Service Engineering of Human Tele-Queues: Empirically Based Stochastic Analysis of Telephone Call Centers

University Of Pennsylvania, Philadelphia PA

Investigators

Abstract

Abstract PI: Lawrence Brown (Proposal ID: 0223304) Title: Service Engineering of Human Tele-Queues: Empirically Based Stochastic Analysis of Telephone Call Centers A call center is a service network in which agents provide tele-services, here to be interpreted as either telephone-based services or, more generally, any online-service with customers and servers being remote from each other. Sound scientific principles are prerequisites for sustaining the complex socio-technical enterprise of the call center. The overall goal is to contribute to the theory that supports these principles and to the creation of new ones. The proposed research builds from an analysis of two unique records of call center operations. These contain complete call-by-call data for the operation over approximately one year of two call centers of different size and operational character. This data will be analyzed in order to understand the mathematical queueing and network characteristics of such centers as they actually operate. Special attention will be given to developing an understanding of the role of scale and efficiency in relation to such operations. The operational features of these two call centers will then be adapted into realistic mathematical models for their operations. On the basis of preliminary work already done, it is expected that these mathematical models will differ in important respects from those in current use for analyzing and predicting performance of such large-scale service operations. In particular, it is expected that new theory will be specifically developed to model the queueing characteristics of centers that operate in the high-volume high-efficiency domain. Call centers can be viewed naturally and usefully as queueing systems. The most widely used queueing model for a call center is the Erlang-C model, also denoted the M/M/S queue. The Erlang-C model is deficient as an accurate model of a call center in several respects. One of these is that it does not accommodate customers' impatience while waiting. The Erlang-C model also involves other assumptions such as Poisson arrival times and exponential service times. These will be tested against the data, and modified theory will be developed as needed. The statistical analysis will be derived according to the primitive elements of a call-center: system arrivals, queueing behavior, and services. The interactions of these primitives will then be analyzed, and alternate queueing models will be examined.