Reducing total fulfillment at costs at Amazon EU through network design optimization

Abstract

A key supply chain management issue encountered by any business requiring a distribution system is in designing its distribution network. A distribution network configuration has both direct and indirect ongoing effects on how a firm operates influencing everything from supplier relationships and contracts to customer interface. A configuration affects both day-to-day operational and longer range strategic and tactical decision-making. From a pure cost perspective, a configuration has a significant impact on total fulfillment costs. The effects of network configuration as well as the challenges and value behind the application of network design optimization techniques are well-illustrated by my 6-month experience working for the largest online retail distributor, Amazon.com, in their European (EU) operations. This paper further documents the process followed in identifying areas for improvement in Amazon's current EU fulfillment networks for the purpose of enabling total fulfillment cost reduction. The challenges and results from my experience are similarly included.

(cont.) Two main projects were ultimately selected and documented in this paper: The first was aimed at minimizing transportation costs around the existing UK network configuration, while the second was targeted at minimizing total fulfillment costs through the alteration of EU network designs through the focused adjustment of product and inventory distributions. The first project has to date enabled significant minimization of UK transportation costs. The second project dealt with two complicated mathematical formulations ultimately intended for optimization, one of which is not yet covered in literature. In this case, further research and investigation is required for its practical implementation; nonetheless, the developed formulation was applied to a simplified scenario for the purpose of future study including validation and extension. The ultimate objective of this paper however is to demonstrate the hidden potential and value behind the application of underutilized analytical techniques to network design through the tailored development and implementation of practical decision-support systems.