One of the largest food distributors in the Middle East had difficulty forecasting demand and accurately placing orders to profitably fulfill demand across their distribution network.
Challenges included the accuracy of demand forecasts, and the long lead times on order placement. In addition, the forecasting process was lengthy and manual with limited functionality in Excel. This impacted the work effort, the quality of forecasts, and forecasting cycle times. Accuracy suffered because orders were typically placed with suppliers for a one-year period with only minor changes to delivery schedule over the year based on stock on hand. The forecast was also compromised because it was difficult to accommodate size and complexity differences between customers of different types, such as chains and independent retailers. The distributor used promotions to increase sales and clear inventory that was near the expiration date, but the basic, less sophisticated forecast, made it difficult to predict the impact of promotions. Forecasts were at a high level and did not incorporate SKU/customer combinations.
These challenges made it difficult to set optimal inventory levels, increasing the risk of excess stock, shortages of product, lower fill rates / service levels and increasing costs associated with products past their expiration dates.
This solution included:
Statistical forecasting using an automated method to select the best statistical model to generate the most accurate forecast with the following features:
