Ad creatives and testing

How do I check creative fatigue in Meta ads?

Check each ad against its own first two weeks: rising frequency, falling link CTR, a flat landing page conversion rate and a climbing CPL point to fatigue.

Published

To check creative fatigue in Meta ads, follow one ad week by week against its own first two weeks of delivery: rising frequency, falling link click-through rate, a landing page conversion rate that stays flat and a CPL that climbs for two weeks is fatigue. Ads Manager can flag it, but only in some setups, so the reliable check is a small weekly table per ad and per landing page.

The four numbers to read

Fatigue happens before the click, which is why the pattern is recognisable. The people who have seen the ad many times stop clicking it, while the ones who do click still convert at their usual rate.

  • Frequency. Meta defines it as impressions divided by reach, so it is the average number of times one account saw the ad. It is an estimate built on sampled data, not a count.
  • Link CTR. This is the number that moves first, because it rests on hundreds or thousands of clicks rather than dozens of leads.
  • CPM and CPC. Both usually rise as the ad gets less engaging, since the same spend buys worse inventory.
  • Landing page conversion rate and CPL. The conversion rate should stay close to its baseline. If it holds while CPL climbs, the loss is upstream of the landing page.

Why the check belongs at ad level

An ad set or campaign CPL averages tired ads with fresh ones, and Meta’s delivery keeps shifting spend between them. A tired ad can be getting more expensive while a new one gets cheaper, and the average then hides both movements.

What the table showsWhat is happening
One ad’s CPL up, its CTR down, conversion rate flatthat creative is tiring
One ad’s CPL up on every landing page it usesthe problem is in the creative
One ad’s CPL up on a single landing page onlylook at that landing page, its form and its tracking
Every ad’s CPL up in the same week, CPM up toothe auction or the season, not the creative
Every ad’s CPL up on the same landing page in the same weekthat landing page changed or broke

The habit that follows: compare each ad with its own past, never with the account average and never with another ad’s best week. Meta itself warns about the breakdown effect, where an ad that looks expensive in one period is in fact producing the cheapest results of that period.

What Meta’s own signals cover

Meta’s fatigue article describes three indicators, and each has a limit worth knowing before you rely on it.

The Delivery column can show Creative limited or Creative fatigue. Meta compares the ad’s cost per result with ads you ran before: above that cost but under double gives Creative limited, double or more gives Creative fatigue. The feature covers ad sets with one creative, so a lead generation ad set with several concepts in it gets no status. The comparison also runs against past ads that may have used a different offer, a different landing page or a different season.

Before launch, Meta can warn that fatigue may set in during the first 7 days of a campaign. Use it as a signal about the asset, not a forecast about your account.

Ad relevance diagnostics give three rankings per ad: quality, engagement rate and conversion rate, each compared with ads competing for the same audience. They need at least 500 impressions and they do not feed the auction. A ranking that slides on the same ad for several weeks is a useful extra column.

Frequency has no published threshold. Meta says the metric may average 1 to 2 per ad set or run much higher depending on budget, audience size and schedule. Read it relative to the ad’s own history: when it is up by half on the first two weeks, put the ad on the review list and let CTR and CPL decide.

Rule out the other causes first

A rising CPL has four other explanations, and each needs a different fix.

  1. The cheap part of the audience is used up. Meta’s delivery takes the highest volume opportunities first and moves to more expensive ones when those run out, so costs rise with audience depth as well as with creative.
  2. A learning reset. After a significant edit, delivery is less stable and costs are usually higher for a while. Check the Last significant edit column before blaming the creative.
  3. The auction. If CPM rises on every ad in the account in the same week, that is competition or season. Meta also warns that when you optimize for conversions, CPM alone may not indicate performance, because the system can buy pricier impressions to find cheaper conversions.
  4. The landing page or the tracking. If CTR holds and leads drop, the click still works and the problem sits after it: the page, the form or the event setup.

A weekly routine, in six steps

  1. Build one table per ad ID and landing page pair, with spend, impressions, frequency, link clicks, the leads your own system received and the CPL computed from those leads.
  2. Set each ad’s baseline on its first two full weeks. If the ad set was still learning for part of that time, start the baseline after it left learning.
  3. Compare each ad with its own baseline, and sort the table by the size of the gap rather than by CPL.
  4. Read CTR and frequency first. A working starting rule is link CTR down 20% on the baseline for two weeks running, or frequency up by half. Set your own thresholds and keep them.
  5. Confirm on CPL with enough leads. Below about 20 leads a week, read CPL over two-week windows instead of single weeks, because a small count moves a long way on luck alone.
  6. Act once the other causes are excluded. Refresh the creative, add a concept, widen the audience or cut the ad. Cut when the ad is above target and has spent at least three times the target CPL since the rise began.

What to change once an ad is flagged

The cheapest change first, and the numbers say which one.

  • New hook. Same concept, different first line, opening frame or headline. Use it when CTR fell but the people who clicked still convert. Change the visual with the words, since Meta warns that creative it cannot tell apart may go to the same people again.
  • New format. The same message as a static, a carousel, a talking-head video or a screen recording, for when hook refreshes wear out faster each time.
  • New concept. When every angle fades within days, the angle is spent, and the next round belongs to concept testing. A Meta ads creative testing framework for lead gen covers how to run that round without crowning the luckiest ad.
  • New landing page. When CTR holds but the conversion rate drops, or when the refreshed creative promises something the landing page does not. Meta Andromeda chooses which ad a person sees; the landing page behind it stays your decision, as Meta Andromeda and creative diversification sets out.
  • Wider audience. When frequency climbs fast because the audience is small, which is one of Meta’s own recommendations.

Two habits keep the refresh from creating a new problem. Leave the original ad running while the new one ramps up, and add new ads in one batch, because each new ad restarts the ad set’s learning phase. Duplicating the tired ad is not a refresh: the same image or video counts as the same exposure, and the duplicate restarts learning without reaching new people.

DropCPL syncs spend per ad from the Meta Ads API and counts leads on your own landing pages, so the weekly table above exists per creative and per landing page without a manual export. More questions on this topic, including how many creatives to test, are in ad creatives and testing.

Questions on this topic

All questions on this topic →

FAQ

Does Ads Manager tell me when a creative is tired?+

Sometimes. A Creative limited or Creative fatigue status can appear in the Delivery column, and Meta warns before launch if fatigue looks likely in the first 7 days. The feature only covers ad sets with one creative, so an ad set testing several creatives gets no status at all.

What frequency is too high on Meta ads?+

Meta publishes no threshold. Its own definition says frequency may average 1 to 2 per ad set, or much more, depending on budget, audience size and schedule. Treat it as a flag relative to the ad's own history rather than a number to hit.

Can a falling CPL still mean fatigue?+

It can hide fatigue in an account where new ads are improving. Read the ad rows, not the ad set row: a tired ad can get more expensive while a fresh one carries the average down. Compare each ad with its own past.

Should I pause the old ad when I launch the replacement?+

Meta says keeping the original ad active may maximise results while the new one ramps up. Pausing it removes the delivery that still works. Add replacements in one batch, since a new ad is a significant edit that restarts the ad set's learning phase.

Ready to find out what
converts, for real?

Free to start.

Start free →