Every ad gets exactly one label, decided by the same rule for everyone. It is never a guess and never tuned per customer.
The labels
- Learning. The ad is less than 14 days old, or its budget changed less than 14 days ago. Ad platforms need time to settle after either, so it is too early to judge.
- Too early. Past learning, but there are not enough orders to trust the number: fewer than 30 matched orders and spend under 4 times your average order value. Klirry says how many more days it needs.
- Pause. Enough data, and the ad's return is below your break-even point. Every day it keeps running costs you money.
- Wait. Enough data, and the ad is right around break-even, within 0.15 ROAS of it (ROAS is what you sell for every 1 spent on ads). Worth a look, not urgent.
- Keep. Enough data, and the ad is comfortably above break-even. This is where you would consider spending more.
All of this looks at your last 14 days: short enough to catch a problem quickly, long enough not to overreact to one bad day. The same numbers are on the methodology page.
What you can do with a decision
- Mark done records that you acted on it.
- Snooze 7d hides it for seven days.
Both are written to your decision history. Klirry itself never changes anything in your ad account; see why we never touch your ads.