Foreviq
Monthly demand forecasts for every customer and product, built from sales history and open orders.
Image preview unavailable- forecasting methods compared on every product line
- 19
- error on a later period it never saw
- 4.6%
Most sales forecasts are last year's numbers with a gut feeling on top. They also count demand twice: once in the forecast, and again when the orders arrive.
Foreviq tries 19 methods on every customer and product line, from seasonal smoothing to machine learning, picks the best one on earlier months and checks it on a later period it never saw. Every forecast comes with an 80% planning range.
Open orders are netted against the forecast, so a fully ordered month counts once. Plans move from draft to review to approved, and the assistant can propose a new forecast but can't publish one without a person.
I built Foreviq with Python, FastAPI and StatsForecast behind a React interface. In local pilot runs, I checked imports, forecasting, order accounting, exports and backup recovery.
Image preview unavailable
Image preview unavailable
Image preview unavailable