An ad rarely dies on a Tuesday. It fades. The same creative that pulled a 1.4% click-through in its first week drifts to 0.7% in its third, the cost to reach a thousand people climbs, and the spend keeps going out the door because nothing in Ads Manager is broken— it’s just tired. On our own account, one ad did exactly that over nine days: frequency crept from 1.9 to 3.8, CTR roughly halved, and CPM rose about 38%. None of it tripped a warning, because from Meta’s side everything was working as configured.
The three numbers that say “tired”
Fatigue isn’t a vibe; it’s a shape in three metrics moving together. Any one of them alone can mislead — read them as a set.
Frequencyis how many times the average person has seen the ad. Past roughly 2.5–3 in a week on a cold-ish audience, you’re buying repeats, not reach. CTRfalling while frequency rises is the audience telling you they’ve seen it — a 30–40% drop off the first-week rate is the line most accounts feel. CPM creeping up as the other two move confirms it: Meta charges more to keep pushing a creative the auction likes less. One number wobbles; three moving in formation is fatigue.
Why the platform won’t tell you
Because nothing is wrong, by the platform’s definition. The ad is approved, the ad set is delivering, the budget is spending. A fatiguing ad and a healthy one look identical in the top-line numbers until you put them on a time axis and watch the slope. That read has to come from your side of the table — which is the whole reason to look at the trend, not just today’s row.
Is frequency past ~3 on this ad set, this week?
Ads Manager → add the Frequency column, set the date to last 7 days
High frequency is the earliest signal and the one Meta hides by default — the column isn’t shown unless you add it.
Is CTR down 30%+ from its first-week rate?
Ads Manager → Ad → chart view → CTR over time
Compare the trailing few days to the opening few, not to a benchmark. The audience’s own baseline is the honest comparison.
Is CPM climbing while CTR falls?
Breakdown → By day
Rising CPM with falling CTR is the confirmation — you’re paying more to reach people who respond less.
How many days has this exact creative been live?
Not the campaign — the creative. Age is context: a nine-day-old ad at frequency 3.8 is a different story from a two-day-old one.
Is there a second ad in the set to rotate to?
One ad means fatigue has nowhere to go but up. Two or three gives delivery somewhere to move when the first tires.
0 of 5 checked
What “fresh” actually has to change
Here’s where most refreshes waste the effort: they recolor the button, swap the background, nudge the headline — and the fatigue barely moves, because the audience recognises the same ad wearing a new coat. What resets fatigue is a genuinely different first impression: a new hook in the opening line, a different image or first video frame, a different format entirely (a static becoming a short video, a product shot becoming a person). The auction and the audience both reward the thing that reads as new in the first half- second, not the thing that’s technically a new asset ID.
So the fix ladder is: rotate to an existing second ad if you have one; if you don’t, make a real variant — new angle, new format — not a tint; and only widen the audience once the creative itself is genuinely different. Widening a stale creative just finds new people to tire out.
The nine-day slide above came from Admiral’s Meta audit reading our own account — it prices creative fatigue from your history and flags the ad before the CPM does. And it now closes the loop it used to only name: from the same brand it scanned, Admiral generates a fresh on-brand image or short video to test against the tired one, attaches it to the ad, and leaves it paused for your call. Rolling out account by account, with a monthly cap so it sharpens the creative without running up a bill.
Run the free audit