How often should you refresh Meta ad creative according to Meta?

Meta has published no cadence in days or weeks. Its 2023 Analytics at Meta study measured the effect per exposure, not per calendar day: the likelihood of a click falls with the number of times the same person has already seen the same creative, following (N+1)^-0.43, which the authors summarise as the likelihood of a conversion dropping by about 45% at four repeated exposures. A small retargeting audience can reach four exposures in days while broad prospecting takes weeks, so watch the Delivery column status rather than the calendar.

What frequency number means an ad is fatigued?

The Meta study argues against using frequency for this at all. It gives two reasons: frequency is reported at the ad, ad set or campaign level while fatigue happens at the creative level, and frequency is an average over a period rather than the marginal effect of the next exposure. Meta's Ads Manager status publishes no frequency number either. It appears when Meta believes your audience has seen the same ad too many times, counting all recent exposures of the image or video, and cost per result against your past ads decides the label: more than your past ads but less than twice as much shows Creative limited, and twice as much or more shows Creative fatigue.

Should you pause a fatigued ad or add a new one?

Add. The study found that adding a new creative of equal or greater quality to a fatigued ad set produced a dose-dependent improvement in conversion rate, on average 8% for high-fatigue cases, in a split test of roughly 26,000 cases. Meta's Help Centre recommendation for the Creative fatigue status is to create a new ad with a materially different image or video, and it adds that keeping the original ad active instead of pausing it may maximise results.

Does a new background or caption on the same AI video count as new creative?

Probably not, and neither Meta document defines the line. The study says users fatigue from ads with similar versions of the same creative, and the Help Centre asks for a video that is materially different from the original. It also states that Meta counts all recent exposures of the ad's image or video, including those from other campaigns from your Page. A safe reading is that a new headline, caption or script over the same frames is a version of one creative, while a new presenter or setting with a new opening frame and hook is a second creative.

Is there a wear-in period before creative fatigue starts?

Not for direct response, according to the study. Meta's authors looked for the wear-in effect that marketers commonly assume and found no evidence for it with direct response objectives; clicks and conversions appeared to become monotonically more expensive with repeated exposures. They note that brand objectives such as ad recall may behave differently, and that repetition still has value when it is spread across a large number of different-looking ads.

Creative Fatigue: What Meta's Study Actually Says

Quick answer:Meta's published study of creative fatigue, a 10 May 2023 post from its Analytics team, measured the effect per exposure, not per calendar day. It found that the likelihood of a click falls with each repeat view of the same creative along the curve (N+1)^-0.43, which the authors summarise as the likelihood of a conversion dropping by about 45% at four repeated exposures. The average Meta impression lands on someone who has already seen that creative 4.2 times within 30 days. Adding a new creative to a fatigued ad set lifted conversion rate by 8% on average for high-fatigue cases in a 26,000-case split test. It gives no refresh cadence in days or weeks and argues against using frequency as the signal. The week-counted refresh cadences and frequency triggers you see on ranking blogs are not in the paper.

What the paper actually measured

The document is Creative Fatigue: How advertisers can improve performance by managing repeated exposures, published by Analytics at Meta on Medium and credited to three authors, Lucas J., Alex D. and Matt M. Most of it is methodology. The definition comes first: "We define creative fatigue as the effect that occurs due to users seeing the same creative (the visual elements from an ad) repeatedly." The parenthetical matters more than it looks; we come back to it below.

Four findings carry the argument.

Finding	What the paper says	How it was measured
Repetition is the norm	"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."	Exposure counts over a 30-day lookback window, tracked per creative rather than per ad
Decay per exposure	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)". "This means that at 4 repeated exposures the associated likelihood of a conversion drops by about 45%."	Logistic regression on clicks matched to impressions, controlling for ad-set and account exposures (audience saturation), estimated CTR and first-impression CTR
No wear-in for direct response	"However, the above analysis (and further analysis looking for this) found no evidence for this effect with direct response objectives."	Same regression plus further analysis; the authors add that brand objectives like ad recall may behave differently
Adding creative fixes it	"Guidance to reduce fatigue can improve conversion rate by on average 8% for high fatigue cases (as shown in the test below)."	Two-cell split test, two seven-day phases, "across roughly 26 thousand cases"; the lift was dose-dependent, larger where fatigue was higher

Two qualifications rarely make it into summaries. The decay coefficient was "roughly the same magnitude and statistically significant" across static, video and carousel formats and across click, conversion and video-view objectives, so video is not exempt. And repetition "is especially prevalent with ads that optimize for link clicks and offsite conversion events", while brand advertisers see "dramatically less repetition of this sort in part due to frequency caps that advertisers impose." If you buy sales campaigns for an ecommerce store, you are in the high-repetition group by construction.

What the paper does not say

It never states a number of days. It never states a frequency threshold. It does the opposite on frequency, spending a whole section on why not to use it: "Frequency is at an ad/ad set/campaign level. Fatigue happens at a creative level." And frequency "presents an average effect over a period of time and not the marginal effect". The paper illustrates that with an ad that reports a frequency of 3 while the people it is actually reaching on day three have seen it seven times.

Ranking pages for the refresh-cadence query cite this study for the 4.2 and 45% figures and then supply week-counted cadences and 7-day frequency triggers that the study does not contain, printed next to Meta's figures so the two appear to share a source.

The paper cannot give a cadence because its unit is exposures per person, and the days it takes one person to reach four exposures depend on budget, audience size, placements and how many other ads from your Page reuse the same video. A small retargeting audience gets there in days; broad prospecting takes weeks. A calendar rule cannot be right for both, which is why Meta built a status instead.

What Ads Manager flags, and the rule it uses

The operational version of the study lives on a Help Centre page, About creative fatigue recommendations in Meta Ads Manager. It sets the status on two conditions. First, it appears only "When we believe that your audience has seen the same ad too many times", counting all recent exposures of the video across your Page. Second, cost per result against your own past ads picks the label: "When cost per result is more than ads that you've ran in the past, but less than twice as much, you will see a Creative limited status." And: "When cost per result is more than or equal to twice as much as ads that you ran in the past, you will see a Creative fatigue status."

Three details on that page decide how useful the status is to you.

It counts the video, not the ad."We consider all recent exposures of the ad's image or video, including those from other campaigns from your Page." Duplicating a winning AI video into five ad sets does not give you five fresh creatives; it gives one creative five ways to fatigue.
It is narrower than it sounds."This feature is only available for ad sets with one creative except those with Advantage+ catalogue ads (previously dynamic ads), dynamic creative or Meta Advantage+ app campaigns." It is also "not available with the sales objective before an ad set is active." An ad set holding ten variants does not get the status at all, so a buyer who batches variants into one ad set is measuring fatigue blind.
The fix is additive.The first recommendation is "Create a new ad with a new image or video that is materially different from the original creative." The note under it: "Keeping your original ad active instead of pausing or turning it off may maximise results." That matches the study, whose split test added a creative rather than swapping one.

One structural note for anyone who used the flexible format to keep several videos in one ad. Meta's flexible ad format pagestill lists "help prevent ad fatigue" as a benefit of grouping media, and it now opens with: "Starting in March 2026, the flexible format will no longer be available in Ad setup." The remaining lever the study and the Help Centre both point at is the same one: more distinct creatives, added over time. How to lay a batch of variants across ad sets without that container is in testing AI video variants on Meta after the flexible format.

What counts as a different creative

This is the question the study raises and neither Meta document answers. The paper says people fatigue from ads with "similar versions of the same creative", and the Help Centre asks for a video that is "materially different". No test, no threshold. What follows is our reading, and we have not measured it either.

The definition's parenthetical, "the visual elements from an ad", is the clue. A new headline, primary text or caption on the same video is not a new creative under it, and a new script read by the same presenter over the same frames is close to a caption change. The AI-specific trap is the reverse assumption: keeping actor, setting and opening frame constant for brand consistency produces files that differ to a diff tool and not to a viewer. Our working line is that a second creative changes the presenter or the setting and the opening frame, and pairs that with a new hook. That is a rule of thumb; the Delivery column is the only feedback you get on whether it held. One complication: with Advantage+ creative enhancements on, Meta itself varies the delivered version per viewer, which the Advantage+ creative post walks through.

The paper's conclusion is the design constraint: repetition "is optimally distributed across a large number of different looking ads." That is a different brief from "make ten versions of the winner", and closer to what the pre-launch sample size guide asks for. The hook formulas are the cheapest axis to vary if each variant has to look different to a viewer rather than to a diff tool.

What the study means for a batch of variants

Read against the study, a batch of ten renders says one thing. If ten variants share one actor, one script and one hook, Meta's rule says you bought one creative ten times. The batch only buys fatigue headroom if the ten look different to a viewer, which is a scripting cost before it is a rendering cost. Per-model render costs are in the full cost benchmark.

A refresh plan that only uses measured facts

Watch the Delivery column, not the frequency column. The status appears once Meta judges the audience has seen the creative too many times; Creative limited then means cost per result is above your history but under 2x, Creative fatigue means 2x or more. Both are lagging once the campaign is live, because they are computed from cost already paid; the only leading signal is the pre-publish warning Meta shows when it predicts fatigue inside the first seven days.
Keep one creative per ad set so the status can appear, or accept that a multi-creative ad set will not show it and track cost per result per ad yourself.
When it appears, add a creative rather than pausing the fatigued one. The split test measured addition, and the Help Centre says keeping the original active may maximise results. Keep a reserve of finished variants for that moment; the Black Friday calendar shows how far ahead that reserve has to be built.
Make the addition materially different: a new presenter or setting plus a new opening frame and hook. Count headline, caption and script-only changes as zero new creative.
Stop reusing one video across campaigns as though each were fresh. Exposures are counted per video across your whole Page.
Expect no wear-in. For a direct response objective the first exposure is the best one, so a variant that underperforms on its first thousand impressions is not warming up.
If new creative is unaffordable, the paper's alternative is reach: "Alternatively to refreshing creative, targeting new audiences could also resolve this problem." It adds that audiences are finite, which is why creative is the lever it recommends.

None of this tells you to refresh on a Tuesday. Meta measured a per-exposure decay and a per-creative fix, and the only schedule it implies is the one your Delivery column writes for you.

Sources

Analytics at Meta, "Creative Fatigue: How advertisers can improve performance by managing repeated exposures", Medium, 10 May 2023 (authors credited as Lucas J., Alex D. and Matt M.)
Meta Business Help Centre, "About creative fatigue recommendations in Meta Ads Manager"
Meta Business Help Centre, "About the flexible ad format"

All three Meta documents were read directly on 4 September 2026. Help Centre pages change without notice; the quotations above are how you check whether this one has gone stale.

Fatigue is the post-launch half of a testing programme. The creative testing framework covers the pre-launch half, and the 30-second script structure is the fastest way to write the new hook a second creative needs. If the bottleneck is making the second creative rather than knowing when, UGC Vids AI is built for that: Starter is $49 a month for 5,000 credits and the first 3 days are free.

Creative Fatigue: What Meta's Own Study Says vs What the Refresh-Cadence Blogs Say