Ad fatigue statistics: which numbers are real, and which ones everyone is copying from each other
Most ad fatigue statistics on the internet trace back to nothing. Here are the figures that have an actual source, the ones that do not, and how to generate the only numbers that matter, which are your own.
Very few widely quoted ad fatigue statistics survive being traced to a source. The figures that do have documented provenance come from the platform itself: Meta defines Creative Limited as an ad's cost per result rising above previous ads, and Creative Fatigue as cost per result reaching or exceeding twice previous ads, and both labels apply only to ad sets running a single creative. Almost everything else in circulation, including the widely repeated claim that a specific percentage of ad spend is wasted on fatigued creative, is an estimate passed between articles without an original study behind it. The only statistics that reliably describe your account are the ones computed from your account: each ad's CTR against its own trailing baseline, frequency slope, and reach saturation.
I went looking for the source of an ad fatigue statistic I had been repeating. I did not find one. I found four articles citing each other in a small circle, and a fifth that cited a webinar that no longer exists.
This post is what I would have wanted at the start of that afternoon: which numbers are documented, which are folklore, and how to produce figures that actually describe your account rather than someone's rounded guess.
The numbers with a real source
These come from Meta's own documentation, which makes them verifiable rather than merely popular. They are worth knowing precisely because they define what the platform will and will not tell you.
- Creative Limited appears when an ad's cost per result is higher than your previous ads, but less than double.
- Creative Fatigue appears when cost per result reaches or exceeds twice your previous ads.
- Both labels apply only to ad sets running a single creative, with some Advantage+ catalog exceptions.
Three sentences, and they tell you more about detecting fatigue on Meta than most statistics roundups manage in two thousand words. The thresholds are defined in terms of cost per result, and cost per result is the last thing to move. That is not a statistic about fatigue, it is a statistic about the detector, and it is the one that should change your behaviour.
The number I am not going to pretend is a statistic
You will see a figure quoted for the share of ad spend wasted on fatigued creative. Our own homepage quotes a range for it. It is an industry estimate, it circulates without a traceable original study, and I am not going to dress it up as measured fact in a post about which numbers are real.
It is directionally useful for sizing the problem. It is not evidence. If you are putting a number in a deck for a client or a board, that distinction is the difference between a credible slide and one that falls apart when somebody asks where it came from.
A statistic without a source is an opinion wearing a lab coat. This industry owns a great many lab coats.
Why benchmark statistics fail on your account specifically
Even honestly-produced aggregate benchmarks have a structural problem: they describe a distribution you are not necessarily in.
A 0.8 percent click-through rate is healthy on one account and an emergency on another. It depends on the offer, the audience temperature, the placement mix, the format, and the country. An industry average CTR tells you where the middle of somebody else's sample sits. It does not tell you whether your ad is declining, which is the only question fatigue detection is actually asking.
This is why every threshold worth using is relative. Not "CTR below X is bad" but "CTR down 30 percent against this ad's own trailing fortnight is bad." The second one survives being moved between accounts. The first one does not.
The four statistics worth computing for yourself
These are the ones that describe your account. All four can be built from ad-level insights you already have access to.
- CTR decay: the mean CTR of the last 3 days against the median of the trailing 14 days, excluding those 3. Expressed as a percentage change, this is your earliest reliable warning.
- Frequency velocity: the slope of average frequency over the last 5 days. The slope matters more than the level, because a frequency of 3.2 that is climbing fast is a different situation from a 3.2 that has been stable for a month.
- Reach saturation: how much of your delivery is going to people who have already seen the ad. Rising saturation across every ad at once is audience exhaustion rather than creative fatigue.
- Days from CTR turn to CPA move: measure this once on your own historical data. It is the single most useful number you will ever have about your account, because it tells you exactly how much warning you actually get.
That last one deserves the emphasis. Everyone repeats that leading indicators move before cost per result. Almost nobody measures how far before, on their own account, which is the number that determines whether a weekly check is sufficient or hopeless.
How to compute the fourth one
Pick five ads that ran their full lifecycle and were eventually paused or replaced. For each, find the date CTR began a sustained decline against its trailing baseline, and the date cost per result crossed a threshold you would have acted on. Subtract. Do it five times and take the median.
It is an afternoon of work with an exported CSV. What you get back is a number you can defend in any room, specific to your account, that tells you how often you actually need to be looking. That beats every aggregated benchmark in this article, including the ones with sources.
The honest summary
There is no good public dataset on ad fatigue. The platform documents its own thresholds and those are worth knowing. Everything else circulating as a statistic is either an estimate being laundered through repetition, or a benchmark describing a population that may not include you.
Which is inconvenient for anyone writing a statistics roundup, and clarifying for anyone running an account. Your data is the only data that describes your account. It is also, conveniently, the data you already have.
The Fadar backtest computes all four of these across your last 90 days: which ads faded, the date each one turned, how many days before cost per result followed, and the euros in the gap. It is free, it is read-only, and if your account turns out to be healthy that is a real result rather than a failed sales call.
Fair questions.
What percentage of ad spend is wasted on ad fatigue?
The figures circulating for this are industry estimates rather than findings from a traceable study, and they should be described that way. The honest answer is that the share is highly account-specific, and the only reliable way to size it is to replay your own history: identify ads that continued running after their CTR began declining, and total the spend between that date and the date they were replaced.
Are there reliable ad fatigue statistics?
Few. The most reliable published figures are Meta's own definitions of its delivery labels: Creative Limited when cost per result exceeds previous ads, Creative Fatigue at twice previous ads, both only on single-creative ad sets. Most other statistics in circulation cannot be traced to an original study.
What is a good CTR before an ad is considered fatigued?
There is no absolute threshold worth using. A 0.8 percent CTR is healthy on one account and a warning on another, depending on offer, audience temperature, placement and format. Useful thresholds are relative: a sustained decline against that specific ad's own trailing baseline, typically measured over a 3-day window against the previous 14.
How long before CPA rises does CTR start falling?
It varies by account, which is why it is worth measuring rather than quoting. Take five ads that ran a full lifecycle, find the date CTR began a sustained decline and the date cost per result crossed a level you would have acted on, subtract, and take the median. That single number tells you how often you genuinely need to check.
Where do ad fatigue statistics usually come from?
Frequently from other articles. Tracing widely quoted figures often leads to a short chain of posts citing each other, ending at a source that is no longer available or was never a study. Before using a statistic in a client deck, follow it to its origin, because someone eventually will.
Fadar came out of buying Meta ads and watching the same thing happen on every account: the creative starts dying days before the cost per result admits it. These field notes are what that looks like from inside an ad account, written from real numbers rather than a keyword list.
More about Mart →Fadar watches every Meta ad for fatigue and pings Slack, Telegram, or Discord in euros the day one starts to fade. The 90-day backtest is free.
Run the free backtest