SKU-Based Forecasting National Sales for an Agricultural Manufacturing Company

This research was developed on a Mexican-based, global company manufacturer of agricultural and construction heavy equipment, selling its products throughout the country. Due to a lack of standardization, the year-basis forecasting process is developed only by family products and neither time series nor forecasting methods were previously used. Research’s objective was to design and implement adequate forecasting models for the 80/20 of national sales SKUs, as developing those, would reduce the difference between the predicted and actual values at a product level (SKU). Throughout its implementation, the general process for creating the adequate forecast for each SKU was established. For each SKU, the forecasting model was defined and developed through data analysis and application of specialized statistical commercial software. In order to validate the developed forecast models, the results were compared to monthly actual sales and to the previously forecasts made by the firm’s demand planner: comparing the results of his forecasted models, the proposed models showed a better result in 78% of the forecasted product families. Additionally, the time of the forecasting process was reduced by 33%.