Assessing the Added Value of Different Data Types for Predicting Customer Attrition

• Activity Number: 224
• Type: Topic Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Michel Ballings*+ and Dirk Van den Poel
• Companies: Ghent University and Ghent University
• Keywords: predictive analytics ; customer intelligence ; survey data ; classification ; data types ; customer churn

Abstract

In an increasingly digital world, every action a customer takes is stored in data repositories, leaving companies with an ever-growing amount of data. Although this data has already been assigned the status of a strategic asset, limited assessment has been made concerning the differential value of different data types as resources in a customer intelligence context. In this study, we first elaborate on how data is important in an analytical Customer Relationship Management context, second we provide a taxonomy of different data repositories using the customer's purchasing process as an operational yardstick and third, we empirically determine the added value of each data repository in predicting customer churn. The latter issue is investigated by comparing alternative statistical as well as data mining classification techniques. The outcomes of this study provide clear recommendations as to which data types companies should invest in for optimizing the value of their resources.