Week one, the new ad prints a 4x return and the team moves budget toward it. Week five, the same ad is sliding toward break-even with nobody having touched the image, the copy, or the targeting. The reflex is to call it ad creative fatigue and brief a new batch of creative. Sometimes that is exactly right. Often it is not, and the chart that looks like fatigue is actually a measurement change wearing a fatigue costume.
Both problems produce a declining ROAS line. Only one of them is fixed by a new video.
What real ad creative fatigue looks like in the data
Ad creative fatigue has a specific, three-metric signature, and it shows up gradually rather than overnight.
Frequency climbs. Meta defines frequency as impressions divided by reach, the average number of times one person has seen the ad. As a campaign keeps spending into the same audience pool, frequency rises by definition, and Meta's own guidance is to watch it alongside results because performance dropping as frequency rises is the standard fatigue pattern.
Click-through rate erodes. The same creative stops earning clicks from people who have now seen it five or ten times. This is usually the first metric to move, before cost per result visibly worsens.
Cost per mille creeps up. Platforms charge more to deliver an ad that engagement signals say an audience no longer wants, so CPMs rise even though targeting and budget have not changed.
Tolerance for frequency varies by funnel stage, and the numbers below are illustrative media-buying heuristics rather than platform-published rules: as a rough guide, prospecting audiences tend to show the first signs of fatigue starting around a frequency of 2.5 to 4, while retargeting pools, smaller and already familiar with the brand, often hold up fine to 6 or 8. Many media buyers use a working rule of thumb of keeping mid-funnel frequency under roughly 4 across a rolling week, specifically to avoid the blindness that triggers fatigue in the first place. The number matters less than the trend: frequency rising while click-through rate falls on the exact same asset, over consecutive weeks, with nothing else in the account touched.
That last clause, nothing else in the account touched, is where most fatigue diagnoses go wrong.
Three things that fake a fatigue curve without the creative aging a day
An attribution window or model shifted underneath the report
Attribution settings change what gets counted as a conversion, not what actually happened on the ad. Google Ads' own documentation on data-driven attribution notes that switching models, or Google's periodic default changes, can distribute credit differently across the customer journey, and conversion reporting under non-last-click models can run 15 to 48 hours behind what a last-click view would show. A founder glancing at "this week versus last week" during that lag sees a dip that has nothing to do with the creative.
Meta has made comparable changes to how it classifies attribution windows in its own reporting over time. A shift like that can make a perfectly healthy ad look like it suddenly stopped converting, because the counting rules moved, not the audience's response to the ad.
A broken audience inflated frequency on a pool that quietly shrank
An exclusion audience can stop doing its job silently. If the event it was built on gets renamed anywhere in the account, past buyers never leave the retargeting pool, frequency on that pool climbs fast, and the resulting decline reads exactly like textbook creative fatigue on every chart pulled. Anlyto's breakdown of diagnosing Meta ads before blaming the creative walks through a delivered case where this exact fault, not the ad, was the entire problem. Rebuilding the audience fixed the decline; a new creative batch would have changed nothing.
Automatic rotation moved spend, not performance, onto fewer assets
Google's asset performance labels rank each asset in a group as Low, Good, or Best relative to the others in that same group, based on the trailing 14 days, and the system automatically favors assets labeled Best. If one asset's relative ranking slips for reasons unrelated to audience fatigue, a landing page update, a seasonal shift in what a competing asset is testing, the system can concentrate impressions elsewhere. The original asset's numbers decline not because the audience grew tired of it but because it is simply being shown less. Anlyto's PMax reporting breakdown covers how much of this reallocation stays invisible even in the improved 2026 reporting view.
The diagnostic order: check this before you touch the creative
Work through these in order, because each step rules out a cheaper, faster fix before you commit to a creative refresh.
1. Pull the frequency and click-through trend over three full weeks, not three days. A short window can catch normal day-to-day noise or reporting lag and mistake it for a trend.
2. Check what changed in the account on the date the metric moved. A renamed event, an edited exclusion audience, a new attribution setting, or a budget shift are all dated changes. If one lines up with the drop, start there.
3. Check whether spend concentrated onto fewer assets in the same window. If a sibling asset's performance label flipped to Best around the same date, reallocation is a likely cause.
4. Only once the first three are ruled out, treat it as creative fatigue. At that point frequency has climbed gradually, click-through has eroded on the same asset over weeks, and nothing else in the account moved on the same dates.
This is the same discipline behind Anlyto's broader diagnosis of Meta ads not converting: a declining number is a symptom, and treating the wrong symptom wastes a creative production cycle on a problem a five-minute change-log check would have caught.
Once it's confirmed as real fatigue
A refresh cadence beats a reactive scramble. As an illustrative starting point, not a fixed rule, many media buyers rotate top-of-funnel prospecting creative every three to four weeks and bottom-of-funnel or retargeting creative every six to seven weeks, since smaller retargeting pools tolerate repetition longer before frequency becomes a problem. Brands that keep a pipeline of new variants entering rotation every week rarely treat fatigue as an emergency, because a replacement is already queued before a winner decays.
Keeping that pipeline running, and reading the data correctly enough to know when a decline is actually fatigue versus a tracking problem, is the ongoing work behind Anlyto's growth partner retainer, which pairs tracking health monitoring with creative strategy so the two diagnoses never get confused. Where the issue turns out to be execution rather than measurement, that same partnership covers ads management directly.
The decision, in one pass
If frequency is climbing, click-through is eroding gradually on the same asset over weeks, and nothing else in the account changed on those dates, it is creative fatigue. Refresh it. If the drop was sudden, or it lines up with a renamed event, an edited audience, or an attribution setting, it is a measurement problem wearing a fatigue costume, and a new ad will not fix it.
If you are not sure which one you are looking at, a free analytics audit checks the tracking and attribution layer first, so the next decision is made against a number that is actually real.
