In this paper we propose a decision tree based approach to modeling service levels in insurance call center operations. Our approach allows call center managers to determine which factors they control have the greatest impact on service levels (ex. handle time, average hold time, etc.). We also propose a sliding window to allow managers to interpret the effects changes in resource allocations have on service levels. To test our solution we analyze data collected from a large U.S. insurance company. The initial results provide good insight into factors affecting service levels.