Creative fatigue, also called ad fatigue, is a drop in an ad's results after the audience has seen it too often1. Guides tend to quote frequency limits and refresh schedules as if they were laws. They are rules of thumb. This article sits under our creative testing framework, which makes decisions on margin, not CTR.
How do you know it's fatigue and not something else?
Fatigue is a pattern on one ad or concept over time. Look for all of these together:
- Cost per result rises week on week for the same ad.
- Response falls, for instance click-through rate (CTR), meaning clicks divided by impressions, on the same audience.
- Exposure climbs. The audience sees the ad more often.
- Nothing else explains it.
That last point does the work. Before you blame the creative, rule out the usual suspects:
- A price, offer or stock change.
- A change in budget, audience or placement.
- A tracking break.
- Seasonality or a promotion that ended.
- A slower landing page.
Our view: one bad day proves nothing. Look at a trend over time, on the same ad, with the same audience and offer.
Rising cost and falling response on the same ad over time is the signal. One bad day is not.
What do Meta's 'Creative fatigue' labels mean?
Ads Manager can flag ads with "Creative fatigue" and "Creative limited" labels. Jon Loomer, an independent Meta ads consultant, explains that both are based on cost per result rising compared with the ad's own past, with "Creative fatigue" the stronger rise; he calls his definitions simplified, because Meta does not spell out the trigger1. They are about cost, not about the look of the ad.
So the label is a prompt. It tells you cost moved. It does not tell you why. Run the checks in the section above before you act on it.
Are there real frequency thresholds?
Frequency is the average number of times a person saw your ad in a stated period, such as seven days. Guides online give "safe" frequency numbers. We have no source that ties any of them to Meta, so we don't quote one.
The sources we read disagree. Loomer says "there isn't a 'good' or 'bad' frequency" and calls refreshing every ad on a fixed cadence a myth1. A creative-analytics vendor's Andromeda playbook lists a frequency number as a warning sign without citing a source for it2. The same vendor's testing guide recommends monthly volumes of new concepts, a share of budget for tests and fixed refresh intervals per platform, presented as working rules rather than study results3. The vendor sells creative analytics and fatigue alerts. Use its figures as a conversation starter, not a rule.
Our view: read frequency alongside cost per result, not alone. A high frequency with stable cost is not a problem. A modest frequency with rising cost might be.
What does fatigue cost in margin and payback?
In our view, fatigue can show in cost per new customer before it shows in ROAS. Platform ROAS counts purchases from new and returning buyers alike, so returning buyers can hide the problem.
Translate it in three steps.
- Track new-customer CAC (nCAC) for the ad or ad set, not only CPA (cost per acquisition, ad spend divided by counted conversions). Our convention for nCAC: acquisition spend divided by first-time customers only.
- Take your contribution margin per order. Our convention: revenue minus variable costs before marketing, which means cost of goods, shipping, payment fees, returns and discounts.
- Recalculate the payback period. Our convention: the number of months of contribution margin needed to repay nCAC.
Example (illustrative): a first order earns €30 of contribution margin. nCAC rises from €40 to €55 as an ad tires. The gap on the first order grows from €10 to €25. If repeat orders cover it later, you wait longer for payback. If they do not, each new customer now costs you €25 net. Multiply by monthly new customers and you have the cost of not refreshing.
That number is the case for production budget. It is also the case for stopping an ad when the refresh would cost more than the fatigue.
How do you fix it: new angles, formats, audiences
Order the fixes by how much they change what the viewer sees.
- New angles. A different reason to buy, a different customer problem, a different objection. This is a new concept.
- New formats. The same angle as video, static, carousel or user-style footage.
- New audiences. The same ad in front of people who have not seen it.
- Small edits. New headline, colour or crop. Cheap, and usually the weakest.
New concepts usually matter more than small edits. Loomer's fix for a fatigued ad set is to add "completely new and unique ads" or start the ad set over1. The creative-analytics vendor reads Meta Andromeda, Meta's retrieval system for choosing ads, as grouping similar creatives together and rewarding varied concepts2. That is a vendor's interpretation, not a Meta statement. It is a useful reason to test genuinely different ads rather than ten versions of one.
Keep the winning idea. Change the way it is shown.
How do you plan production so you don't fall behind?
Treat creative as a running cost of acquisition, like media.
- Set a rhythm from your own data. Look at how long your ads usually hold before cost rises, then plan the next batch ahead of that point.
- Run concepts in parallel. Keep new ideas in test while the current ones still earn.
- Brief from learning. Write down why each ad won or lost, so the next brief starts from evidence.
- Budget production with media. A refresh that arrives late costs more than one planned.
Our performance creative service covers production and testing. Plan the cadence from your own trend lines, not a vendor's schedule.
