What model does this forecaster use?

Holt-Winters exponential smoothing with damped trend by default, additive seasonal component when 24 or more months of history are provided. Below 3 months of data the tool degrades to honest compound-growth projection (single rate extrapolation) and emits no fabricated confidence intervals.

Where do the confidence bands come from?

The 80% prediction interval is derived from the residual variance of the fitted Holt-Winters model. It describes model uncertainty given the fitted parameters. It does not capture parameter risk, regime shifts, or external shocks.

Is the forecast generated by AI?

Forecast numbers come from a deterministic Holt-Winters engine fit to your monthly history. Felix provides the narrative interpretation. Numbers from the model, narrative from the agent.

How much history do I need?

3 months minimum for the statistical mode. 6 to 12 months yields tighter confidence bands. 24 months unlocks the seasonal variant. Below 3 months the tool runs an honest compound-growth projection that uses a single growth rate, with no confidence intervals.

What happens when I connect Shopify?

The Shopify mode pulls 12 or 24 months of order history from your store, sums revenue per calendar month, and pre-fills the form for review. You see the synced data before fitting. Months with no orders emit explicit zeroes so the fitter sees an unbroken series.

Can the forecast be wrong?

Yes. The model assumes the historical pattern continues. External shocks, ad-spend changes, launches, and unusual holiday timing are not modelled. The 80% prediction interval bounds typical noise; outliers can fall outside it.

Yes. Three uses per day without signup, ten per day with a free account, unlimited with Pro. Inputs are not retained beyond the session.

IndexSee which brands AI assistants actually recommend.Who AI actually recommends.See all

Monthly Revenue Forecast

See where your trajectory leads, with honest confidence bands. Holt-Winters fit on your monthly history with 80% prediction intervals from residual variance. Numbers from the model, narrative from the agent.

FelixForecasting