Predicting retailer orders with POS and order data: The inventory balance effect

Evidence is presented that retail echelon inventory processes translate into a long-run balance or equilibrium between orders and POS, which is referred to as the inventory balance effect, allowing for more accurate order forecasting.

Abstract

Despite advances in retail point-of-sale (POS) data sharing, retailers' suppliers struggle to effectively use POS data to improve their fulfillment planning processes. The challenge lies in predicting retailer orders. We present evidence that retail echelon inventory processes translate into a long-run balance or equilibrium between orders and POS, which we refer to as the inventory balance effect, allowing for more accurate order forecasting. Based on the inventory balance effect, this research prescribes a forecasting approach which simultaneously uses both sources of information (retailer order history and POS data) to predict retailer orders to suppliers. Using data from a consumable product category, this approach is shown to outperform approaches based singularly on order or POS data, by up to 125%. The strength of this novel approach – significantly improved forecast accuracy with minimal additional analysis – make it a candidate for widespread adoption in retail supply chain collaborative planning and forecasting initiatives with corresponding impact on fulfillment performance and related operating costs.