All posts

記事Meta Ads

Creative fatigue on Meta: how to tell it apart from a bad audience

Creative fatigue is the point where an audience has seen an ad so often that fewer people click it and Meta charges more to keep showing it.

A Meta ad that gets more expensive every week is either worn out or aimed at the wrong people, and the two need opposite fixes. Here is how to tell which one you have from four columns in Ads Manager, in about fifteen minutes per campaign.

Ruslan8 min read

What is creative fatigue on Meta?

Creative fatigue is what happens when the same people see an ad too many times. Fewer of them click, Meta charges more per thousand impressions to keep showing it, and the cost per result climbs. It builds over days, not hours, and it shows up in the numbers before it shows up in your budget report.

The trouble is that each of those symptoms has another cause as well. A falling click-through rate can be a season. A rising frequency can be a small audience doing exactly what you asked. A rising CPM can be the whole auction getting more expensive. Fatigue is the combination, and the check below is about reading the combination.

How is fatigue different from a bad audience?

A fatigued ad worked and stopped working. A badly targeted ad never worked. That is the whole test: look at the same ad at the start of its run and now. If the early days were good and the recent days are not, the creative is wearing out. If the early days were already poor, a new creative will not fix it.

DiagnosisWhat the numbers showFix
Creative fatigueFrequency rising, CTR falling, CPM and cost per result rising, all on the same adNew creative in the same ad set, then pause the old one
Small audienceFrequency high from the start, CTR and cost per result stableNothing, or widen the audience if you need more reach
Wrong audienceCTR low and cost per result high from day one, frequency normalChange the targeting; a new creative will not help
Auction shiftCPM up across every ad in the account at once, CTR flatWait it out, or adjust budgets; the ads are fine

The last row is the one people miss. Late November and the end of a quarter push CPMs up for every advertiser. If all of your ads got more expensive in the same week, that is the auction, not your creative.

How do I check for creative fatigue in Ads Manager?

Open the ads, add four columns, break the report down by day and compare the first days of the window with the last. In Meta Ads Manager:

  1. Open the Ads tab, not Campaigns or Ad sets. Fatigue belongs to one creative, and an ad set average hides it.
  2. Set the date range to the last 14 days.
  3. Choose Columns → Customize columns and add Frequency, CTR (link click-through rate), CPM and Cost per result, beside Impressions and Amount spent.
  4. Choose Breakdown → By time → Day. Each ad now has one row per day.
  5. For each ad, average the first three days and the last three days of each column, and compare them.

Three days at each end smooths out a single odd day without hiding the trend. Skip any ad with fewer than about 1,000 impressions in either three-day slice: at that volume a handful of clicks moves the CTR by a third, and you would be reading noise.

What numbers mean an ad is fatigued?

No single number does. Call an ad fatigued when at least two of these move together and one of them moves a lot. One signal on its own means watch it, not pause it.

SignalFirst 3 days → last 3 daysStrong
Click-through rateDown 25% or moreDown 50%
FrequencyDaily average of 2.5 or more3.5
CPMUp 20% or moreUp 40%
Cost per resultUp 30% or moreUp 60%

These are working thresholds, not Meta’s. Meta publishes no frequency at which an ad is worn out, and the right number depends on the offer: a retargeting ad for a cart someone abandoned can run at a high frequency and still pay. Use the table as a starting point and move it once you have a few months of your own history.

Frequency in Ads Manager is an average over the date range you picked. Over 14 days a frequency of 5 means about one view every three days per person, which is rarely a problem. Compare frequency per day, from the daily breakdown, not the figure for the whole window.

A second opinion from Meta

Ads Manager also shows three ad relevance diagnostics: quality ranking, engagement rate ranking and conversion rate ranking. Each one ranks your ad against ads competing for the same audience. An engagement rate ranking that slides from Average to Below Average while frequency climbs is the same story told by Meta. The diagnostics need at least 500 impressions and cover only the last 35 days, so they confirm a verdict rather than make one.

A worked example

A shop spends $150 a day across one prospecting campaign with three ad sets and twelve ads. Cost per purchase has gone from $31 to $44 in two weeks. The daily breakdown, first three days against last three:

AdCTRFrequencyCPMCost per purchase
Spring video1.42% → 0.81%3.8$9.10 → $12.40$28 → $52
Bestsellers carousel1.10% → 0.74%2.9$8.30 → $11.60$33 → $49
Testimonial2.05% → 1.49%1.9$7.80 → $8.10$30 → $34
Lookalike static0.48% → 0.51%1.4$8.60 → $8.90$71 → $69

Reading it:

  • The spring video is fatigued. All four signals fire, two of them strongly. It was the best ad in the account two weeks ago.
  • The carousel is fatigued too. CTR is down a third and CPM up 40%, with frequency close to 3.
  • The testimonial has one signal, a CTR down 27%, at a normal frequency. Watch it; do not touch it yet.
  • The lookalike static is the trap. Its cost per purchase is the worst in the account, but nothing moved: it was this expensive on day one. That is an audience problem. A new creative in that ad set would cost money and change nothing.

Two ads to refresh, one to watch, and one ad set whose targeting needs another look. A rule that paused everything above $40 per purchase would have killed the testimonial and kept the real problem running.

What should I do with a fatigued ad?

Put the replacement in first, then pause the old ad. Adding a new creative to the same ad set keeps the audience and the learning the ad set already has. Pausing the old ad first leaves the ad set with less to deliver while the new one starts.

  • A new angle, not a new colour. The people who stopped clicking have seen the message. A different hook, a different first three seconds or a different format helps more than the same video recut.
  • Widen the audience when the ad itself still works on new people and the ad set has simply run out of them. Rising frequency on a narrow audience is the sign.
  • Pause, never delete. A paused ad keeps its history, and an ad that has rested for a few months sometimes works again.
  • A frequency cap is not a fix for most campaigns. Meta honours a frequency cap only on reach buying, so on a sales or leads campaign it is not available.

How to run this with an AI assistant

The check above is arithmetic on a daily report, which makes it a good job to hand over. With Meta Ads connected to Adako, it is one question:

Are any of my Meta ads burning out?

Adako pulls the day-by-day numbers for every ad over the last 14 days, compares the first three days with the last three, and uses the four thresholds in the table above. Each ad comes back as fatigued, watch, healthy or too little data, with the figures behind the verdict. The check is a read, and reads cost nothing on any plan.

If you want to act on it, the pause or the new creative arrives as a proposal. Nothing changes in Ads Manager until you approve it, and a new ad is created paused so the replacement is ready before the old one stops. The creative fatigue use case lists the exact steps and limits.

Where it stops: it compares two ends of a window rather than fitting a trend, so a spike on either end moves the verdict. And it cannot tell you whether the new angle will work. That part is still yours.

How often should you check?

Weekly for any ad set spending more than $50 a day, and every two weeks below that. Check sooner after you narrow an audience or raise a budget sharply: both push frequency up faster. An ad set with a small audience and a large budget can wear out a creative in under a week.

The short version

  • Work at the ad level, over 14 days, broken down by day.
  • Compare the first three days with the last three; skip ads under 1,000 impressions.
  • Fatigue is two or more of: CTR down, frequency up, CPM up, cost per result up. One alone is a watch.
  • An ad that was poor from the start has an audience problem, not a creative one.
  • If every ad got more expensive at once, it is the auction.
  • Add the replacement first, then pause the worn-out ad. Never delete it.

Ruslan builds Adako, the MCP server these guides are written alongside, and writes up the ad-ops procedures behind it.

Keep going