Every campaign in a run gets a verdict - the recommended action class - shown as a colour-coded chip.
The verdicts
Review (red, critical) - spending materially above its target CPA. A person should look at it.
Pause (red, critical) - spent money with zero conversions; a one-click pause is offered.
Scale (green, high) - hitting targets and constrained by budget; more budget would genuinely help.
Change (yellow, high) - ready to switch bidding strategy (typically onto tCPA/tROAS after the threshold is met).
Adjust (yellow, medium) - tighten or loosen an existing target.
Reduce (orange, medium) - budget should come down.
Consider (purple, low) - a soft recommendation; the data supports it but isn't definitive.
Hold (grey, none) - performing as expected; no action needed. Hold is a positive verdict - it means the engine looked and found nothing to fix.
How to triage a run:
Reds first - Pause and Review are money actively leaking.
Then greens - Scale is money being left on the table.
Then yellows - Change and Adjust are improvements, not emergencies.
Skim Considers; skip Holds.
RELATED ARTICLES
Reading a decision card
Everything behind a recommendation, on one card.
Reading the dashboard
Every account's state at a glance - pacing, CPA, verdicts, lifecycle mix.
One lever at a time
Bid and budget are never touched together - by design.