The ever-increasing amount of individual-level customer data generated by reward/l- alty programs opens new perspectives for customer relationship management (CRM). Without any question, the potential bene?ts of these data and analytical models for - plaining, extending, and predicting customer behavior is very high. However, recent analyses have shown that a high fraction of CRM projects result in negative return on investment. One of the main reasons for this dilemma is that these data require advanced analytical processing to fully leverage their potential ("analytical CRM"). Yet, research and practice is still in its early stages with respect to analytical CRM. In particular, the so-called "non-contractual settings" remain widely unexplored. Lit- ature refers to a "non-contractual setting" when customer relationships are not governed by a contract that predetermines the monetary value and/or length of the relationship. Examples include hotels, airlines, and most retailers. The most obvious consequence for CRM is that the end of a customer relationship is not directly observable, i.e., a c- tomer can switch providers without notifying the focal provider. Consequently, analysis of customer retention, and future buying behavior is even more problematic than in contractual settings.
