Demand Forecasting for Grocery Chain

The problem

FreshFoods was throwing away 8% of perishable inventory due to spoilage while simultaneously experiencing frequent stockouts of popular items. Store managers ordered based on gut feel, leading to massive inconsistency. They needed store-level demand predictions they could trust.

What we built

We built a forecasting system that predicts demand for each SKU at each store 14 days ahead. The model incorporates historical sales, promotions, weather, local events, holidays, and competitor pricing. Forecasts update daily and integrate with their ordering system.

Tech stack

Results

Spoilage decreased 35% in the first year, saving $6M annually. Stockouts dropped from 12% of SKUs to 4%. Store managers now spend 80% less time on ordering decisions. The system paid for itself in 4 months.