Meta Ads

Meta creative fatigue: read frequency, find the real winner, revive it

When a Meta ad that sold well starts to slide, the reflex is to call it fatigue and order a recut. Sometimes that’s right. Often the ad isn’t tired at all, or it was never the cold winner it looked like. This guide shows how to tell the cases apart in Ads Manager, and what to make next when the ad really is done.

By the Dokiri Studio team · November 6, 2026 · 8 min read

In this article

What fatigue is, and what it isn’t

Creative fatigue is something precise: most of the people an ad can reach have already seen it, and the ones who were going to act already have. Each extra impression lands on someone who scrolled past last time, so click-through slips, cost per purchase rises and spend slows, on that ad first.

Media buyer Bluesense Digital describes every ad as having a finite lifetime spend: it ramps up to a daily peak, holds, then fades. In his experience, push it hard and it burns out in weeks; run it on a modest budget and it can last months (video). That’s a working model, not a Meta rule, but it explains why you need several live ads, not one.

Plenty of drops look like fatigue and aren’t:

  • Everything fell at once. Every ad dipping on the same day points to the account: auction costs, a holiday, the site, tracking. A tired ad fades alone.
  • Spend just jumped. After a big budget rise, results often lag for a few days while delivery finds new people.
  • The offer or the page changed. A price rise, a sold-out size or a slow page hurts every ad pointing there.
  • It was never a cold winner. Its numbers came from people who already knew you. See the third section.

Read frequency the right way

Meta defines frequency as the average number of times each person saw your ad: impressions divided by reach, and an estimate (Meta’s definition). An average hides a lot, so the number means nothing until you’ve fixed three things.

  • The window. Seven, 14 and 30 days give very different numbers. Pick one and keep it whenever you compare.
  • The level. A campaign’s frequency usually sits above its typical ad’s, because the same people see several of its ads. Compare ads with ads.
  • The audience. Past buyers and recent visitors see your ads again and again by design. Blended in, they make a healthy prospecting ad look tired, or hide a tired one.

With those fixed, Bluesense Digital’s rule of thumb for cold campaigns is an ad-level frequency of about 2 or less, read on new people only (video). His window has moved between videos, from 14 days to 30, so treat it as a starting point and calibrate on your own good periods. Long purchase cycles tolerate more repetition: for something people mull over for weeks, like a sofa, he sees campaign frequency near 5 hold up; for impulse buys he wants it under 3.

Is your best ad really a cold winner?

Broad campaigns don’t only reach strangers. Meta can show an ad mostly to people who already follow you, visited the site or bought before, and they convert far more often. The ad tops the ROAS column while doing a retargeting job. Bluesense Digital calls these fake cold winners and checks them in three steps (video). Here is our version, with the decision at the end.

  1. 1

    Look at its frequency

    Compare it with the other ads in the same campaign, over the same window. A prospecting ad running far above its neighbours is being served to a small, familiar pool.

  2. 2

    Split it by audience

    Meta lets you define an engaged audience and your existing customers from custom audiences, then break results down into new audience, engaged audience and existing customers (Meta). Availability depends on the campaign type, so look in the Breakdown menu. Read the new-audience row on its own.

  3. 3

    Compare attribution on that row

    Put the default attribution next to 7-day click only, and next to incremental attribution if your account offers it (Meta). Incremental results try to count only the sales the ad caused.

  4. 4

    Decide what the ad is

    If it still clears your break-even point on new people, it’s a real cold winner: brief more concepts like it. If it collapses, keep it running as the retargeting ad it is, but stop counting it as prospecting and stop copying it.

In one account he walks through, an ad reporting a double-digit ROAS came down to under 2× once only new people and incremental sales were counted. Your gap may be smaller, but you won’t know until you look. The break-even ROAS calculator gives you the line to judge against.

Symptom, likely cause, fix

Use this table when results slip. The cheap checks come first: rule out the account and the site before you blame the creative.

SymptomLikely causeFix
Every ad drops on the same day, frequency flatNot fatigue: auction costs, season, site or trackingCheck CPMs, checkout and the pixel before touching creative
One ad: frequency rising, click-through and 3-second plays fallingThe ad is tiredNew openings on the same angle, a new face, a new format
Several similar ads fatiguing togetherToo alike, sharing one pool of peopleAdd a different concept (persona, angle or offer), not another variant
Frequency high early on a narrow personaThe audience is too small for the spendBroaden the persona, or spend less on that concept
Best-ROAS ad, frequency far above the restRetargeting in disguiseRun the three-step check; brief cold concepts elsewhere
Spend piles onto one ad, the rest get almost noneMeta picked a favourite, not a problem by itselfLaunch diverse new concepts; leave the low-spend ads on
Hook still holds, purchases fallThe offer, the page or the stock changedCheck that the page continues the ad and the offer still stands

Reviving a tired winner: new hooks, faces, formats

Once an ad really is tired, the tempting move is a recut: same footage, new order, new music. Bluesense Digital’s experience is that recuts rarely work, because the market has already seen that footage and skipped it. What works in his accounts is new footage at the front: a different person, setting or format, leading into the body that already sells (video).

The body carries the persuasion; the opening decides who stays to hear it. A new opening is a new ad to everyone who skipped the old one, and it costs far less than a whole new ad. Levers, from cheapest to boldest:

  • A new hook kind on the same angle: a question instead of a demo, the result instead of the pain point.
  • A new face. The person on screen is part of who the ad reaches: a retired gardener and a student in a shared flat pull different viewers.
  • A new place. The same script outdoors instead of in a kitchen.
  • A new format. A story instead of a showcase, an unboxing instead of three reasons to buy.
  • A new angle or offer. A seasonal reason, a bundle, a gift with purchase. That’s a new concept, often what the account needed anyway.

With Dokiri Studio: a batch of new openings

Dokiri doesn’t read your ad account: frequency, segments and ROAS stay in Ads Manager. What it does is the making. From your product, a batch of 5 or 10 concepts, each with its own format, person, place and opening from eight hook kinds (question, pain point, result first, pattern interrupt, POV, bold claim, story, demo). No hook kind comes back more than once in 5, and near-duplicates are rewritten or dropped.

To revive a winner, write its angle and offer in the Angle, offer or audience field, so every concept keeps what sold, and let the batch change the rest. Scripts are written for free; you keep or drop each one and pay only for the videos you make. In the ZIP, ads.csv has an opening column, so in Ads Manager you’ll know which opening each result came from (export to Meta).

Angle for a revival batch
Our winning angle: dog owners who want the hair off the sofa before guests arrive. Keep that promise and the offer: a second brush head free with the first order. New people, new rooms, new openings. Dry, practical tone.

The limits, plainly. Dokiri makes new 9:16 videos with narration and AI-drawn people; it can’t splice a new opening onto your existing footage, and a video can’t be re-edited once made. If your winner is a founder or creator video, the faithful revival is the same person filmed somewhere new, and that’s a camera job. And an AI-drawn person should never be presented as a real customer.

Frequently asked questions

What frequency is too high on Meta?

Meta sets no threshold. A common rule of thumb for cold campaigns is about 2 or less at ad level over 14 to 30 days, read on new people only. Calibrate it on your own good periods: products people take weeks to buy tolerate more repetition.

Should I turn off a fatigued ad?

Not while it still earns above your break-even point. Let it run on a smaller share of spend while new ads take over, and turn it off once it costs more than it brings.

How many new hooks should I make for a winner?

Practitioners don’t agree, and advice ranges from three to ten per body. Start with three to five that differ in kind, not just in wording, and judge them on new-audience results.

Can Dokiri tell me when an ad is fatigued?

No. Dokiri doesn’t connect to your ad account or report results. It makes the new ads; frequency, audience segments and attribution are read in Meta Ads Manager.

Try it

Tools in this article

Turn your product into ads today

Drop a product photo or a store link: the scripts are written for free, and you pay only for the videos you make.

Make my first ad

Keep reading

All articles