.…”

Never Say Never—Status Quo and Research Agenda for Relationship Termination in B2B Markets

“…One approach is the NBD/Pareto model (e.g., Schmittlein et al 1987;Reinartz and Kumar 2000). More current studies have reported alternative approaches to measure the inactivity of a customer (Fader et al 2005;abe 2009). The defiance of all approaches is to define adequate defection indicators due to the fact that several repurchase decisions can take years.…”

Nonparametric hierarchal bayesian modeling in non-contractual heterogeneous survival data

“…This model is attracting the attention of researchers and practitioners because of its increasing importance in new types of marketing, such as CRM, and One-to-One Marketing. Their work is highly regarded and follow up research has been conducted [4][5] [6][7] [8]. Abe [8] proposed a hierarchical Bayes extension to the Pareto/NBD model to estimate the impact of the customer's characteristic variables on profitable lifetime duration.…”