Sales & inventory forecasting
Forecasting that shows its accuracy
Most forecasting tools hand you a confident number and quietly hide how often they are wrong. Ours reports the error rate it scored on your own data, so you know exactly how much weight to put on it.
No account. No card. Your figures never leave your browser.
What we make
Sales forecaster
Paste your sales history and get a forecast with a 95% confidence interval, a backtested accuracy score, scenario planning and a printable report. Runs entirely in your browser.
Open the forecasterInventory planning for Shopify
The same engine wired straight into a Shopify store: days of stock left per product, when to reorder and how much — with the forecast accuracy shown per SKU, so you know which recommendations to lean on. Not released yet.
Get notifiedHow we think about forecasts
It shows its work
Every forecast carries the error rate it scored against your own recent history. A number without an error bar is a guess wearing a suit.
It refuses to guess
Thin history, or a product that sells three times a month? It says so, instead of printing a confident figure you might order stock against.
It picks the model by testing
Trend and seasonal models are scored against periods held out of the fit. Whichever actually predicted them best is the one that gets used.
Nothing here is a black box
The methods are ordinary statistics, not a mystery: ordinary least squares, Holt-Winters seasonality, Bayesian shrinkage, bootstrap and Monte Carlo intervals. All of it is written up in plain English, because a forecast you cannot interrogate is a forecast you should not trust.