An Optimization-Based Approach to Efficient Clearance Inventory Allocation

Abstract

Allocating clearance inventory effectively remains a critical challenge in retail environments characterized by short decision cycles, fluctuating demand, and operational constraints. Decisions made during the clearance period are particularly impactful, as they determine the final opportunity to recover value from unsold products before they lose relevance or perishability. This thesis presents a mathematical optimization model designed to support the redistribution of discounted articles across a network of stores, with the objective of maximizing revenue while satisfying constraints related to stock availability, store capacity, and observed demand at the article-size level.Developed in collaboration with a leading global fashion retail company, the model was built to align with existing business processes and balances analytical rigor with simplicity in implementation. The model incorporates business-defined parameters and is tested using real operational data from selected distribution centers. It demonstrates significant improvements over the current practice of single-item allocation and addresses the computational challenges posed by the high dimensionality of real-world retail problems. By implementing efficient iterative procedures and demand-scaling mechanisms, the model ensures tractability while capturing the complexity of the business environment.