Introduction to the Special Issue on Call Center Management

Abstract

Over the course of the last two decades, call centers have become the preferred way for most businesses to communicate with their customers. Information technology made this possible, but OR/MS techniques needed to facilitate smooth operation with a balance between costs and service level. Most large call centers, with say more than 200 employees, use some type of automated workforce management tool, in which queueing models, simulation, and mathematical programming methods are implemented. Still, much remains to be understood regarding the dynamics of call centers. This special issue explores a number of topical issues related to the management of call centers. Out of the twelve papers in this issue more than half deal with issues related to what is called workforce management: the operational scheduling of call center employees. Two of these deal with forecasting, the others focus on staff scheduling, mostly by means of simulation-based methods. The other papers concern various issues, such as outsourcing and forms of training. We provide an overview of the individual papers below. Call centers are evaluated according to certain service-level objectives. Usually 80% of the calls should be answered within 20 seconds, and the abandonment percentage should not exceed a certain level. Strictly optimizing according to these objectives sometimes leads to counterintuitive results in which a firm maximizes performance by providing poor customer service to a certain class of calls. Milner and Olsen explore this in the context of a situation with contract and noncontract customers. Workload prediction is the crucial first step of the operational workforce planning cycle. Up to now it received surprisingly little attention in the scientific literature. The next two papers deal with this important topic. Taylor compares a number of different univariate forecasting techniques using data from a British bank and shows that simplistic methods work well under some circumstances, while other settings require more sophisticated forecasting tools. Soyer and Tarimcilar are the first in the literature to address the problem of quantifying, over time, the effect of marketing campaigns on the call centers. Because such campaigns can substantially alter demand patterns, their work is relevant to many sales-oriented call centers. Gurvich, Armony, and Mandelbaum analyze rules of thumb for routing and staffing in a model where all agents can serve all customer classes, but where there are different service-level requirements (the socalled V-model). They provide a heuristic algorithm to support staffing and assignment decisions that is nearly optimal for large systems and quite effective for small systems. While most staffing and scheduling systems are based on a stationary queueing model, Atlason, Epelman, and Henderson provide an alternative method, based on simulation and cutting planes, which takes the time-varying behavior into account. Cezik and L’Ecuyer present an approach based on similar ideas, but with a focus on operations with multiple types of calls and different possible skill combinations for call center representatives. Feldman, Mandelbaum, Massey, and Whitt also present a simulation-based method for staffing in the case of varying parameters. Bhandari, Scheller-Wolf, and Harchol-Balter consider staffing with permanent and temporary operators in order to cope with the fluctuations in offered load. Aksin, de Vericourt, and Karaesmen compare outsourcing contracts based on call volume and service capacity. Ren and Zhou also consider outsourcing contracts, but they relate the call center effort (such as additional training) to the call resolution ratio, which is the percentage of calls that received a good answer or led to a purchase. They show how to design the contract such that the call center is stimulated to staff and exert effort according to the outsourcing company’s objective. The paper by Murthy, Challagalla, Vincent, and Shervani is a statistical field study of different types of training. They show that simulation-based training outperforms more traditional forms of training. Finally, Jouini, Dallery, and Nait-Abdallah look at a center of a French telecom provider that introduced small independent teams, each with their own customers, instead of one large pool. This decrease in scale meant a loss of efficiency, which remained limited because of the way unidentified calls were assigned over all agents. However, this loss of efficiency was outweighed by motivational and staffretention benefits, making this organizational structure also a viable strategy for other call centers. This special issue was initiated by the late Perwez Shahabuddin.