Point-of-sale demand forecasting

Abstract

Nike Always Available (AA) is a significant global business unit within Nike that allows retail customers to purchase athletic essentials at weekly replenishment intervals and 95% availability. However, demand fluctuations and current forecasting processes have resulted in frequent stock-outs and inventory surpluses, which in turn affect revenue, profitability, and brand trust. Potential root causes for demand fluctuations have included: -- Erratic customer behavior, including unplanned promotional events, allocation of open-to- buy dollars for futures (i.e., contract) versus replenishment (i.e., AA), and product inventory loading to protect from anticipated stock-outs; -- Lack of incentives and accountability to encourage accurate forecasting by customers. Current forecasting processes, which utilize historical sell-in data (i.e., product sold to retail customers) were found to be significantly inaccurate - 100% MAPE. The goal of this project was to develop a more accurate forecast based on historical sell-through data (i.e., product sold to consumers), which were recently made available. Forecast error was drastically reduced using the new forecasting method - 35% MAPE. A pilot was initiated with a major retail customer in order to test the new forecast model and determine the effects of a more transparent ordering partnership. The pilot is ongoing at the time of thesis completion.