How often should you refresh your ad creative?

How long does the average ad creative last before it gets pulled?

Most do not reach a month. Motion's analysis of 1.29 billion dollars in Meta ad spend, covering 578,750 creatives across 6,015 advertiser accounts from September 2025 to January 2026, found roughly half are turned off before 28 days.

The same report puts winners at about 5 percent of everything launched, counting a winner as a creative that takes at least 10 times the account's median spend, with a floor of 500 dollars. Read those together and the median creative is not fatiguing at all. It is cut early, long before anyone asks whether it went stale.

Motion counts what advertisers did, not why they did it. An ad switched off on day 9 could have fatigued, or could have lost to the ad sitting next to it in the same ad set. A spend benchmark cannot separate those, and the report does not claim to.

We plan around that ratio. When 5 percent of what you ship carries the account, the useful question stops being how long a good ad survives and becomes how many you can put into the auction this month.

What frequency level actually predicts ad fatigue on Meta?

Exposures to a single creative, rather than any account-wide frequency figure. Admove's August 2026 guide reports diminishing returns starting around three exposures to the same ad and negative returns past six.

Because that count is per creative, one universal number falls apart the moment you compare two ad sets. A retargeting pool of a few thousand people runs through six exposures in days. A broad prospecting audience may never get there, and cutting on the same schedule throws away an ad that was still working.

So watch the pair, CTR and CPM, against the account's own baseline rather than an industry chart. When CTR slides while CPM climbs on the same audience, you are paying more to show something people have already seen. That is the trigger. A date is not.

How much does repetition really hurt click-through rate?

The figure in circulation is a 45 percent drop in click likelihood after four exposures to the same ad. It reaches us through Admove's guide of August 17, 2026, which credits Meta's own research, rather than through anything Meta published where we could read the method.

We treat it as direction, not as a coefficient. Nothing we read links the sample or the placements behind that 45 percent, so we would not build a forecast on it. Quote it if it helps, and say where you got it.

The audience side is easier to check. A 2025 Harris Poll, cited in the same guide, found 61 percent of US adults say they are less likely to buy from a brand showing repetitive ads, and 88 percent say repetitive ads make them pay less attention. Stated attitudes are weaker evidence than spend data.

Does the refresh schedule change by platform or audience type?

Yes, and the published windows differ by weeks. Admove's guide lists 2 to 3 weeks for Meta broad audience campaigns and 3 to 7 days for TikTok, with Google Performance Max and LinkedIn at 3 to 4 weeks per asset.

Every row above comes from that one August 2026 guide, so read it as one publisher's working numbers rather than five independent studies. The spread is the part we use. A TikTok slot and a Performance Max asset are not on the same clock, and running both on a single refresh cadence starves one of them.

So the cadence belongs to the placement, not to the account. Put a TikTok slot on a three-week clock and you spend two of those weeks buying attention that ran out in the first.