Retail demand forecasting across promotions, shortages and new products

SupChains describes a retail proof of concept spanning about 150 stores, 500 products and 332 million forecast points. The model used promotions, prices, shortages, new-product launches and store opening hours; the client reportedly measured a 33% forecast-error reduction versus its existing provider.

The case highlights the importance of a controlled forecast benchmark, client-owned validation, treatment of censored demand during stockouts and scalable data pipelines before production replacement.