Accurate Force and Answer Consistency Algorithms for Operator Services J.J. Gordon

Abstract Operator services are big business. In the United States operator salaries per annum amount to approximately one billion dollars. Service providers constantly strive to cut costs while maintaining customer satisfaction. Queueing theory provides two tools to help them do this: force algorithms for accurately provisioning their teams, and answer consistency algorithms for evaluating the consistency of the service provided to customers. We analyze force and answer consistency algorithms based on three models: Erlang C, M/G/c, and M/G/c with abandonments. Using two weeks of directory assistance (DA) field data, we investigate how well these algorithms model the behavior of a real DA operator team. Although the M/G/c model performs best, our analysis suggests that certain modifications to the simple M/G/c model could significantly enhance its practical value.