On this pageJump to a section
There is no winning ad. There is an ad that is winning this month.
Every account I have managed since 2014 eventually teaches the same lesson, and it usually arrives as a bad week. A creative that crushed for six weeks starts to slip. Cost per result creeps. Someone says, "we need to find the next winner," and the hunt begins again.
That is the wrong frame. The accounts that last do not hunt for a winner. They run a creative pipeline, because they know every ad decays.
What is ad fatigue, and why does it happen?
People see the same thing too many times and stop reacting.
What Meta says
Meta's developer guidance on optimizing campaigns lists ad fatigue as a reason costs rise: audiences become overexposed to the same ads. Its recommendation is to regularly rotate in new visuals to find strong performers and counter fatigue.2
Why it feels sudden
Fatigue builds slowly and shows up all at once. An ad reaches its best audience first. Over time the pool of people likely to respond shrinks, the same people see it again, and the system has to reach further to spend your budget. Costs rise in a way that feels like a cliff, even though the slope started weeks before.
How do you spot fatigue early?
Watch three numbers together. One alone can mislead you.
| Signal | What it may mean |
|---|---|
| Frequency rising | The same people are seeing the ad more often |
| Click-through rate falling | The creative is less interesting to the people reached |
| Cost per result rising | You are paying more to get the same outcome |
| All three together, offer unchanged | Strong sign of fatigue |
| Only cost rising, other numbers steady | Check the market, season and competition first |
Rule out the other causes
Before you blame the creative, check what else changed: the offer, the page, the audience, the season, a tracking change, a competitor, a holiday. Fatigue is a diagnosis you reach after you rule out the others, not before.
Watch the comments
Comments are a free early warning. When people start writing "I keep seeing this," the data usually follows.
How should you plan creative refreshes?
Before you need them. Production is slower than decay.
Build a pipeline, not a project
Treat creative like inventory, the idea behind a creative bank. At any moment you should have ads in market, ads in testing, ads in production and ideas in the backlog. If the only thing in your pipeline is what is live today, you are one slow week away from a scramble.
- BacklogIdeas ranked by concept, not by headline
- ProductionAds being made before the old ones fade
- TestingNew concepts running beside current ads
- In marketProven ads holding the budget
Match cadence to spend
Higher spend and smaller audiences burn through creative faster. I will not give you a universal number because it does not exist. Start with a steady rhythm, such as new tests every week or two, and adjust by watching how quickly your best ads fade.
Why does Andromeda reward distinct concepts?
Because the system is choosing among far more ads than it used to, and it likes variety.
What Meta has said
Meta described Andromeda as the stage that selects ads from tens of millions of candidates down to a few thousand relevant ones. It said the number of creatives in its systems is expected to grow significantly.1
Increased ad diversity can improve people's experience with ads and drive better advertiser outcomes.
What that means for how you make ads
That is Meta's statement, and the inference is mine: if the system values diversity, ten ads that are really one idea may count for little. I would rather have four ads built on four reasons to buy than ten ads that swap a headline.
Dara Denney, who says she has spent over $100 million on Meta ads, warned about what happens when a team only iterates, in a video she posted on November 21, 2024:3
all of your creatives end up looking and feeling the same to users
Her fix is a split of 90% net-new tests to 10% iterations.3 That is her rule of thumb from her own accounts, so test it in yours. The direction is what I take from it: new concepts first, small edits second.
VideoDara Denney
10 Key Iterations I Make AFTER Finding a Winning Ad Creative on Meta
Her ten iteration types, from messaging swaps to format changes. The video is sponsored by Storyblocks, which she says at the top.
Four levels of creative change
- Concept. The core reason to buy: the problem you lead with, the promise, the angle.
- Format. Video, static, carousel, a person talking, a demo.
- Hook. The first line or first second.
- Detail. Colors, fonts, small wording. The least likely to change results much.
Changes at the top of the list move results. Changes at the bottom mostly extend an ad's life by a little. When refreshing, spend your effort at the top.
What makes a concept actually new?
A new reason to care, not a new coat of paint. Take any offer and ask what different people might be thinking when they first see it.
- The problem angle. Lead with the pain: the thing that keeps them up or costs them money.
- The outcome angle. Lead with the result they want, described in their words.
- The objection angle. Lead with the doubt: "I have tried this before and it did not work."
- The myth angle. Lead with something most people believe that is wrong.
- The proof angle. Lead with a real result or a real story, shown with permission.
Five angles, five different ads, even if the offer is identical. A person who ignored one may stop for another. And the system has genuinely different candidates to choose from.
When should you turn off an ad, and when should you leave it alone?
Less often than your anxiety suggests.
Do not judge on noise
Meta says the learning period stabilizes with a minimum of 50 events over a 7-day period, and that significant edits can trigger a new learning period.2 A few bad days on a small number of events tell you very little. Set a rule in advance for how many results you need before you judge.
50
Do not edit winners in place
Meta's guidance suggests keeping adjustments minor and avoiding significant edits that can reset learning.2 When you want to try a new angle, launch it as a new ad. Leave the proven ad running until it tells you it is done.
What is a simple creative calendar?
- Week 1: Launch two new concepts alongside current ads.
- Week 2: Review the early signals. Do not cut anything without enough results.
- Week 3: Shift budget toward whatever is working. Brief the next two concepts.
- Week 4: Retire what has faded. Log what you learned about each concept.
Then repeat. Every ad decays, so the calendar is the strategy. After a few cycles you will know your account's rhythm better than any benchmark could tell you. If you want that rhythm wired to your tracking and follow-up, that is part of the system I build for clients.
The myth of the winning ad is comforting because it promises an end to the work. The truth is better. Once you have a creative pipeline, you stop fearing the decay, because the next ad is already in line.
Keep building.Ekai
Questions people ask
What is ad fatigue on Meta?
Ad fatigue is the decline in performance that happens when the same people see the same ads too often. Meta's guidance describes rising costs when audiences have been overexposed to the same ads and recommends regularly rotating in new visuals.
How do I know when an ad is fatigued?
Look for a rising frequency alongside falling click-through rate and a rising cost per result, with no change in your offer or audience. One signal alone is not proof, but the three moving together usually is.
How often should I refresh Meta ad creative?
There is no universal schedule. Plan for new creative on a steady cadence based on your spend and audience size, and watch the signals. Accounts with larger spend and smaller audiences tend to need fresh creative sooner.
Why do distinct concepts matter with Meta Andromeda?
Meta says its Andromeda retrieval system selects a few thousand ads from tens of millions of candidates, and that increased ad diversity can improve people's experience and advertiser outcomes. Genuinely different concepts give the system more to choose between than small variations.
Sources
- 1
- 2
- 3
