A data-driven approach to addressing missing item data in ecommerce

Abstract

As part of managing, planning for, and carrying out e-commerce order fulfillment, retailers and their fulfillment engines depend on product item weights and dimensions. Missing or incorrect item data can lead to poor, costly decisions, for example, leading to inaccurate shipping cost estimates resulting in more costly fulfillment decisions, or showing incorrect information to customers. However, retailers often struggle to get accurate and complete item weight and dimensions, with many items and high item turn-over, and as a result many item characteristic values are missing or inaccurate. We analyze the impact of missing item weight and dimension data on fulfillment results and cost and propose and evaluate a data-driven approach for addressing the missing data, which has been implemented and is currently being used in a production e-commerce order fulfillment system.