By the time ROAS drops, you have already paid for the decline. Here are the earlier signals.
An ad that was working stopped working, and you found out from a ROAS report a week after it started costing you money.
5 min read
ROAS is a lagging indicator of creative fatigue. It is the last metric to move, because it sits at the end of a chain — impressions, clicks, landing page visits, add to carts, purchases — and each link absorbs some of the decline before passing it on.
By the time your ROAS report shows the drop, the ad has typically been decaying for one to two weeks. The money spent during that window is not recoverable. The point of watching earlier signals is to shorten that window, not to eliminate it.
Here are the indicators in the order they move, with the thresholds that make each one actionable.
The order in which a fatiguing ad falls apart
| Stage | What moves | Typical timing |
|---|---|---|
| 1 | Frequency climbs as the reachable audience is exhausted | Days 5–10 |
| 2 | Click-through rate falls against its own baseline | Days 7–14 |
| 3 | Cost per click rises to compensate for the weaker response | Days 10–16 |
| 4 | Cost per purchase rises | Days 14–20 |
| 5 | ROAS finally drops enough to notice | Days 18–25 |
Anyone watching only stage five is making decisions with a two to three week lag on a channel that changes weekly.
Signal 1: frequency
Frequency is the average number of times a person in the audience saw the ad over the period. It is the earliest signal because it is a cause rather than a symptom.
Meta's own research indicates performance typically starts to decline past a frequency of roughly 3 to 4 for direct-response campaigns. For ecommerce cold prospecting specifically, practitioner data puts the working band tighter, at roughly 1.8 to 2.5 over a rolling seven days.
1.8–2.5
Working 7-day frequency band for ecommerce cold prospecting
Triple Whale creative fatigue analysis
3–4
Frequency at which Meta's own research indicates direct-response performance typically declines
Practitioner summaries of Meta research
- Below 2.0 on cold prospecting: healthy, keep going
- 2.0 to 2.5: brief the replacement creative now, while the current ad is still working
- 2.5 to 3.5: ship the replacement
- Above 3.5 on cold traffic: you are paying for repetition
Retargeting tolerates far higher frequency, because the audience is small by design and the shopper already knows you. Judge retargeting on cost per purchase rather than on frequency.
Signal 2: click-through rate against its own baseline
The mistake here is comparing your click-through rate to an industry benchmark. Benchmarks vary enormously by category, audience and placement, and yours is not theirs.
Compare the ad to itself. Take the click-through rate from the first five to seven days of meaningful delivery as the baseline. A fall of 20% to 30% below that baseline is a widely used fatigue threshold, and it holds up in practice because it is relative rather than absolute.
One important distinction. A high frequency with a stable click-through rate is not fatigue — it is an ad that keeps working on repeat exposure, which some very good creative does. Watch the pair together, not either alone.
Fatigue versus a bad ad, and why the difference is expensive
These need entirely different responses and are routinely confused.
| Bad ad | Fatigued ad | |
|---|---|---|
| Started | Underperformed from day one | Performed well, then declined |
| Frequency | Often still low | Elevated and rising |
| Click-through trend | Flat and low | Was high, falling |
| Right response | Cut early on small spend; the concept is wrong | Refresh the execution; the concept still works |
| Cost of the wrong response | Spending more testing a losing idea | Abandoning a winning idea because one execution wore out |
The second error is the more damaging one. Brands regularly kill a concept that was genuinely resonating because they read a fatigued execution as a failed idea, and then spend months hunting for something that works when they already had it.
What to actually refresh, in order of leverage
A new ad does not have to mean a new production. Ranked by effort against effect:
- Hook — the first two seconds of a video, or the top third of a static. Highest leverage by a distance, and cheapest to change.
- Opening frame or thumbnail. Changes who stops scrolling, which changes everything downstream.
- Primary text and headline. Cheap, fast, and frequently underrated.
- Format — the same content cut as a static, a carousel, a short vertical video.
- Angle — same product, different reason to buy. Convenience instead of price, results instead of ingredients.
- Whole new concept. Highest effort, and the only one that requires a real production cycle.
A workable refresh cadence
- Cold prospecting: plan a refresh every two to three weeks, briefed at frequency 2.0 and shipped by 2.5 to 3.5
- Retargeting: every one to two weeks, because the audience is small and sees everything
- Evergreen best performers: leave them alone while the click-through baseline holds; some run for months
- Seasonal and promotional: built and retired with the promotion, and never left running past it
Set a weekly twenty-minute review: frequency by ad set, click-through trend by ad, and the state of the concept queue. That is the entire discipline. Everything more elaborate than this is optional; everything less than this is guessing.
Common questions
What frequency means my creative is fatigued?
For ecommerce cold prospecting, costs typically start rising past a 7-day frequency of about 2.5, and above 3.5 you are largely paying for repetition. Retargeting sustains much higher frequency, so judge it on cost per purchase instead.
How much of a CTR drop indicates fatigue?
A 20% to 30% fall below the ad's own first-week baseline, with frequency rising, is the widely used threshold. Compare the ad to itself rather than to an industry benchmark.
How often should ecommerce brands refresh ad creative?
Cold prospecting typically every two to three weeks; retargeting every one to two weeks. Genuine evergreen performers can run far longer, and should be left alone while their click-through baseline holds.
Is a fatigued ad the same as a bad ad?
No, and the difference matters commercially. A bad ad never worked and the concept is wrong. A fatigued ad worked and wore out, which means the concept is still good and only the execution needs replacing.
Does changing the headline count as new creative?
It counts as a new execution, and it often moves performance more than the effort suggests. The hook and the opening frame are the highest-leverage changes; a full new concept is the most expensive and not always the most effective.
How Glimmio handles this
Glimmio watches frequency and click-through decay at ad level against each ad's own baseline, so fatigue surfaces as a recommendation while there is still time to act rather than as a ROAS report after the fact.
The evidence comes with the finding — the ad, its baseline, its current reading, and the threshold it crossed — so the decision to refresh or to cut is made on numbers rather than on instinct.
- Manual approval by default — nothing runs unattended
- New campaigns and ads are always created paused
- 7-day recovery window on eligible changes
- 48 permissions across 13 roles, scoped per client account
Go deeper on this
The product pages and setup guides that cover what this article describes.
Searches this answers
- meta ads creative fatigue signs
- ad frequency threshold ecommerce
- when to refresh ad creative
- ctr decline facebook ads
- creative testing cadence d2c
Read next
A Creative Testing Framework That Survives Small Budgets
You know you should be testing creative but every framework you read assumes a five-figure monthly test budget.
ReadHow to Find Wasted Meta Ad Spend in One Afternoon
You are spending five or six figures a month on Meta and you have a strong suspicion that a meaningful slice of it is doing nothing.
ReadAI Ad Creative Without Losing Your Brand
AI creative tools produce output fast, and most of it sounds like it could belong to any brand in your category.
Read