An Empirical Study of Ordering Behavior of Retail Stores

We present an empirical study of the ordering behavior of retail stores in a supermarket chain to investigate whether store managers deviate systematically from order advices generated by an automated replenishment system. We consider concern for forecast errors and handling workload to be possible motivations for such deviations. Using orders and shipments data, and POS sales data for 19,417 item-store combinations over five stores, we construct metrics to measure two aspects of ordering behavior: order shifting and order consolidation. Our study finds that store managers systematically made corrections on automated order advices either by shifting orders from peak days to non-peak days, or by changing the order size. Both types of corrections are shown to be explained significantly by expected factors, such as case-pack size and average demand per item-store combination. The shifting of orders is also dependent on store characteristics (like the delivery schedule and sales volume) and product characteristics (like product variety, profit margin, and availability of excess shelf space). The insights of our study would be useful in improving decision rules in retail inventory management, and in determining the value of store managers’ intervention in automated systems.