|M.Sc Student||Fisher Amit|
|Subject||Customer Lifetime Evaluation in E-Commerce and its|
Application for Scheduling
|Department||Department of Industrial Engineering and Management||Supervisor||Dr. Opher Etzion|
e-Commerce companies understand that their customers are their most important asset, and that it is imperative to know their value. One of the widely acceptable methods in conventional marketing is to evaluate such customer value using models known as Customer Lifetime Value (CLV) models. A CLV model provides an embedded value of a customer, by projecting and discounting the future net cash flows for a customer across all product holdings.
In the past, Markov Chain Models (MCM) that use a set of three variables - Recency, Frequency and Monetary (RFM), were used in order to calculate this value. Customers are then clustered into value groups and the value of a customer (CLV) is calculated for each of those classes. In this research we present the e-CLV model, an enhanced MCM for the calculation of CLV for the e-Commerce domains. This model adds to the RFM variables several new e-Commerce and QoS variables required for evaluating customer value in an e-Commerce environment.
We also show that for a popular auction site in Israel our model succeeded in predicting the future income generated from their customers and to successfully assign customers to value classes with a high degree of accuracy. Furthermore, by using simulation, we show that our model achieves even greater accuracy when customers’ behaviors are influenced by the site’s QoS. In Addition, we introduce how our model can include Service Level agreements (SLA) constraints, and how we can use the e-CLV in a simple scheduling mechanism in order to improve the site’s revenues.