Demand Forecasting Engine — National Retailer
61%→88%
Demand forecast accuracy
£4.2M→£0.9M
Annual stock write-offs
14%→4%
Peak period out-of-stock rate
22%
Reduction in total inventory holding cost
The Challenge
A national retailer with 280 stores was running demand planning on 18-month-old statistical models that couldn't incorporate online behaviour signals, competitor pricing, or local demand patterns. Forecast accuracy of 61% resulted in £4.2M annual stock write-offs and 14% out-of-stock rates during peak periods.
What We Built
We built a demand forecasting engine ingesting 40+ signals: POS data, web traffic, search trends, competitor pricing, weather, events, and promotional calendars. The model generates store-SKU-week level forecasts, flags confidence intervals for buyer review, and feeds directly into the replenishment system — with weekly model retraining on new data.
“We went from guessing to knowing. The forecasting engine sees signals our planners couldn't see manually — and the stock position shows it.”
— Chief Merchandising Officer
Want a Result Like This?
SmartPath AI builds and deploys production AI systems for enterprises. Schedule a strategy session to discuss your specific use case.
Schedule Strategy Session