More than 30% of the food in the Netherlands in the chain is wasted. The same goes up for the flower- and plant industry. There is practically no supermarket anymore that does not have fruit and vegetable and flower and plant shelves, but especially the phenomenon of high waste is a very big problem.
Accurate demand planning is rapidly gaining in importance in this rapidly changing world. The combination of dynamic customer demand, changing weather patterns, store and product characteristics, promotions and various events (such as holidays, Valentines day) makes it increasingly difficult to predict expected sales per product using human intuition and experience, because no account can be taken of the impact of all those factors.
The management believed that intelligent demand forecasting would be of great benefit to stay competitive in the future and solve two core problems: reducing the waste and thus increasing the gross margin and more-over making the forecasting process more scalable for the future.
Up to that point, the client had no formal demand forecasting in place other than gut-feel and experience, and so the first portion of the project involved setting up a robust demand forecasting system. SYMSON has an internally generated suite of forecasting tools which were put to work for this customer. The AutoML approach ensures that only the best model is chosen for the SKU in question, with every SKU having its own forecasting model.
The SYMSON data science team achieved an average of 87% accuracy during the pilot period which is a fantastic result. These forecasts are to be fed into the Demand Forecasting tool to help the category managers of the wholesaler decide which amount of product they should send to each store.
SYMSON enables this wholesaler to scale their demand forecasting with less manual labor and to minimize their waste and at their same time optimizing their sales in the retail stores.