Modeling end-to-end order cycle-time variability to improve on-time delivery commitments and drive future state metrics

Abstract

Dell is accelerating investments to simplify and improve one of the core competencies it was founded on, customer experience. One goal within this initiative is to increase the percentage of orders that are ontime to a committed Estimated Delivery Date (EDD). EDDs for products vary greatly with the complexity of the customer purchase orders. In order to remain competitive, Dell has set an aggressive goal to provide better on-time delivery performance. Dell needs to quote more accurate lead time commitments to customers and increase the stability of high variability steps in the end-to-end order supply chain. The EDD lead time, from customer order to proof of delivery, consists of a payment (processing) phase, manufacturing (build, inbound logistics, warehouse) phase, and a logistics (delivery) phase. Each of these segments are managed by different organizations within Dell. Understanding what the end-to-end future state looks like will allow functional teams to set improvement targets to achieve Dell's on-time goal. This study has three main objectives: (1) determine the key drivers of variability in the current state process, (2) identify opportunities for more detailed EDD range generation, and (3) quantify targets for individual process steps to drive towards the target future state. Three high volume Build to Order (BTO) regional product lines were chosen as cases to analyze. BTO product lines, compared to Build-to-Stock (BTS), inherently have a more variable supply chain for the processes examined. To meet the main objectives, this thesis examines the hypothesis that a simulation model based on historic order data can be used to quantify existing cycle time performance in the supply chain and deliver targets to achieve Dell's on-time performance target. Key drivers of cycle time variation were identified through process mapping and design of experiment statistical analysis. Results from the modeling and sensitivity analysis produced actionable recommendations for each of the three objectives and lead to a pilot project to improve EDD commitments for an existing desktop product line. Direct to customer shipping, inbound logistics method, and day of week were identified as attributes that were significant drivers of variability and were underutilized in the EDD commitment process. This provided an opportunity for smarter lead time setting. A pilot project for a desktop line adjusted lead times to incorporate direct to customer shipping and day of week, resulting in a 30-40% on-time performance improvement. Finally, modeling results quantified cycle time distribution targets for each process step to achieve Dell's future state goal for on-time delivery. Dell is building on this project by analyzing more regional product lines and exploring opportunities to incorporate machine learning.