Cross training for efficient and flexible service system portfolios

This paper proposes using cross training to create efficient server portfolios for multi-class service systems which usually use specialized access channels and servers for different types of service. Such a system is seen to satisfy customers with the high service quality, but it isn't robust enough to the volatile demand coming from each channel. Workforce flexibility, which can be obtained by cross training, provides the possibility of centralizing demands from all channels and thus increases system responsiveness. However, cross training may decrease system efficiency, in the sense that it lowers servers' average skill level. It is desirable for service systems to satisfy responsiveness and efficiency simultaneously. Therefore, we seek a reasonable tradeoff betweens these two criteria. In this paper, cross training decisions are supported by multi-criteria optimization models, and thus efficient server portfolios can be generated if servers are cross trained in this way.