Meta doesn't measure creative fatigue with frequency, and says you shouldn't either

Meta marks an ad Creative fatigue when its cost per result reaches twice what your past ads cost, and Creative limited when the cost is higher than your past ads but less than double. Neither status uses frequency. Meta's own analytics team published a section titled "Dangers of measuring Frequency" explaining why the metric cannot detect creative fatigue at all — and the status only appears on ad sets running a single creative, so plenty of accounts will never see it.

Ask when to refresh a creative and you get a number. One guide starts planning the refresh at a frequency of 2.0 for small audiences. Another puts the danger zone at 2.5, and performance degradation at 3.0. A third acts at 3.5 to 4 and tolerates 5 to 8 in retargeting. They are confident, they are specific, and they disagree with each other by a factor of four.

None of them is Meta's number, because Meta doesn't have one.

What Meta's fatigue status actually measures

Meta's help centre documents the rule behind the Delivery column status, and it is relative to your own history rather than to any constant:

Status	Meta's condition
Creative limited	cost per result is "more than ads you ran in the past but less than twice as much"
Creative fatigue	cost per result is "more than or equal to twice as much as ads you ran in the past"

Three details in that page matter more than the thresholds themselves.

It counts across your campaigns. Meta says it considers "all recent exposures of the ad's image or video, including those from other campaigns from your Page". Re-uploading the same asset under a new campaign does not give you a clean slate; the exposure history follows the image, not the ad.

It can warn you before you launch. If Meta predicts fatigue within the first 7 days of a campaign, it says so before you publish.

It arrives through opportunity score, Meta's recommendation surface, not as a standalone fatigue alarm — which is part of why so few advertisers have seen it.

Why frequency can't see fatigue

In May 2023, three analysts at Meta published the research behind all of this, and it contains a heading most of the industry has managed not to read: "Dangers of measuring Frequency". They give two reasons.

Frequency is measured at the wrong level. "Frequency is at an ad/ad set/campaign level. Fatigue happens at a creative level." If the same image runs in three ads, a viewer fatigues on the image while each ad reports a comfortable frequency of its own. Meta notes that users "can fatigue from ads across campaigns, placements, optimizations and even those with similar versions of the same creative."

Frequency is an average, and averages hide the people being hit hardest. Meta's worked example: 100,000 users see an ad once on day one, 50,000 of them see it twice on day two, 25,000 of them see it four times on day three. Total impressions divided by reach gives a frequency of 3. But the people being reached on day three have now seen it seven times. As Meta puts it, frequency "presents an average effect over a period of time and not the marginal effect".

So the metric the whole category rotates creative on is, by its author's account, aggregated at the wrong unit and smoothed in the wrong direction.

What the research actually found

The same paper is worth reading for the numbers that aren't thresholds.

Repetition is the normal case, not the failure case. "The mean count of user/creative previous exposures across all Meta ad impressions is 4.2 with over 19% of ad impressions having been seen more than five times", over a 30-day lookback. The average impression on Meta is already a repeat.

The decay has a shape, and it starts immediately. Controlling for audience saturation and for baseline conversion rates, Meta modelled click likelihood against prior exposures to the same creative: "the reduction in likelihood of a click follows the shape of (N+1)^-.43 where N is the number of times the user has seen the same creative prior to that impression (ie. indexing starting with zero)". Meta's own reading of that curve: "at 4 repeated exposures the associated likelihood of a conversion drops by about 45%." (The model predicts clicks; that sentence says conversions — Meta's wording, not ours.)

There is no wear-in. This is the finding that should change how people work. The belief that a creative needs a few exposures to warm up before it fatigues did not survive contact with the data: Meta "found no evidence for this effect with direct response objectives", and "clicks and conversions appear to become monotonically more expensive with repeated creative exposures". Brand objectives like ad recall may behave differently. For performance, every exposure after the first is worse than the one before it.

And adding creative causes the recovery. Meta ran a two-cell split test across roughly 26,000 cases and saw a dose-dependent result: the more fatigued the ad set, the bigger the conversion-rate improvement when a new creative was added. Their headline figure is an average 8% conversion-rate improvement in high-fatigue cases.

Note what that experiment did not test: swapping the tired creative out. Meta is explicit elsewhere that "keeping your original ad active instead of pausing or turning it off may maximize results". Adding beats replacing.

The catch: you may never see the status at all

Meta's fatigue recommendations only run on ad sets with one creative. The page excludes Advantage+ catalog ads (previously dynamic ads), dynamic creative and Advantage+ app campaigns, and says the feature is unavailable with the sales objective before an ad set is active.

Read that against how most accounts are now built. If you run dynamic creative, or Advantage+ catalog ads, the absence of a fatigue warning tells you nothing whatsoever — the diagnostic was never running. A quiet Delivery column is not evidence of a healthy creative.

A diagnostic that doesn't depend on a constant

Everything above points the same way: stop looking for a threshold and start comparing an ad against your own history.

1. Benchmark cost per result against your own past ads, which is the comparison Meta's status makes. Doubling is Meta's line for fatigue; your own tolerance is probably lower, and it should be a ratio to your baseline rather than a number you read somewhere.
2. Count exposures at the level of the creative, across campaigns. If the same image is live in four ad sets, no single ad's frequency describes what anyone has actually seen.
3. Treat frequency as context, not a trigger. It is a genuine input into how much repetition exists. It is not, per the people who built the system, able to tell you when a creative is spent.
4. Add, don't swap, and make the addition materially different. Meta's recommendation is a "new image or video that is materially different from the original creative". A recolour of the same layout is the same creative to a viewer who has seen it four times, which is the average.

That last point is where most creative refreshes quietly fail. Producing another variant is easy; producing one that a fatigued viewer experiences as new is the actual job — and it is a different job from producing more.

Sources

• About creative fatigue recommendations in Meta Ads Manager — Meta Business Help Center: the Creative limited and Creative fatigue cost-per-result conditions, cross-campaign exposure counting, the 7-day pre-publish prediction, the single-creative eligibility limits, and the "keeping your original ad active" note (accessed 2026-09-16)
• Creative Fatigue: How advertisers can improve performance by managing repeated exposures — Analytics at Meta, 10 May 2023: "Dangers of measuring Frequency", the 4.2 mean exposure count, the (N+1)^-.43 decay, the absence of a wear-in effect, and the ~26,000-case split test (accessed 2026-09-16)

---

Related: why every Meta safe-zone guide disagrees — the same pattern, where Meta's own page holds a number nobody copies, and the sixteen layouts almost every performance ad uses, which is where "materially different" starts.