Not logged in.

Contribution Details

Type Conference Presentation
Scope Discipline-based scholarship
Title Time-varying covariates for the Pareto/NBD model
Organization Unit
Authors
  • Patrick Bachmann
  • Markus Meierer
  • René Algesheimer
Presentation Type keynote
Item Subtype Original Work
Refereed Yes
Status Published in final form
Language
  • English
Event Title INFORMS Marketing Science Conference 2016
Event Type conference
Event Location Shanghai, China
Event Start Date June 16 - 2016
Event End Date June 18 - 2016
Abstract Text Today valuing customers is essential any firm and enables marketers to identify key customers. Customer lifetime value (CLV) is a central metric for valuing customers. It describes the long-term economic value of customers and gives managers an idea of how customers will evolve over time. Among the various approaches to assess CLV in non-contractual business settings probabilistic customer attrition models stand out due to their ability to simultaneously forecast a customer’s actual lifetime and future transactions. However, empirical evidence suggests that standard probabilistic customer attrition models do not outperform basic management heuristics. One of the main reasons for the limited predictive performance of the Pareto/NBD model is the assumption of independence of observed external determinants. Regularity patterns and non-random direct marketing activities are the two broad categories of contextual factors identified in previous literature to affect the predictive performance of latent probabilistic customer attrition models. Time-varying covariates would allow to control for these effects. Recently an extension to include time-varying covariates for continuous non-contractual business setting non has been presented. However an implementation of time-varying covariates for the continuous non-contractual business setting is not available. In this paper we propose a latent attrition model that allows for the first time to model time-varying contextual factors in continuous non-contractual settings. Our approach is an extension of the standard Pareto/NBD model. Our extension allows time-varying contextual factors to influence both, the purchase and the attrition process. Our findings have strong implications for both, marketing practice and research. Besides giving detailed recommendations on when to use which modeling approach, we also provide practical advices for estimating these models.
Export BibTeX