What a budget change will return. Drawn from data, not guessed.

The curve: what each extra dollar returns

On Google, the curve comes from Google's own campaign forecast, corrected to your store's real revenue. On Meta, it is built from Meta's delivery estimate and the account's own history. The page says which one you are looking at. A rough estimate is always labeled as one.

Three points: now, target, peak

Current spend, proposed spend and the spend where profit peaks, each with the revenue it returns and what the last dollar earns at that point. The recommendation is the peak, unless one of your spend limits says otherwise.

How sure it is

Every curve comes with a confidence score and a range. A budget change can skip a strategist's review only when confidence is high and the change is inside your limits. A low-confidence change is suggested, never applied.

Your targets and limits

Your budgets and targets by channel and campaign are stored once and checked against every recommendation. A target can be reset from a lift test, so it is based on the revenue the ads caused rather than the revenue the platform claims.

When it refuses to forecast

When a campaign has too little data, the page says “not enough data to forecast”. A new campaign gets a single labeled estimate, not a curve that looks more certain than it is.

Check the prediction against what happened

The chosen budget and its prediction are written to the record. Thirty days later the real result sits next to it. Every forecast is scored, and each one improves the next.