Reducing food waste with inventory forecasting & optimization

Forecasting the expected sales of ultra-fresh products – such as bread – poses a unique challenge compared to dry goods. While fresh items are often the highest priority for customers, decisions around production and supply are inherently riskier due to their short shelf life. Balancing supply and demand is crucial, as excess fresh products lead to waste. Standardized solutions don’t provide good enough predictions, leading to manual interventions at the local store level. This takes up a lot of store management time without necessarily improving daily forecast accuracy.

We created a scalable solution with 300+ underlying forecasting models, each uniquely tailored to specific store and bread type, all delivered within a unified framework. Models are adjusted to capture local trends and allow for substitution effects. This is achieved by seamlessly blending a top-down forecast for the entire product group with existing promotional forecasts and historical trends of non-promotion products. On top of this, a lost sales algorithm corrects for historic out-of-stock moments, enhancing the precision of this forecasting approach.

The system enables more efficient centralized decision-making, based on forecasts with a high degree of accuracy at a granular level. It minimizes both waste and lost sales by dynamically optimizing write-offs and stock-outs, all based on performance and targets. And by removing the need for manual intervention at store level, store employees have more time to focus on their key task: serving their customers with fresh bread.