Estimating cross-training call center capacity through simulation

Abstract

Call centers have grown world-wide during the past decade. One of the most important aspects considered by call center managers is the optimization of its operators, which implies covering the highly variable demand and finding an efficient way to assign people to certain shifts in order to achieve a desirable service level and abandonment rate. Another challenge is determining which system setup is appropriate for the specific call center. Should we have a single-skill call center or multi-skill call center? If we do have the latter, how many multi-skill agents should we have on staff? In this case study, we generate and analyze discrete-event systems simulation-optimization models to test the behavior of a real-world call center under the actual configuration and under different levels of cross-training. The model results help call center managers by: 1) determining the optimal number of operators needed for different staff configurations in order to achieve the targets for service level and abandonment; 2) providing information about the trade-off between the key measurements in the call center; and 3) providing useful information about the number of operators needed and used for each hour of operation to estimate the number of four-hour shifts required to achieve the performance targets. Our experimental findings from this case study suggest that a bi-skill call center is economically better in the long-run compared to a full-skill or single-skill call center. This case study augments the call center body of knowledge by providing additional managerial insights for the operations management community.