Your customer is seeing the same ad for the 15th time

Yes — it is a documented phenomenon both in the academic literature and in the platform's own documentation. Repetition effects aren't linear: they climb to a point, then fall. Meta-analysing 312 effect sizes from experimental studies, Schmidt and Eisend found that attitude reaches its maximum at approximately ten exposures, while recall increases and does not level off before the eighth exposure (Schmidt & Eisend, 2015).

That number shouldn't be carried straight into your ads dashboard: the study pools laboratory experiments, and what it measures is attitude and recall, not purchases. The transferable finding is the shape of the curve, not the figure — as the same paper notes, the literature agrees on an inverted U-shaped curve for attitude and a logarithmic course for recall. So "repetition is bad" is wrong too: repetition has a productive range; the problem is running past it.

Field data points the same way. Braun and Moe modelled the individual impression histories of 5,803 people during a single automobile brand's ten-week display campaign in 2009, and showed that each repeat impression wears out a creative's effect, while that wearout is gradually restored as time passes (Braun & Moe, 2013). The data is neither cosmetics nor social — but the mechanism, wearout plus restoration during a hiatus, is modelled behaviour rather than folklore.

Meta defines it this way in its own help centre: creative fatigue occurs when an audience has seen the same creative too many times; people may be less likely to engage with your ad, which can lead to a higher cost per result (Meta Business Help Center).

There is no universal threshold — worth stating plainly, because the "never let frequency pass 3" rules circulating online rest on no source at all. By Meta's own definition, frequency is the average number of times each Meta account saw your ad, calculated as impressions divided by reach, and it is an estimated metric (Meta Business Help Center).

The word "average" is doing heavy lifting. A frequency of 3 does not mean everyone saw the ad three times; a large share may have seen it once while a small core has seen it fifteen times. The scenario in this article's title is precisely that tail, hiding under the average.

The same page notes frequency may average 1 to 2 per ad set or run much higher depending on budget, audience size and schedule; its actual advice is not to hit a number but to monitor frequency alongside your results — if performance begins to drop as frequency rises, your audience may be experiencing ad fatigue, and changing creative or targeting may be wise.

The arithmetic also explains why smaller brands hit this wall sooner: frequency = impressions / reach. Spend a meaningful daily budget against a 30,000-person retargeting pool and frequency climbs within days; the same budget against a three-million-person audience stays under 1 for months. Because retargeting audiences are small by definition, creative fatigue shows up there first.

In the Delivery column. When Meta believes your audience has seen the same ad too many times, your ad set or ad status shows either Creative limited or Creative fatigue. What separates them is cost: when cost per result is higher than ads you've run in the past but less than twice as high, you see Creative limited; when it is at least twice as high, you see Creative fatigue (Meta Business Help Center).

Two details on that page matter in practice. First, Meta considers all recent exposures of the ad's image or video, including those from other campaigns from your Page — so running the same video across three campaigns accumulates fatigue in aggregate. Second, the feature only covers ad sets with a single creative; Advantage+ catalogue ads, dynamic creative and Advantage+ app campaigns are excluded. If fatigue is predicted for the first seven days, the warning can also arrive before you publish.

Not every drop is fatigue

Seeing a decline in the dashboard and a creative wearing out are not the same event. Meta notes that under the greater daily budget flexibility it is gradually introducing to some ad accounts, spend may run up to 75% over the daily budget on some days, that breaking results down by hour can show oscillations which average out over time, and that results are best read at campaign level across a full week (Meta Business Help Center). A two-day dip is usually noise, not fatigue.

The second possibility is more uncomfortable: the drop may come from measurement rather than creative. Missing purchase events or broken attribution make performance look worse than it is and get a healthy ad switched off. We covered where measurement quietly breaks in our piece on Meta Pixel measurement errors.

Because how much repetition an audience tolerates depends on how well it knows the brand. Across two experiments using television ads and computer internet ads, Campbell and Keller found that repetition of advertising attributed to an unfamiliar brand showed decreased effectiveness, while the same advertising attributed to a known, familiar brand had its repetition wearout postponed (Campbell & Keller, 2003).

The mechanism is worth noting too: the authors report that negative thoughts about tactic inappropriateness arose with repetition, particularly for an ad for an unfamiliar brand, and that these partly drove the decrease in effectiveness. A familiar brand's persistence reads as a reminder; an unknown brand's reads as pestering.

The practical consequence for a new cosmetics brand: a repetition level a known brand carries comfortably can be your fatigue point. Creative production cadence has to be planned as part of the budget rather than as a luxury — it's also the precondition for sustaining a creative testing loop at all.

Meta itself recommends three routes, and they're alternatives to one another: create another ad with a new image or video that is materially different from the original, expand your audience to reach people who haven't seen the ad too much yet, or try Advantage+ creative, which automatically produces variations from a single image or video (Meta Business Help Center).

The same page carries a note that runs against most brands' reflex: keeping your original ad active instead of pausing or turning it off may maximise results. "Refreshing" and "killing the previous ad" are not the same move.

Rotation has a measured payoff as well. Braun and Moe simulated a setup that chooses which creative to serve next based on a person's impression history, and reported a 12.7% increase in expected website visits and a 13.8% increase in expected conversions for their data (Braun & Moe, 2013). That is not a forecast for a cosmetics brand on Instagram — it is a model result from one brand's display campaign. The transferable idea is not the percentages but that sequencing itself is a lever.

Resting a creative is a tool too

In the same model, wearout isn't permanent: each week that passes restores 2.7% of the campaign-level wearout and 8.8% of the wearout specific to a given creative. So parking a creative that worked and bringing it back later is, on this model, a defensible move rather than sentimentality.

Timing matters on its own. Using an individual-level data set built from online field experiments, Sahni found that the likelihood of a product's purchase increases when its ads are spread apart rather than bunched together — even when spreading them apart shifts some ads away from the purchase occasion (Sahni, 2015). Those experiments ran on a restaurant search platform, so the finding doesn't transfer one-to-one to beauty e-commerce, but it does show that compressing a budget into three days versus spreading it over two weeks is not a neutral choice.

• Read frequency next to cost per result, never on its own. The decision signal isn't "frequency hit 4", it's "cost per purchase is rising as frequency rises".
• Look weekly. At campaign level, over a full week; hourly and daily breakdowns will send you chasing false alarms.
• Watch the Delivery column: Creative limited is an early warning, Creative fatigue means cost per result has already doubled.
• If you run the same video across campaigns, remember fatigue accumulates across them — count your creative inventory at brand level, not campaign level.
• When you refresh, produce something genuinely different: not new music or a new cover frame, but a different angle, a different face, a different opening. Where to start is covered in our five-hook test piece.
• Put creative production on a line. The cure for fatigue is inventory, and for small budgets the format that makes it sustainable is usually UGC — if you want help on the production side, our creative production service exists to build exactly that line.
• Archive winners instead of deleting them. A creative brought back after a few weeks' rest costs less than one produced from scratch.
• Check the audience side too: exclude purchasers, widen the lookalike. Sometimes what's exhausted isn't the creative but the audience.
• Verify measurement before you switch anything off. Is the drop real, or are conversions going missing? Every decision made without that distinction is blind.

One frame sits under all of it: creative fatigue isn't a mistake, it's the natural depreciation of the attention you bought. Performance falls not because the ad got worse but because your audience learned it. If you want the wider view of how the ad structure fits together, our Instagram ads guide for beauty brands is a good starting point.