Predict and optimize: a smart inventory management model for beauty retail

Abstract Effective inventory management is critical for retail companies, particularly in the beauty sector, where balancing stock availability and cost control directly impacts profitability. While inventory management is recognized as a key driver of operational efficiency, it faces challenges since demand fluctuates and future needs are highly uncertain. In this paper, we examine the inventory management problem for a beauty retailer, and show how we can improve upon their current approach to inventory management, by combining stochastic optimization methods with modern prediction approaches. Currently, the company uses a periodic review system, in which inventory levels are reviewed at set intervals and replenishment orders are based on current stock levels and historical demand distribution. Although this approach is cost-effective, it struggles to adapt to changes in demand over time and periodic patterns within the replenishment cycle and during lead times, which can result in lost sales or overstocking. Our model enhances the traditional periodic review system by integrating stochastic demand forecasting and dynamic re-optimization, allowing for more responsive and informed ordering decisions. By accounting for demand uncertainty and adjusting inventory levels accordingly, the model aims to minimize overall operating costs while maximizing service levels. For a beauty retailer, where customer loyalty is a key consideration, we show that the new method leads to higher service levels, which reduces inventory costs and increases customer satisfaction.