Live-chat agent assignments to heterogeneous e-customers under imperfect classification

Many e-commerce firms provide live-chat capability on their Web sites to promote product sales and to offer customer support. With increasing traffic on e-commerce Web sites, providing such live-chat services requires a good allocation of service resources to serve the customers. When resources are limited, firms may consider employing priority-processing and reserving resources for high-value customers. In this article, we model a reserve-based priority-processing policy for e-commerce systems that have imperfect customer classification. Two policy decisions considered in the model are: (1) the number of agents exclusively reserved for high-value customers, and (2) the configuration of the classification system. We derive explicit expressions for average waiting times of high-value and low-value customer classes and define a total waiting cost function. Through numerical analysis, we study the impact of these two policy decisions on average waiting times and total waiting costs. Our analysis finds that reserving agents for high-value customers may have negative consequences for such customers under imperfect classification. Further, we study the interaction between the two policy decisions and discuss how one decision should be modified with respect to a change in the other one in order to keep the waiting costs minimized.