Skip to content
← Growth Brain

A recommendation you can argue with.

Every recommendation carries the proposed action, its expected impact on the Goal's objective, the evidence behind it, a confidence level, the spend it affects, its risk and constraint checks, and a rollback plan where one exists. The expected impact comes from the budget engine and the business metrics service; the model assembles the case and writes the explanation. Constraint checks run against the Campaign Constitution before you see the proposal, so nothing is shown that could not be executed. Rejecting a recommendation is recorded with the same detail as accepting one.

Confidence is stated, not implied

Each proposal names the data it rests on and how much of it there is. A budget shift backed by three weeks of collected revenue reads differently from one backed by four days of platform-reported conversions, and the recommendation says which it is. Low confidence does not hide the proposal; it changes what the Constitution permits next.

Affected spend is bounded up front

A recommendation names the amount it moves and the campaigns it touches before you approve anything. Protected campaigns are excluded at proposal time, not caught at execution. The daily movement limit caps what a single accepted action can change.

Rollback is part of the proposal

Where an action is reversible — budget splits, audience suppression, creative rotation, landing-page routing — the rollback plan is written before execution. Where it is not reversible, the recommendation says so plainly. Reversibility is one of the inputs to whether Guarded Autopilot may act without an approval.

In this capability

  • Expected objective impact
  • Evidence and confidence
  • Constraint pre-checks
  • Rollback plans

One goal. Every channel. Real business truth.

Connect your ad accounts, say what a good customer is worth, and let the Growth Brain work the objective inside limits you set.

No ad spend commitment · Connect one channel to start · [email protected]