Forecast of sales volumes for a supermarket chain

We support a supermarket chain in making its supply chain fit. Pacemaker Forecast makes it possible to forecast the sales volume of over 25,000 products in wholesale and retail branches of a supermarket chain more precisely.

Challenge

The customer operates over 10,000 wholesale and retail stores. Accurate forecasts of the sales volumes of the entire product range of the individual branches are required in order to optimally design the supply chain and the associated planning processes, from purchasing to action planning to resource planning in the warehouse.

The wide variety of products offered by a full-range retailer in retail and wholesale presents various challenges in preparing reliable forecasts. Extremely different sales patterns and varying sales frequencies of the individual product range items require a variety of different methods for a reliable forecast.

The different sales frequency of the individual articles represents a particular difficulty. This is compounded by the fact that many assortment items have varying and, in some cases, extremely short sales histories, such as newly introduced items. The “one-size-fits-all” solution currently used reveals problems, especially with rather low-traffic items. As part of the project, it should therefore be evaluated for two selected retail branches and two wholesale department stores whether the classic forecasting methods previously provided by SAP could be improved by using machine learning methods.

Approach

To improve existing forecasts, a variety of models were trained on the entire product range of two wholesale and retail branches in order to identify the optimal forecast model for each individual item. In addition to classic influencing factors such as promotional periods, holidays and special sales periods (e.g. pre-Christmas period), the models were also expanded to include weather effects.

Upshot

By choosing the right model, the forecast error for the entire range of warehouses (wholesale) could be improved by up to 4% on average compared to an SAP forecast. At individual item level, significantly more significant improvements were achieved, particularly in the B2C retail business. Here, the average prediction error was improved by 6 — 9%. The noticeably improved forecasts help customers better plan downstream processes in purchasing and logistics.

Jetzt unverbindliche Erstberatung vereinbaren

Ganz unabhängig davon, wo Sie gerade stehen. Unser Team steht Ihnen gerne in einem kostenlosen Erstgespräch zur Verfügung. Gemeinsam schauen wir in knapp 30 Minuten auf Ihre Herausforderungen und unsere Lösung.