Creative fatigue metrics are the signals that tell you a Meta ad is wearing out, and the one that actually triggers the platform’s own alert is not frequency, it is cost per result. Meta’s business documentation marks an ad with a creative fatigue status when its cost per result reaches twice that of your past ads. Frequency, click-through rate and the first-time impression ratio are useful early symptoms, but they only carry meaning when they move together.
That single fact reframes the whole subject. You are not watching an exposure counter tick towards a magic number, you are watching a cost drift. This article gives the metric that really triggers the alert, the one measured decay law that exists at scale, the crossing of signals that flags a dying creative, and a detection cadence you can run on your own account.
Which metric actually triggers creative fatigue?
Cost per result, measured against your own recent baseline. The metric that circulates most, frequency, is a symptom rather than a trigger, and Meta does not publish any frequency threshold at all. What it does publish is a rule tied to money: an ad earns the creative fatigue status when its cost per result reaches roughly two times the cost per result of your past ads, with an intermediate status when it drifts above that baseline without doubling.
The practical consequence is that you should anchor every other metric to this one. A frequency of 4 on an ad whose cost per result is still flat is not a problem. A frequency of 2 on an ad whose cost per result is climbing towards twice your baseline is a problem. The counter tells you how much repetition is happening; the cost tells you whether that repetition is hurting you.
This is also why chasing a universal frequency cap fails. Cost per result already integrates frequency, competition, seasonality and creative quality into a single number that reflects what the auction is actually doing to you. It is the metric the platform acts on, so it is the metric that predicts what the platform will do with your ad.
At what frequency does ad fatigue start?
There is no measured answer, only convention. Practitioner benchmarks tend to flag prospecting campaigns once frequency creeps past 2.5 to 3 over a rolling seven-day window, while retargeting audiences absorb far more before they tire, as Adamigo and other operators report. These are heuristics passed from account to account, not thresholds backed by a published sample. Treat them as a prompt to look closer, never as proof.
The deeper problem is that average frequency hides its own distribution. A mean frequency of 2 sounds safe, yet it can conceal a segment of your audience seeing the ad eight or more times while most people see it once. Analytics at Meta reports that about 19 per cent of impressions are viewed five or more times by the same person within a 30-day window, and that mean exposure per creative sits at 4.2. An average smooths those heavy repeaters into invisibility, which is exactly where fatigue concentrates.
A better proxy is the first-time impression ratio, the share of daily impressions going to people who are seeing your ad for the first time. A healthy prospecting campaign typically runs between 65 and 80 per cent, and dropping below 50 per cent means more than half your budget is landing on people who have already seen the ad, according to practitioner benchmarks from Flighted. A ratio sliding towards 20 per cent is a stronger fatigue signal than any single frequency number, because it measures the flow of fresh eyeballs rather than an average that erases its own tails.
What CTR drop signals fatigue?
No fixed percentage on its own. The reliable signal is not one metric crossing a line, it is several metrics crossing each other. Triple Whale describes the clearest early warning as a frequency and CTR inversion: frequency climbing while CTR falls week over week, joined by rising CPM, a declining first-time impression ratio and engagement decay.
If you want thresholds to build an alert around, practitioner conventions offer a starting point. TheOptimizer treats a CTR sitting 20 to 30 per cent below baseline as the fatigue zone, and flags a 10 per cent CTR drop over seven days paired with a 15 per cent cost-per-acquisition rise as an early warning. Atria similarly points to CPM climbing 15 to 40 per cent over two weeks with no targeting change as a red flag. These are conventions, not laws, and the point of listing them is not to pick a winner but to show that the useful reading is always relational.
The discipline this imposes is simple. Wire an alert to the pattern, not to a lone metric: for example, CTR down 20 per cent week over week while frequency rises and the first-time impression ratio falls. A single line moving is often noise, a seasonal blip or a reporting artefact. Three lines moving in the fatigue direction at once is a decision.
What the one large-scale dataset actually shows
There is exactly one publicly documented, large-scale measurement of repeated exposure, and it describes a smooth decay rather than a cliff. Analytics at Meta models the fall in conversion probability as a law in (N+1)^-0.43, meaning performance drops continuously from the second exposure, hard at first then flattening. On the same data, conversion likelihood falls by about 45 per cent after four repeated exposures of the same creative.
Two details matter for how you read your metrics. First, that 45 per cent figure is about conversion probability, not click-through rate. Someone can keep clicking an ad they have seen four times while having become markedly less likely to buy, which is precisely why a stable CTR can mask a rotting cost per result. Second, applying anti-fatigue guidance lifted conversion rate by roughly 8 per cent on average for high-fatigue cases in Meta’s own test. That gain is real and measured, and it is modest, which is a healthier expectation than the doubling that vendors sometimes imply.
The academic work points the same way. A September 2025 arXiv paper, A Path Signature Framework for Detecting Creative Fatigue in Digital Advertising, reframes fatigue monitoring as a geometric change-detection problem on performance trajectories, and stresses that delayed detection translates directly into avoidable opportunity cost. The recurring theme across both the platform data and the research is that fatigue is a drift to be caught early, not a threshold to be hit.
How often should you refresh creatives?
On a signal-driven cadence rather than a fixed calendar. As a rough starting point, practitioners find top-of-funnel ads tire faster, around every 3 to 4 weeks, while bottom-of-funnel ads last longer, closer to 6 to 7 weeks, because their smaller, warmer audiences see fresh creative less often. But a date on the calendar is a blunt instrument next to the crossing of signals described above.
Refreshing also has a volume cost that the cadence question tends to ignore. Motion analysed more than 550,000 ads from over 6,000 advertisers, representing roughly 1.3 billion dollars of spend across Facebook and Instagram, and found that only about 5 to 8 per cent of creatives become real winners while around half never earn meaningful spend. Renewing before fatigue bites therefore means testing enough concepts to keep finding those winners, which is a budget question as much as a design one. The pressure only grows as reach gets more expensive: Meta CPM rose 20.03 per cent year on year across every vertical Triple Whale tracked, so a worn creative costs more to replace each year. We cover that cost side in detail in our guide to how much Facebook ads cost.
One caution on what refreshing means. The decay law is about repeated exposure to a message, so recolouring a button or recropping an image reproduces the same stimulus your audience already tuned out. A refresh that changes only the surface will fatigue at the same speed, because the thing that fatigued was never the pixels, it was the idea.
Creative fatigue versus audience saturation
These are different failures that look identical on a performance chart, and mixing them up wastes budget. Creative fatigue means your specific ad has worn out while the audience pool is still fresh. Audience saturation means you have exhausted the reachable audience within your targeting, whatever creative you run. LeadEnforce frames a clean test: if a new creative in the same audience restores CTR and lowers cost, the problem was fatigue; if performance only recovers when you broaden to a fresh audience, the problem was saturation.
The order of diagnosis matters because the wrong fix scales the wrong problem. Replace a creative when the real issue is saturation and the new ad fatigues just as fast on the same tired audience. Expand the audience when the real issue is a single dying creative and you have simply spread a fixable problem across more people. Saturation usually shows up when you push spend, which is why it belongs to the scaling conversation rather than the creative one; we unpack that mechanism in our piece on why ROAS drops when scaling.
In short
- Anchor to cost per result, not frequency. Meta assigns its creative fatigue status when an ad’s cost per result reaches twice your baseline. There is no official frequency threshold, so treat the popular 2.5 to 3 numbers as prompts, not proof.
- Read the crossing, not a single metric. Fatigue is CTR and the first-time impression ratio falling while CPM, frequency and cost per result rise, all together. One line moving is noise; three moving is a decision, and the measured decay is a smooth drift rather than a cliff.
- Separate fatigue from saturation before you act. If a new creative in the same audience revives performance, it was fatigue; if only a fresh audience helps, it was saturation. The wrong fix scales the wrong problem.
If your ads are slipping and you cannot tell whether it is the creative, the audience or the account structure giving way, that diagnosis is exactly what we run inside our B2B paid acquisition work. Book a diagnostic and we will read your cost per result, your first-time impression ratio and your creative cadence together, then hand you a refresh plan grounded in your own numbers.