Retail case study

Revenue forecasting across different stores and cities.


The client, a major B2B2C distributor of fresh food and consumables, wants to be able to accurately forecast the arrival of orders and pickups of online sales for different products in the supermarket: Fresh, PGC, Vegetables, all in 15 minute increments. At the same time, they want to be able to accurately forecast the revenue day by day for the different stores across different cities. Being able to accurately have these forecasts is crucial for avoiding waste, reduce costs, improve profitability and optimize staffing and logistics across the entire supermarket chain.

Solving the issue requires the following characteristics:

  1.  Accurately provide forecast in 15 minute increments for the pick up and sales of the different product groups
  2. Accurately provide forecast for the revenue of the day for each of the different stores


Predictive Layer gather the following data points:

  • History of the sales at each store
  • Realtime weather data
  • Realtime market information regarding the country, region

With this data, Genius Forecaster was able to begin the training process following the characteristics mentioned above: accuracy and flexibility in the horizon forecast.

Once trained, the solution was ready to be tested by the client in parallel with benchmarking against their existing forecast (in other words: validate the PL forecast against actual forecast). Client validated the accuracy of the system and is now waiting to integrate this in their future staffing engine. Genius Forecaster improved the accuracy of the standard models by more than 20%. It outlines a potential reduction of cost staffing exceeding 15% with an equivalent service quality.

Trusted by leading global brands

Shneider Electric
Arcelor Mittal
General Electric