Replenishment Recommendation in Convenience Stores

This study examines the impact of a replenishment recommendation system on inventory management in convenience stores. We collaborate with a convenience store chain to implement the system in some locations while leaving others unchanged. We find that the introduction of the system increases reorder points and reduces order size, leading to a 2.9% improvement in service levels without requiring additional inventory and without negatively affecting sales or revenue. Additionally, it reduces managers’ daily ordering time by 36.5%, thus enhancing overall store efficiency. Our study also highlights the important role of managerial discretion in complementing the system. For popular items, managers often adjust recommendations by placing orders ahead of any recommendations or increasing the recommended quantity. Interview with managers suggests that they override in anticipation of demand increases that the system has yet to detect. For non-popular items, given their slow-moving nature, managers are more likely to postpone recommendations due to concerns over excess inventory. Our study demonstrates that, for small retailers without the resources to invest in advanced algorithms, even a basic moving average algorithm can be effective by allowing managerial discretion to complement the algorithm in responding to demand fluctuations. This also underscores the need to enhance recommendation accuracy for both top-selling items, given their significant contribution to sales, and the typically overlooked less-popular items due to their slow-moving nature.