Home improvement e-commerce · the same national retailer
Sales forecasting that knows the calendar, the map and the weather
Machine learningPredictive analyticsSQL ServernopCommerce data
The problem
Years of order history sat in the store’s database, but planning still worked from last year’s totals. Demand moves differently by region and by season, and events like Black Friday distort the averages for everyone.
What we built
A forecasting model over the store’s order history that predicts demand day by day and month by month, with separate components for festival and sale peaks, for each region, and for climate seasons by region.
The result
Forecasts the retailer can plan stock and campaigns against, instead of annual averages. The model also surfaced effects nobody was planning for, such as a sales rise in one region during hurricane season, now treated as a predictable seasonal peak. Figures are withheld under the client agreement.