Facebook Ad Frequency Benchmarks by Niche
See how long winning Facebook creatives really last in six niches, measured on 2,660 finished ads from 805 brands, and the day to plan your refresh.
Every article about Facebook ad frequency gives you a number. Keep it under 3. Keep it under 2.5. Clothing tolerates 4, beauty tops out at 2.8. None of them tell you where the number came from, and there is a reason for that.
Meta does not publish frequency for anybody else's ads. Not in the Ad Library, not through any API, not to any tool. So when a page tells you that beauty brands fatigue at 3.1, nobody measured that. It is a number with a confident voice attached.
What Meta does publish is the day a creative started and the day it stopped. Brandsearch indexes both, across 120.8M Meta ads. The gap between those two dates is the only hard record of the moment a brand decided a creative was finished, which is the decision frequency is a proxy for in the first place. So we measured that instead. Across 2,660 winning creatives from 805 brands that have now completed their runs, the median lifespan ranges from 32 days down to 12.
The six niches
Ordered by how long the median winning creative survived before the brand replaced it. Ads is how many finished creatives were measured, Top 10% is what the longest tenth of them managed.
| # | Niche | Ads | Median | Refresh at | Top 10% |
|---|---|---|---|---|---|
| 1 | Health & Supplements | 474 | 32 days | day 26 | 163 days |
| 2 | Home & Garden | 470 | 28 days | day 23 | 136 days |
| 3 | Sports & Outdoors | 395 | 28 days | day 22 | 135 days |
| 4 | Beauty & Skincare | 479 | 25 days | day 20 | 162 days |
| 5 | Fashion | 430 | 22 days | day 18 | 185 days |
| 6 | Electronics & Tech | 412 | 12 days | day 9 | 90 days |
How to read the numbers
This is runtime, not frequency. Frequency is impressions divided by reach, and it exists in exactly one place you can see it: your own Ads Manager. Nothing here claims to know what a competitor's frequency was. It says how long their creative lasted, which is the outcome you actually care about.
Refresh at is 80% of the measured median. It is the old operator rule, which used to be pinned to a fatigue point nobody had measured. Now it is pinned to one that has been. If Beauty & Skincare replaces its winners at a median of 25 days, you want the next creative built and approved on day 20, not briefed on day 26.
Rank badges are worldwide, spend is not. Every rank in the screenshots below is computed across everything that brand runs in every market. The euro figures next to them are only the slice EU and UK disclosure rules force advertisers to publish, so a creative can rank first globally and still show a modest number. Never read the small euro figure as weak performance when the rank says otherwise.
These are winners, not all ads. The population is the ads a niche surfaces when you sort it by reach and by scaling signal, and every one of them had finished its run by 12 September 2026. That selection matters, and the gap it creates turned out to be the most useful thing in the whole pull.
1. Health & Supplements: 32 days
Supplements run the longest of the six, and the chart shows why. Cymbiotika sells pure-ingredient supplements for energy, focus and stamina, and its best-ranked creatives sit at the far right of the plot, past 200 and 300 days, while the low-ranked ones bunch up under 30. Across the niche as a whole the median of 32 days is the highest here, and 27% of Health & Supplements winners ran past 90 days, also the highest.
That fits the product. A supplement is a considered, repeat purchase with a long education cycle, so a creative that explains the mechanism keeps earning its impressions after a fashion ad would have burned out. Of the 593 finished Cymbiotika creatives in the pull, 122 carry a worldwide rank badge, and those ranked ones ran a median of 47 days against 9 days for the brand's creatives as a whole.
2. Home & Garden: 28 days
HexClad sells hybrid stainless and non-stick cookware, and its chart is the sparsest of the six for a reason: only 48 of its 581 finished creatives earned a rank badge at all. The ones that did lasted. HexClad's ranked creatives ran a median of 93 days, the longest of any brand here, against a median of 5 days for everything it ships.
That is the whole argument in one brand. HexClad is not running long creatives because cookware buyers tolerate repetition. It is killing almost everything inside a week and then riding the survivors for three months.
Pull any competitor's ads ranked against how long each one ran
What the catalog is doing underneath
The Products tab explains which creatives are worth keeping alive. HexClad's catalog runs to 203 products, and the top four slots are all one item in four colourways, the HexMill Pepper Grinder at $169, each holding its rank for about a week. Ranks five through eight are a $39 oven mitt set.
The rank age is the part worth reading. A bestseller that has held #1 for a week is a launch, and launches are what justify a new creative. A catalog where the top rank has not moved in months is one where creative refresh is the only lever left.
3. Sports & Outdoors: 28 days
On sells Swiss running shoes built around a cushioning system that activates on landing. Its chart is the clearest picture of what high-volume testing looks like: a dense wall of creatives under 30 days, and only 38 of 587 finished ads carrying a rank badge.
On's median creative ran a single day. Its ranked creatives ran a median of 37 days. That is not a contradiction, it is a testing programme: the brand is buying information at one-day resolution and concentrating budget behind the few that survive contact. Reading On's median runtime as its fatigue point would be a 37x error.
4. Beauty & Skincare: 25 days
Jones Road Beauty sells clean no-makeup makeup, free from sulfates, parabens and phthalates. Its plot has the tightest cluster of the six, most of it under 30 days, with one ranked creative holding out near 190. Beauty is also the most video-heavy niche in the pull at 64%, against 40% for Electronics.
The niche median of 25 days sits mid-table, but beauty has the second-widest tail: 22% of its winners ran past 90 days. High visual fatigue and long-running winners are not mutually exclusive, and the brands that manage both are rotating the hook rather than the product. Catching the moment a rotation stops working, from outside the account, is the signals that a competitor's winner is burning out.
The same idea, said 136 ways
The Hooks view is where rotation becomes legible. Jones Road runs 136 distinct opening lines across the 815 ads we scanned, with a further 1,185 catalog ads where Meta fills the text in per product. The top hook, "Miracle Balm is truly a miracle.", carries 51 ads on its own. Then "Meet your new morning essential." at 39, and "Not your typical foundation." at 36.
That is roughly six creatives per hook. When a brand rebuilds the same opening line into fifty ads, the creative is not what is being refreshed, the format is. It is the cheapest version of a refresh available and it is invisible unless you count the ads behind each line.
Find the hooks already working in your niche
5. Fashion: 22 days
Comfrt sells buttery soft hoodies, blankets and loungewear, and at 5.1k active ads out of 11.9k indexed it is the highest-volume advertiser in this list. Fashion is the most top-heavy niche in the pull: a 22-day median with 41% of winners dead inside two weeks, and yet the longest top decile of all six at 185 days.
Fashion is seasonal, promotional and crowded, so most creatives are disposable by design. The few that are not tend to be identity pieces rather than offers, and they sit at the right-hand edge of Comfrt's chart at 300 days while the promotional wall churns underneath.
Stop reading about winners. Find them yourself.
Search 6.5M+ brands, their ads, revenue, and products — all in one place.
Try Brandsearch free6. Electronics & Tech: 12 days
This is the row that contradicts the received wisdom, including the version this page used to carry. Electronics is routinely described as the niche where buyers tolerate repetition, on the reasoning that a demo video keeps working. Measured, it rotates faster than anything else here by a factor of two: a 12-day median, 53% of winners dead inside a fortnight, and the shortest top decile at 90 days.
Raycon sells wireless and open-ear audio, and its chart shows the pattern cleanly. A wall of creatives under 30 days, then a tight cluster of its best-ranked ads around 60 to 70 days. Its ranked creatives ran a median of 48 days against 6 days for the brand overall. The plausible read is that electronics competes on price and spec against a fast-moving field, so the offer inside the creative expires before the creative does.
Check your own rotation against the brands in your category
What the six niches have in common
Three things fall out of the pull that no single niche row shows.
The ranked and the unranked are different populations. Across all six named brands, creatives that earned a worldwide rank badge ran a median of 21 to 93 days. The same brands' creatives overall ran a median of 1 to 9 days. If you average those together you get a number that describes neither, which is roughly how the internet arrived at "keep it under 3".
Most creatives are never given the chance to fatigue. Between 29% and 53% of each niche's winners were finished inside 14 days, and those are the winners. Fatigue is not what kills the average Facebook ad. A decision on day four does. This is also why a competitor's active ad count tells you about their testing budget rather than their creative quality.
A short median does not mean a short ceiling. Fashion has the second-shortest median of the six and the longest top decile of any of them, 22 days against 185. Electronics has the shortest of both, 12 days against 90. So a niche's median tells you about its testing culture, and its 90th percentile tells you what is actually possible there when a creative lands.
The phase ladder, in days rather than frequency
Frequency means different things at different points in a creative's life, which is why a single threshold fails. Runtime bands are the same idea with a unit you can actually observe on a competitor.
Days 1 to 14. Most of the population dies here. Frequency readings are close to meaningless: a new creative can show a frequency of 5 against a tiny warm audience. Judge early CTR and cost per click instead.
Days 14 to 30. The scaling window, where your own frequency starts to matter. Rising frequency with flat or rising CTR is the system finding new people. Rising frequency with falling CTR is the same audience being hit again.
Day 30 and beyond. You are now past the median in five of the six niches. The creative has earned its budget. The refresh trigger here is not a frequency number, it is a CTR drop against its own peak.
Kill or refresh, and how to tell
The cleanest signal is not in your Ads Manager at all, because it moves first.
Rank movement leads cost metrics. A creative parked around rank 150 for weeks that suddenly sits near rank 400 is being shown less. Meta has seen a performance dip and reallocated, and that happens before your cost per acquisition has moved enough for you to trust it. Tracking a brand in Brandsearch accumulates that rank history daily, which is the one thing a snapshot of a competitor's ads can never give you.
Then the three-day rule, on your own account. If CTR is down 30% from peak for three consecutive days, or cost per conversion is up 50% for three days, act. Two of those signals together is a kill, not an optimisation. One is a refresh.
Find the replacement before you kill. Budget sitting idle costs more than a tired creative. The practical version is to know where the ads in your niche are actually pointing before you need to know.
HexClad runs 197 landing pages. The #1 page, a 13-piece hybrid cookware set, has 857 ads pointing at it and €1.7M of disclosed EU and UK spend behind it. The #2 page, a summer sale, carries 322 ads. That ranking is the brand telling you which offer it believes in, and it is a better brief for your replacement creative than any hook you could guess at.
Pull any competitor's landing pages, ranked by the ads pointing at them
The money math, and why positive ROAS can still lose
This one is arithmetic rather than measurement, and it is worth doing once.
A creative converting at 3.5 ROAS early in its life that decays to 1.8 ROAS late in its run, against a break-even of 2.0, is losing money while the dashboard still shows a positive number. The decay is gradual and the break-even line is invisible unless you have drawn it, so the ad reads as a winner for the entire period it is quietly eating margin.
Work out your break-even first and the kill decision stops being a judgement call. Run your own number through the break-even ROAS calculator, then set the alert at that number rather than at a frequency.
Your own frequency number, and what to do with it
Frequency is real and it is worth watching. It is just yours.
It is impressions divided by reach over whatever window you select, it lives in your Ads Manager, and no competitor research tool can show it to you for anybody else. The decision rule does not need a benchmark: if frequency is climbing and CTR is holding, the delivery system is still finding new people and you are scaling. If frequency is climbing and CTR is falling, you are paying to annoy the same audience, and the measured runtime for your niche tells you roughly how long you have before everyone else in that niche would have moved on.
For a deeper read on the kill decision itself, see when to kill a Facebook ad by phase signal and how to replace a dying ad using competitor research. For the cost side, CPM benchmarks by niche and what a niche's ad economics look like before you test cover the inputs this page assumes.
How we measured this
The population is Meta ads from Shopify DTC brands with at least $1M in estimated revenue, pulled per niche under two sort modes, by EU and UK reach and by scaling signal, then merged and deduplicated. Cumulative reach structurally favours long-finished campaigns, so using it alone would have biased every median upward.
A creative counts as measured when it has both a start and an end date and its end date is more than 10 days before the index date of 12 September 2026, meaning the run is genuinely over. The status field alone is not sufficient: an ad can carry status: active with an end date seven months old, and treating that as live is how a stopped campaign becomes a "current example". Runtime is end date minus start date, cross-checked against the API's own total_active_time field, which matched on every row.
Two biases point in opposite directions and both are left in. Ads still running on the index date are excluded because their runtime is not final, which pulls the medians down, and that was 6 to 27 ads per niche out of roughly 470. Selecting on reach and scaling rank favours creatives that accumulated reach, which pulls them up. The number is the median of winners, not of all ads, and the gap between those two is discussed above rather than hidden.
One view did not make it in. The Discovery board crashed the capture browser on three separate clean profiles, so there is no screenshot of the raw sorted list and no figure in this article is sourced from it. Everything above traces to a brand's own Ad Rank, Hooks, Landing pages or Products view, or to the API query below.
Query this yourself
Every number in the table came out of one endpoint. This is the actual call for one niche, not a cleaned-up version:
curl -s -H "X-API-Key: $BRANDSEARCH_API_KEY" \
"https://api.brandsearch.co/v1/meta-ads/search?platform=shopify\
&brand_type=dtc%20brand\
&niche=Beauty%20%26%20Skincare\
&min_revenue_min=1000000\
&status=active\
&sort_by=eu_total_reach&sort_order=desc\
&page_size=100&page=1\
&fields=id,brand_id,status,start_date,end_date,total_active_time,reach_rank,is_video,creative"
Request end_date and status together or you cannot tell a live campaign from a dead one. The median is then just end_date - start_date over the rows that have finished. The same data is available over MCP, so you can ask Claude, Cursor or ChatGPT for your own niche's rotation cadence directly.
FAQ
What is a good Facebook ad frequency?
For your own ads, under 3 over a 7-day window is a reasonable starting point for cold prospecting, but the number matters far less than its direction against CTR. Rising frequency with steady CTR is scaling, rising frequency with falling CTR is saturation.
Can I see a competitor's ad frequency?
No. Meta does not expose frequency for other advertisers through the Ad Library or any API, so no tool can show it to you, including this one. What you can see is how long each of their creatives ran, its worldwide rank, and how many ads they have live.
How often should I refresh Facebook creatives?
Use the refresh column above as the planning date rather than the deadline: between day 9 and day 26 depending on niche. That is 80% of the measured median run in that category, which leaves time to build and approve the replacement before the current creative is spent.
Does high frequency always mean ad fatigue?
No. Frequency rising while CTR and conversion rate hold is normal during scaling. Fatigue is frequency rising while performance falls, and the two are easy to confuse if you watch frequency on its own.
Why do electronics ads rotate fastest?
Measured across 412 finished winners, Electronics & Tech has a 12-day median run against 22 to 32 days for every other niche here. The likely reason is that the offer inside the creative, usually a price or a spec, expires faster than the creative itself in a category that competes hard on both.
