The frequency benchmark worth using is weekly frequency per creative, not campaign frequency: decline begins around 2.5 on Meta prospecting and around 3.0 on TikTok. Campaign frequency is an average across every ad in the set, so a campaign sitting at a comfortable 2.0 routinely contains one creative that has been served six times to your best segment. Work out the per-creative number and the benchmark becomes actionable instead of reassuring.
How do you calculate ad frequency
Frequency is impressions divided by reach, over a stated window. The window is the part people drop, and dropping it is what makes the number useless.
Lifetime frequency only ever goes up, so it tells you nothing about whether pressure is building this week. Use a rolling seven days. On Meta you get it directly by adding Frequency to an ad-level breakdown with the date range set to the last seven days; per-creative reach is reported at ad level, which is the whole reason the ad-level view is worth opening.
Two derived numbers are worth keeping next to it. Impressions per day is your daily budget divided by CPM, times 1,000. Accumulation rate is budget over audience: it is the speed at which you are spending down a finite pool of attention, and it is the only reason two accounts running identical creative see creative fatigue arrive weeks apart.
Ad frequency benchmarks by band
| Weekly frequency per creative | What it means | What to do |
|---|---|---|
| Below 1.0 | Delivery is spread thin, learning is slow and results are noisy | Consolidate: fewer ads in the set, or more budget, before you judge anything |
| 1.0 to 2.0 | The healthy operating band for prospecting | Nothing. Keep producing for the next rotation |
| 2.0 to 2.5 | Approaching the point where decline starts on Meta | Start production now. This is a supply trigger, not a pause trigger |
| 2.5 to 3.0 | Decline has begun on Meta feed and Reels; TikTok is at its ceiling | Promote a fresher variation and take budget off the top-frequency ad |
| 3.0 to 4.0 | Fatigue is arriving, and CPM is usually already moving | Replace, do not tune. Nothing in the media buy recovers this |
| Above 4.0 | Past the fourth exposure, where CTR drops 45 percent | Pause the creative. Retargeting a warm pool is the only defensible case |
Those bands are for prospecting. Retargeting a warm pool legitimately runs higher, because you are paying for reminder rather than discovery, and 4 to 6 a week on a two-week window is normal there without anything being wrong.
What the platform ceilings actually are
Our 2026 fatigue benchmark puts the ceiling at 2.5 weekly on Meta Reels and Meta feed, around 3.0 on TikTok, and 3 to 7 per week on YouTube in-stream. The same report has Meta internal research finding CTR drops 45 percent after a fourth exposure to the same creative, which is the figure that makes the "above 4.0" row in the table above a hard stop rather than a preference.
The ceilings differ because the placements differ. TikTok tolerates a higher number partly because the feed is faster and partly because hook rates there run higher, 33 percent median in our benchmark against 28 percent on Meta, so a larger share of impressions are actually being watched rather than passed. YouTube in-stream sits higher again because a skipped pre-roll barely registers as an exposure.
Do not treat any of these as targets. They are the point at which the curve bends, and you want to be replacing creative before you reach them, not on the day you do.
Why campaign frequency hides an over-served ad
Here is the arithmetic that convinces most teams to change what they look at.
An ad set runs for a week: 500,000 impressions against 200,000 reach. Campaign frequency is 2.5, which looks like the top of the healthy band. Five ads are live, so the intuitive read is that each one is sitting near 0.5 and nobody is over-served.
Now look at delivery. Meta concentrates spend on whatever is winning, and 60 to 70 percent of impressions landing on one ad is completely ordinary. Say the winner took 325,000 of those impressions. Its own reach was 130,000, because it was shown repeatedly to the segment that responded. Its frequency is 325,000 divided by 130,000, which is 2.5, five times the number you assumed and exactly at the point where decline starts. The other four ads sit near 0.4 each, which drags the campaign average down and makes the set look healthy.
This is why the campaign number is comforting and wrong. The average is computed across ads that are not being served, so adding weak creative to a set improves your frequency report without changing a single person's experience of your advertising. The ad your audience is actually tired of is the one the average hides.
Working out when you will hit the ceiling
Frequency is forecastable, which makes it the only fatigue signal you can schedule around. A worked example, all figures illustrative:
- Daily budget 700, CPM 14. Impressions per day = 700 divided by 14, times 1,000 = 50,000.
- Weekly impressions = 350,000.
- Audience 250,000, of which the platform typically reaches 60 percent in a week = 150,000 reach.
- Weekly frequency = 350,000 divided by 150,000 = 2.33.
- Impressions needed to hit 2.5 = 2.5 times 150,000 = 375,000, which at 50,000 a day is 7.5 days.
So this ad set crosses the Meta ceiling in about a week and a half, and if 65 percent of delivery goes to one creative, that creative crosses it sooner. Halve the budget and you get roughly three weeks. Double it and you get four days. Our ad fatigue calculator runs the same calculation with your own numbers, including the concentration of spend on the winner.
The practical use of this is planning production, not planning pauses. If you know the winner crosses 2.5 on day eight, the replacement has to be in review by day six, which means it has to be briefed by day four.
What to do with the number once you have it
Set the alert at 2.0 rather than 2.5, because production has a lead time and delivery does not. Read frequency at ad level weekly, not at campaign level, and never at lifetime. Keep 8 to 20 live variations per active campaign so there is somewhere for budget to move when one creative crosses the line. And when it does cross, replace the creative rather than capping it: a cap paces delivery and pushes budget to worse impressions, which flattens the curve without resetting it.
The reason most teams do not act at 2.0 is that they have nothing to promote. That is the part we work on. Genyad turns footage you already own into new video variations, each one a fresh script, shot selection, voiceover and caption set rather than a re-cut of the same timeline, at one credit per variation. It does not read your frequency for you: no dashboard, no predicted performance scores, and no direct publishing to Meta or TikTok. The number comes from your platform reporting, and the replacement comes from your library.
Frequently asked questions
What is a good ad frequency on Facebook or Instagram?
Between 1.0 and 2.0 weekly per creative in prospecting. Our benchmark report puts the start of decline at a weekly frequency of 2.5 on Meta prospecting, so 2.0 is a sensible alert threshold and anything above 3 means fatigue has arrived. Retargeting a warm audience runs higher, and 4 to 6 a week there is usually fine.
Is frequency measured per campaign or per ad?
The platforms report both, but only the per-ad figure is useful. Delivery concentrates on the winning creative, so a campaign average is dragged down by ads that are barely being served while the one everybody has seen sits well above the ceiling. Read frequency at ad level over a rolling seven days.
Should I use a frequency cap to control this?
Rarely in prospecting. A cap paces delivery rather than fixing the underlying decay, and it pushes budget to the next-cheapest impression, which typically raises CPM. Caps earn their place on small warm retargeting pools where the alternative is frequency 6 by the middle of the week.
Why does the same creative last longer in some accounts?
Because accumulation rate is budget divided by audience. A campaign spending modestly against a large audience adds frequency slowly and can run the same ad for weeks, while the same creative against a 20,000 person audience at the same budget can cross 3 in days. The creative is identical, the exposure schedule is not.