Proceedings of the 1999 Winter Simulation Conference P. A. Farrington, H. B. Nembhard, D. T. Sturrock, and G. W. Evans, eds. CALL CENTER SIMULATION IN BELL CANADA Oryal Tanir Bell Canada 1050 Beaver Hall, 2 nd floor Montreal, QC H2Z 1S4 CANADA Richard J. Booth Bell Canada 483 Bay Street, 5 th floor North Toronto, ON M5G 2E1 CANADA ABSTRACT Call centers have relied historically, on Erlang-C based estimation formulas to help determine number of agent positions and queue parameters. These estimators have worked fairly well in traditional call centers, however recent trends such as skill-based routing, electronic channels and interactive call handling demand more sophisticated techniques (see Cleveland and Mayben 1997). Discrete event simulation provides the necessary techniques to gain insight into these new trends, and helping to shape their current and future designs. This paper relates the experiences of designing call center simulations in Bell Canada. We the experience of constructing, executing and analysing a large call center model. Problems that we faced are identified and potential solutions are given. The examples are taken from large and small call centers alike in the attempt to bring forth some common problems that a simulationist will face. 1 INTRODUCTION As is in most businesses, Bell Canada’s relationship with its customers is a mission critical part of its success. It is the customer call center where customers experience the real personality of our business. This is key to their overall perception of Bell Canada’s value as a services provider. We meet here to understand their needs, and offer product and service solutions and support. We must create a positive experience (quality of service), promptly (speed of answer), serving many customers (average wait time, customer serving time, and large variable call volumes), at their convenience. Understanding this complex relationship is strategically important. Discrete event simulation provides a way to search deeper into this relationship. We believe the call center environment contains interesting opportunities and challenges for simulation. There are many viable problems within a simulation context. We find simulation can add value to: • Customer queuing strategies, • Agent versus electronic channel utilisation, • Load analysis, • Scheduling impacts, • Process redesign, • Executive Learning, • And many other related problems. There are common steps to follow and hurdles to overcome in all these problem areas. The following sections will identify some of the major learning and understandings that was achieved during the design of simulation models in Bell Canada. 1.1 Background Before discrete event simulation was first considered, Bell Canada’s consumer and small business client centers had returned to a provincial alignment (Ontario, Quebec) under regional Vice Presidents. These Vice Presidents reported to a new Customer Care Services Group Vice President. Her focus was to restore operational excellence to the centers while building flexibility and depth by bringing together the other call center teams like the Direct Marketing and Collections Centers. At that time service levels were borderline, but improving from many months of sub standard performance. Previous company initiatives had a strong cost focus. Service had become a variable rather than a given. The result was service chaos and disheartened personnel. Each of the centers struggled to redefine their business processes. Traditional projection and scheduling techniques were proving limited as the nature of our business was changing. These business drivers altered calling patterns and serving times as the complexity of the call mix and contact increased. Business Transformation 1640