Multi echelon supply chain design for Amazon private brands

Abstract

Retailers across the globe continue to grow their private label portfolio to offer customers an alternative to existing brands. Typically, retailers source private label products directly from manufacturers to remove middlemen from the value chain, thereby capturing greater value and subsequently passing it on to customers. Combined with the growth of e-commerce as the primary method for consumers to shop for products, expanding private label portfolio has made e-retailers to re-think their supply chain. Amazon began its journey in Private-Label Brands (PB) in 2009 with the launch of Amazon Basics. Since then, it has expanded its presence across multiple categories. The majority of these products are imported from Asia-Pacific region (APAC) and require sourcing larger quantities to account for long-lead time between production runs and high variability in demand to maintain competitive costs.

These factors result in PB inventory dwelling for a long period at the Amazon Robotic Fulfilment Centers (FCs), reducing the turns-ratio of expensive storage bins there, which could otherwise be utilized for storing high-velocity products. The growth of PB products raises the need to build more storage space, which is expensive in highly automated robotic FCs. Additionally, since fixed storage cost is proportional to the space occupied in FCs, high 'dwell time' translates to high storage cost. To increase utilization of FC storage bins, the Inbound Supply Chain Team plans to build a low-cost upstream storage (LCS1) to supply the FCs and store excess PB inventory there. Alternatively, Amazon can also use its third party storage center in APAC, another low-cost storage node (LCS2), after sourcing PB products from manufacturers in Asia before shipping to regional markets in US, EU, Japan etc.

This could provide an opportunity for inventory savings from risk pooling by optimizing inventory storage across various nodes in the supply chain. Using multi-echelon inventory optimization techniques, this thesis explores the tradeoffs between using low-cost storage node close to end customers in the US (LCS1) versus that close to manufacturing source in APAC (LCS2). The objective of the thesis is to find the optimal inventory placement strategy across three storage points - FCs, LCS1 in US, and LCS2 in APAC - to achieve the best-in-class customer experience (InStock availability) at minimal inventory storage cost.