SERVICE CENTER DESIGN CHOICES
My research focuses on how managers can understand customer/server decision-making behavior to design better service centers, such as call centers, back-office support centers, and emergency departments. Examples of customer decisions in these centers include which service channel to join, whether to abandon the center after waiting in queue, and whether to return to the center. Examples of server decisions include whether to hand a customer off to an expert or continue working with the customer, which customers to serve from the queue, and how many to serve at once. Once managers understand the theoretical and real-world behavior of the customers and servers, they can make more informed choices regarding: 1) which service channels to offer and how to offer them, 2) how customers should be routed through the system, and 3) when and how servers should perform handoffs.
Although call centers have recently invested in callback technology, the effects of this innovation on call center performance are not clearly understood. In this paper we take a data-driven approach to quantify the operational impact of offering callbacks under a variety of callback policies. To achieve this goal, we formulate a structural model of the caller decision-making process under a callback option and impute their underlying preferences from the call center data of a large US-based bank. Using the callers’ imputed preferences, we conduct counterfactual analyses of how various callback policies would affect the performance measures of this call center. We find that offering callbacks can substantially reduce the average time callers wait on the phone and can significantly increase system throughput by acting as a demand postponement strategy during periods of temporary congestion.
A prevalent customer behavior in service centers is abandonment of a queue to seek service through another channel (or to return later). Moreover, it has been observed from real-world data that customers whose service request requires more time with the server tend to be willing to wait longer before abandoning. To explore the managerial implications of this phenomenon, we formulate a stochastic process that characterizes a system with two classes of impatient customers who differ across the classes in their distribution of service times and patience times. We analyze this process to obtain performance measures such as the percentage of customers receiving service in each class and the expected waiting times of customers in each class. We then use our characterization to perform a numerical analysis of the system and find several managerial implications of administering a first-come, first-served system with multiple classes of impatient customers whose service time requirements are correlated with their patience levels.
To increase revenue or improve customer service, companies are increasingly personalizing their product or service offerings based on their customers' history of interactions. In this paper, we show how call centers can improve customer service by implementing personalized priority policies. Under personalized priority policies, managers use customer contact history to predict individual-level caller abandonment and redialing behavior and prioritize them based on these predictions to improve operational performance. We provide a structural estimation framework for how companies can use individual-level customer history data to quantify the improvements to operational performance of these policies. Using data from a real-world call center, we find that our proposed personalized priority policies have the potential to substantially decrease average waiting times or increase system throughput by reducing the percentage of service requests lost to abandonment.
In many service encounters front-line workers (often referred to as gatekeepers) have the discretion to attempt to resolve a customer request or to transfer the customer to an expert service provider. Motivated by an incentive redesign at a call center of a mid-size US-based bank, we formulate and solve an analytical model of the gatekeeper’s transfer response to different incentive schemes and to different congestion levels. We then test several model predictions experimentally. Our experiments show that human behavior matches the predictions qualitatively, but not always in magnitude. Specifically, transfer rates are disproportionately low in the presence of monetary penalties for transferring, even after controlling for the economic (dis)incentive to transfer, suggesting an overreaction to transfer cost. In contrast, the transfer response to congestion information shows no systematic bias.
- Featured on "Resoundingly Human" INFORMS podcast: Link