Your creative did not die on day 14. It died on impression two.

Meta’s own analytics team published the shape of the decay, and it is not a calendar. Repetition follows the curve (N+1)^-0.43, where N is the number of times that person has already seen that specific asset. Run their exponent out and the single biggest drop in the whole curve is the first repeat, roughly 26 percent between exposure one and exposure two. By four exposures, Meta puts the loss at about 45 percent of conversion likelihood, measured across roughly 26,000 experimental cases.

The same study reports the mean exposure count across all Meta ad impressions at 4.2, with more than 19 percent of impressions landing on someone who has already seen that creative five times or more.

That is not a creative shelf-life problem. It is an exposure accounting problem, and almost nobody is counting.

Why “refresh every two weeks” is the wrong instruction

A rule written in days only works at one budget and one audience size. Addict Mobile makes the arithmetic uncomfortable: a campaign running 100 EUR a day with a single creative against a small audience can reach a frequency of 4 or 5 within a few days. Move the same asset into a broad campaign and it might take six weeks to reach the same exposure count.

So when a blog tells you a creative lasts 10 days, ask which budget and which audience that number was calibrated on. Nobody publishing it says, because in most cases nobody measured it. The most repeated lifespan numbers in this category trace from one blog to another and never reach a dataset.

Half your creatives never live long enough to fatigue

Motion’s Creative Benchmarks 2026, published 17 April 2026 on 578,750 unique creatives, 6,015 advertiser accounts and 1.29 billion dollars of realised spend, found that 50 to 53 percent of creatives are switched off before they reach 28 days. Only 4 to 8 percent become winners, and those winners absorb about 55 percent of total spend.

Read that against the fatigue conversation and something obvious falls out. For half of everything you launch, fatigue is not the diagnosis. The asset never accumulated enough repeat exposures for decay to be the story. It was a concept, offer or targeting failure on arrival.

Which means most teams are applying refresh logic to an account where the real problem is hit rate, and answering it by producing more variations of concepts that were never viable.

What the curve looks like when the asset does survive

For creatives that do live, decay is slow and then sudden. NP Digital, running 50 companies on Google and Meta, measured cumulative CTR decay of 3.4 percent by week 2, 7.3 percent by week 4, 10.8 percent by week 5, 15.7 percent by week 6, and 27.1 percent by week 8.

Nearly flat for a month, then more than triples in the four weeks after. Now compare that to production reality. Smartly’s 2026 Digital Advertising Trends Report, surveying 450 marketing leaders, found 41 percent of marketers still need three to four weeks to launch a campaign and only 3.6 percent can go live in under a week.

Your production lead time is almost exactly as long as the flat part of the curve. Teams that wait to see the drop are structurally a month late.

Five things to do with this

1. Split the creative report into two cohorts at day 28 before diagnosing anything. Anything killed earlier is a concept problem. Only survivors are eligible for a fatigue conversation.
2. Move the refresh trigger from a date to average frequency, and set it at 2 on cold prospecting, not at 4. By the time you hit Meta’s own platform average of 4.2, roughly half the conversion likelihood is already spent.
3. Hold 6 to 8 active creatives per campaign. Addict Mobile reports that as the 2026 standard, where 3 or 4 was enough two years ago. Rotation keeps per asset exposure low without demanding more net new production.
4. Brief the replacement at week 4, while the incumbent still looks healthy, and swap at week 5 or 6. Given a three to four week production cycle, briefing at the first visible dip means shipping into week 8.
5. Retire assets, not concepts. A concept that produced a winner deserves new executions before it is abandoned. Motion’s own framing is that low hit rates are a statistical feature, not a quality signal.

The limitation

Almost every fatigue benchmark in circulation is walled garden social. Motion’s 578,750 creatives are Facebook and Instagram. Meta’s decay function is Meta’s auction and Meta’s own frequency management. NP Digital’s curve is Google and Meta.

There is no published decay curve for open programmatic, oRTB or in app mediated supply. Advertisers buying outside the walled gardens are applying Meta derived refresh rules to inventory with completely different frequency dynamics and different audience overlap, without the first impression diagnostics those platforms provide. Anyone who tells you otherwise is extrapolating.

There is also a credible counter argument worth knowing. Brand side research on wear-out, including System1 work documented by The Research Agency, shows strong creative holding top positions for well over a year with no sign of decay. That is television and brand, and it does not transfer cleanly to direct response mobile UA, but it should stop anyone from writing “all creative decays” as a law.

Where the structure of the deal changes the incentive

Most media relationships make creative decay the advertiser’s problem alone. Impressions are billed whether the creative is on exposure one or exposure nine, so the cost of fatigue sits entirely on the buyer’s side of the table.

On a CPA model the incentive flips. For example, at SpinX we are paid per verified action, confirmed by the MMP, which means a fatigued creative hits our margin before it hits the client’s ROAS. Frequency management stops being a reporting courtesy and becomes an operational necessity for us. Owned and operated supply is what makes it enforceable, because we can see exposure at the placement level rather than inferring it from a blended frequency number three platforms deep.

That is the practical version of everything above. Fatigue is an exposure problem, exposure is a supply visibility problem, and supply visibility is a structural choice, not a setting.

If creative decay is showing up in your ROAS before it shows up in your dashboard, talk to us about running your next flight on pay-per-result inventory at spinx.io.