How do you spot Meta ad creative fatigue before performance drops?
Updated
Short answer
Watch frequency and engagement on each ad, not only cost. Fatigue usually shows as rising frequency with falling click-through rate and hook rate, followed by rising cost per result. If an ad shows that pattern for several days while audience, budget and landing page stay the same, plan a refresh that keeps the idea and changes how the ad opens.
What creative fatigue is
Fatigue happens when the same people see the same ad repeatedly and respond less each time. Delivery then has to pay more to find people who still respond, so cost per result climbs.
It differs from a change outside the ad, such as a seasonal dip, an expired offer or a tracking problem. The fix for those is not a new creative.
Signals to track for each ad
- Frequency rising across the reporting window.
- Click-through rate falling against the ad's own earlier baseline.
- Hook rate (3-second video plays divided by impressions) falling on video ads.
- Cost per result rising while budget, audience and landing page stayed the same.
- Comments turning repetitive or negative.
Rule out other causes first
Check for budget changes, edits that restarted learning, landing page or checkout changes, stock problems, a competitor promotion and tracking changes.
If several unrelated ads drop on the same day, the cause is more likely outside the creative.
How to refresh without losing what worked
Keep the mechanism that drove results: the angle, the proof and the offer. Change the way in: a new hook, a new opening visual, a different creator or a different format.
Change one thing per refresh when you want to know which change helped. Keep several distinct concepts live so one tired ad does not carry the account.
Where Critiqa fits
Critiqa's Creative Multiplier starts from a source creative, records what is locked and what may change, and builds controlled variations, including a Refresh aimed at fatigue. Connected performance evidence can include spend, CTR, hook rate and hold rate.