Customer description‍

A national retail formula with more than 300 stores, active in FMCG. The supply chain is complex, with daily changing demand and a great deal of dependence on local conditions per location.

Challenge‍

Stores were struggling with structural shortages of popular products and the loss of slower-selling items. Inventory was centrally planned based on averages, so that local differences were not sufficiently exploited.

Solution‍

A solution was developed that provides an AI-driven forecasting model that estimates expected demand per store and per product group. This allowed inventory management to be tailored to local conditions, such as weather, purchasing behavior and promotions.

Approach

1. Collection of store data
   Sales figures, checkout dates, weather information, promotional calendars and seasonality per store were bundled into one data model.
2. Develop a prediction model
   For each product group, a machine learning model was trained on shopping behavior, with adjustments based on deviations and external influences.
3. Integration with ordering system
   The model was linked to the existing ordering system, so managers received suggestions based on expected sales.
4. Monitoring and Optimization
   Effects on availability, turnover and loss were continuously monitored. The model learned from anomalies and customer feedback.

Results

• 20% less loss of fresh produce and slow movers
• 13% higher availability of in-store runners
• Less manual work for branch managers
• Higher customer satisfaction due to better shelf availability

Learnings‍

With predictive AI, the retail formula got a grip on the fine-tuning of its inventory policy. The result: better margins, less waste and stores that respond smarter to customer behavior locally.

‍