Cross Channel Customer Mapping

In multi-channel setting (for example, web channel and storefront), it is often required to generate an integrated view of a customer across channels for making better CRM, marketing and merchandizing decisions. In order to generate integrated view, it is essential to uniquely identify a customer across channels. Due to various reasons, it is often not feasible to impose a unique identifier on a customer across channels. Moreover, a customer may not provide her true demographic information, which makes it even more difficult to track her. In the absence of a unique identifier and the demographic information, the behavioral profile of a customer can perhaps be used for tracking her across channels (cross-channel customer mapping). We define four channel-independent behavioral attributes namely, brand loyalty, price preference, responsiveness to promotion, and spending range. We call a behavioral attribute of a customer to be channel-independent if it does not change substantially with the channel characteristics in absence of a conscious effort from the customer. We defined the attributes in such a way that they reflect the statistical nature of a customer behavior, are easily computable, and can be incrementally updated over time. We then match these profile attribute values across channels to map the customers. We used two months Safeway transaction data; randomly divided it into two channels in different proportions, and computed these attribute values. We achieved significant accuracy in tracking the customers across channels, for example, more than 90% accuracy for the high valued customers.