Results dropped and you haven’t touched a thing in the account. Most of the time it’s Facebook ad fatigue, and it arrives in three stages: frequency climbs, CTR falls, cost per purchase goes up last. Wait for the third sign and you’ve already paid for a week of expensive delivery.
Facebook ad fatigue shows up in three signs, in this order
The signs appear in frequency first, then in click-through rate, and only at the end in cost per acquisition. Meta calls the phenomenon creative fatigue and describes it as an audience seeing the same creative too many times. The order below comes from the accounts we run; Meta doesn’t publish it.
The people in your audience don’t disappear overnight, but each of them sees the ad a third and a fourth time. Frequency moves first, because it’s a simple division: impressions divided by reach, the average number of times each person saw the ad. It’s an estimated metric, so read it as a seven-day trend rather than an exact daily number.
Then CTR falls, the share of people who click out of the people who saw the ad. Someone who scrolled past the same video twice doesn’t stop the third time. Clicks drop before sales do, because the purchase rate of the people who still land on your site holds steady for a while.
Cost per purchase climbs last, and it climbs from two directions at once. You get fewer clicks on the same budget, and CPM rises because you’re bidding into an audience where you’ve used up the easy buyers. CPM means cost per thousand impressions, spend divided by impressions, multiplied by 1,000.
| The sign | When it shows | What you do |
|---|---|---|
| Frequency climbs | first | check |
| CTR falling | second | prep creatives |
| Cost per purchase up | last | replace |
Meta recommends tracking frequency alongside your results, precisely so the same people don’t see your ads too often while the campaign runs.
At a frequency of 3.5, you check the ad set
We open up an ad set when its seven-day frequency passes 3.5. That’s our agency threshold, not a number published by Meta. The documentation only says frequency may average 1 to 2 per ad set or run much higher, depending on budget, audience size and scheduling.
The threshold moves with the temperature of the audience. On cold traffic, people who have never heard of you, 3.5 is already a lot: they’ve seen the same image three and a half times and nothing happened. On retargeting, a list of people who added to cart a few days ago, a frequency of 4 or 5 over seven days is normal and often profitable, because the list is small by definition.
The second factor is the ratio between budget and audience size. At the same budget, an audience of a few tens of thousands hits a high frequency much faster than one of a million. A broad interest audience in the US can keep a creative alive for two or three weeks; a 30,000-person lookalike won’t come close. Say you put $60 a day on an audience of a few tens of thousands: the 3.5 mark can show up within days.
For an official order of magnitude, in reach and frequency buying Meta sets a default cap of 2 impressions every 7 days. You wouldn’t run that buying type for sales, but it shows what the platform treats as comfortable exposure.
Six checks before you blame the creative
A good share of the “the ad died” cases we get sent are something else entirely. Before you go produce new creatives, run this list. It takes a minute and saves you a wasted week.
- Stock and price. Is the product that carried the campaign in stock, at the same price? One sold-out size cuts conversions without touching CTR.
- The site. The page loads, checkout works on a phone, the promo code from the ad is still valid.
- Measurement. Real orders in the store match what Ads Manager reports. If tracking broke, the sales still happen, you just stop seeing them.
- Account history. Meta recommends you review what edits were made right before delivery slowed down. Repeated pauses during the day confuse the system while it tries to pace the budget.
- Seasonality and competition. Black Friday, the start of the month, a competitor running hard on the same auctions. CPM goes up for everyone.
- Overlap between ad sets. Two ad sets bidding on the same people push frequency up without you changing anything.
The sign that points almost certainly at the creative: frequency and cost per result climb together, while new ads dropped into the same ad set start off well. When every ad in the set declines, including the ones you added yesterday, the problem sits with the audience or with delivery.
The first thing you change is what people see in the first seconds
The order of intervention runs from what gets seen first toward what’s hardest to change. People don’t read the ad copy if they never stopped scrolling, so rewriting copy on a tired visual shows up in the data as no change at all.
- The visual hook. The first two seconds of video or the static image. Another shot, another background, the product in someone’s hand instead of on white, a different opening line.
- The message angle. Same offer, a different problem pushed to the front. Fast shipping instead of price, a free gift instead of quality.
- The format. Video instead of static, carousel instead of a single image, vertical footage for Reels. A concept that died as an image often has a second life as video.
- The audience. A fresh list, a fresh lookalike, broader targeting.
- The offer. Last step, because it changes your margin and invalidates every comparison you made up to that point.
What tends to work in practice, for the hook and for the rest of the asset, we covered in the article on what a creative that sells on Facebook, Instagram and TikTok looks like.
Keep photos and videos in separate ad sets and don’t load more than 5-6 creatives into one ad set. Past that, delivery spreads thin and none of them gathers enough data for you to compare anything.
One change per ad set, then you wait for the data
Changing the audience and the creative on the same day leaves you with two variables and no answer. There’s a technical reason too: Meta treats as a significant edit pausing the ad set, changing the optimization event, the audience or the creative, and large budget or bid changes count as well. Each one restarts the learning phase.
Meta’s documentation also recommends waiting until the ad set has generated roughly 50 optimization events since the last significant edit, the point where costs get more stable. For a store with ten orders a week out of one ad set, that’s over a month. This is why we leave a campaign alone for the first 72 hours and don’t draw conclusions from fewer than three days of data.
When you add the new creative, put it in the existing ad set and let it compete with the old one for a few days. If the old ad set is already expensive, duplicate it with the new creative and turn the old one off only once the new one starts delivering.
The threshold where we kill an ad set with zero sales: 3x the target cost per acquisition on a new pixel with no history, and 2.5x on a pixel that has collected data. To shut off weak ads before they eat the budget, the rules in Ads Manager work overnight too; we walked through them step by step in the article on automated rules in Facebook Ads. Before you declare a creative dead, check how many conversions you need for the gap between two ads to mean anything.
The math for how many new creatives you need per month
You get the number from two figures: how many days a creative lasts for you, and how many ads actually carry delivery in each active ad set. You measure the lifespan once. Write down the day you launched a creative and the day the ad set’s frequency passed 3.5 with CTR sliding. The gap is your cycle.
Say you run $50 a day on Meta, split across two active ad sets, and the cycle you measured is 12 days. Inside each ad set, delivery concentrates on one or two ads in practice, the rest get crumbs. So you’re replacing four creatives every 12 days, around ten new concepts a month. A new version of the same video with a different opening shot counts as half a concept: it buys you one round, not two.
The only official number we found about cadence comes from the documentation for Advantage+ app campaigns, where Meta recommends uploading a high volume of creative assets and refreshing them a few times a month. For online stores there’s no published recommendation, so the arithmetic above is what you have.
Keep a buffer, too. When a creative dies on a Friday and you don’t shoot again for ten days, the ad set runs expensive that whole time or sits paused. Three concepts ready to upload get you out of that.
An old ad that performed can be switched back on
A creative that burned out can come back after a few weeks off. The cold audience refreshes itself, people forget, and the asset that convinced once convinces again. We put it back as a new ad inside an ad set that’s already running, instead of reactivating the old one.
Two conditions. First, the offer and the price shown in the asset are still valid. Second, something changed in the meantime: a different audience, a different opening shot, different primary text. Drop it back in identical on the same small retargeting list and frequency jumps over the threshold again within days.
For the audiences you refresh in parallel, the starting list is in the article on which Facebook and Instagram audiences to build on day one.
Frequently asked questions
What frequency is too high on Facebook Ads?
It depends on the audience. On cold traffic we check the ad set when seven-day frequency passes 3.5; on retargeting, 4-5 is normal because the lists are small. Meta publishes no cutoff, it only mentions that the average per ad set usually falls between 1 and 2 and that it depends on budget, audience size and scheduling. Read frequency together with CTR and cost per purchase, never on its own.
How many days does a creative last on Facebook Ads?
There’s no universal number, because the speed depends on the ratio between budget and audience size. A big budget on a small audience, which is what happens on most retargeting and lookalike lists, burns a creative in a week or faster. A small budget on a broad audience can hold for a few weeks. Measure your own cycle: the days between launch and the moment frequency passes 3.5 with CTR sliding.
CTR is dropping but sales are fine. Is that a bad sign?
It’s the warning sign, the one that shows up before the cost does. The purchase rate of the people who reach your site stays flat for a while, so sales look okay even though fewer people are coming in. Once clicks and sales volume both drop, cost per acquisition is already up. Use that window to get the new creatives ready.
Do I change the image or the ad copy?
Start with the visual. Copy only gets read by people who stopped scrolling, and on a tired asset they stop scrolling less. Swap the first frame of the video or the image, leave the copy alone, and compare. If CTR recovers, you know exactly what did it.
Do I turn the ad set off or let it recover on its own?
An ad set doesn’t recover from creative fatigue on its own, because the variable that broke stays in place. If it has brought zero sales and spent 3x the target cost per acquisition on a new pixel, or 2.5x on one with history, turn it off. If it still brings sales but expensive ones, add the new creative to the same ad set and give it a few days before you switch anything off. Remember that pausing the ad set counts as a significant edit and restarts the learning phase.
What Optimads does here
The hard part is catching in time that frequency climbed on one ad set out of ten, on a day you were busy with something else. In Optimads Autopilot, Finny reads the real numbers in your account, including frequency, CTR and cost per purchase by campaign, ad set and ad, looks at the creatives too, and tells you what to turn off and what to scale. You set rules that pause weak ads or shift budget on their own and get a notification on your phone, and you can generate new creatives from a product photo, with several copy variants. Every campaign you launch starts paused, so nothing is spent until you press go. Try Optimads for 7 days.
Sources
- Despre frecvența reclamelor (Meta Business Help Center)
- Recomandări pentru creative fatigue în Meta Ads Manager
- Despre campaniile reach and frequency, frecvența medie și CPM
- Despre faza de învățare și modificările semnificative
- Bune practici pentru activele creative în campaniile Advantage+
- Ce verifici când livrarea reclamelor încetinește





