Starting Cold: The Power of Social Networks in Predicting Non-Contractual Customer Behavior

Abstract

The last decade has seen a rapid emergence of non-contractual networked services. The standard approach in predicting future customer behavior in those services involves collecting data on a user's past purchase behavior, and building statistical models to extrapolate a user's actions into the future. However, this method fails in the case of newly acquired customers where you have little or no transactional data. In this work, we study the extent to which knowledge of a customer's social network can solve this “cold-start” problem and predict the following aspects of customer behavior: (1) activity, (2) transaction levels and (3) membership to the group of most frequent customers. We conduct a dynamic analysis on approximately one million users from the most popular peer-to-peer payment application, Venmo. Our models produce high quality forecasts and demonstrate that social networks lead to a significant boost in predictive performance primarily during the first month of a customer's lifetime. Finally, we characterize significant structural network differences between the top 10% and bottom 90% of most frequent customers immediately after joining the service.

Keywords: Dynamic Social Networks; Customer Behavior; Non-Contractual Settings; Cold-Start; Predictive Analytics

JEL Classification: M31