A Componentwise Index of Service Measurement in Multi-Component Systems Sridhar Seshadri, 1 Jayashankar M. Swaminathan 2 1 Stern School of Business, New York University, New York, New York 10012 2 Kenan-Flagler Business School, University of North Carolina, Chapel Hill, North Carolina 27599 Received 22 March 2000; revised 14 June 2002; accepted 26 July 2002 DOI 10.1002/nav.10050 Abstract: In this paper we present a componentwise delay measure for estimating and improving the expected delays experienced by customers in a multi-component inventory/ assembly system. We show that this measure is easily computed. Further, in an environment where the performance of each of the item delays could be improved with investment, we present a solution that aims to minimize this measure and, in effect, minimizes the average waiting time experienced by customers. © 2002 Wiley Periodicals, Inc. Naval Research Logistics 50: 184 –194, 2003. Keywords: multi-component inventory/assembly system; random lead times; fixed budget; greedy allocation 1. INTRODUCTION Improving customer waiting times in a multi-item inventory/assembly system is a challenging task since demands for different items may be correlated to one another. This is particularly true in environments such as PC and electronic assembly where randomness in the delivery time may primarily be due to randomness in the kitting process. Researchers have worked on different aspects of this problem including performance evaluation, developing approximations for performance measures and system optimization subject to budgetary constraints (see Hausman [6], Song [10], Glasserman and Wang [5] and Song, Xu, and Liu [11], for example). In such an environment, it becomes difficult for managers to allocate resources to improve the delivery performance with regard to different items because their performance may be interrelated. Zhang [12] recently wrote an article in this context that has important implications for customer order management in many industries. Zhang showed that in a multi-item inventory system, when demands are correlated across items, the expected time delay experienced by customers for order fulfillment will be overestimated if the individual item demands are assumed to be Correspondence to: S. Seshadri © 2002 Wiley Periodicals, Inc.