How to spot and fix creative fatigue in LinkedIn ads with AI
How we spot creative fatigue in LinkedIn ads with a monthly, alert-only AI radar: the four checks it runs, the thresholds we use per format and funnel layer, and when a human should overrule it.
Creative fatigue rarely shows up as a crash. An ad loses a bit of its click-through rate one month, a bit more the next, and by the time someone notices, it has been eating budget for weeks.
We saw how big that problem can get at a B2B software client that spends about €200,000 a month on LinkedIn and runs hundreds of ads at the same time. At one point, 32% of its spend went to ads that had been live for more than 180 days.
That number sounds alarming. On its own, it isn’t. A good ad on a large audience can run for a long time. What matters is whether an ad is performing worse than it used to. That’s exactly what is hard to see in Campaign Manager, and exactly what our Ad Fatigue Radar checks every month.
Below are the four checks it runs, the thresholds we use, and the cases where we ignore what it says.
What is creative fatigue in LinkedIn ads?
Creative fatigue is when an ad that used to work loses its effect because the same people have seen it too often. Clicks, views or conversions drop while the audience and budget stay roughly the same.
It’s easy to confuse with two other problems:
- Audience saturation: the audience itself is used up. Every ad in the campaign drops, not just one. New creative helps less than a wider or fresh audience.
- A weak ad: the ad never worked. There is no decline, because there was never a good period to decline from. That calls for a new concept, not a refresh.
The difference matters, because each problem has a different fix. Creative fatigue is the one you can catch early, if you measure each ad against its own past.
Why industry benchmarks won’t tell you an ad is tired
The usual approach is to compare an ad’s CTR with an industry average. That tells you very little.
Fibbler’s LinkedIn ads benchmarks, based on more than 1,000 B2B advertisers, show how wide the spread is. The bottom quarter of advertisers gets a Sponsored Content CTR of 0.17% or less, the median is 0.39%, and the top quarter reaches 0.83% or more. Formats differ even more: the median CTR is 0.11% for document ads and 0.09% for video.
So a 0.4% CTR can be a great result for one ad and a warning sign for another. The only fair benchmark for an ad is the same ad, a month earlier.
How we spot creative fatigue: the 4 checks in our Ad Fatigue Radar
Our Ad Fatigue Radar is an AI agent that runs once a month over every live ad in an account. It doesn’t pause anything. It flags the ads that need attention, and a person decides what happens next. Jasper Hissink, Director of Demand Marketing at Unmuted, runs it for our clients.
Each flag comes from one of four checks.
1. Compare each ad with its own last 30 days
The radar compares an ad’s last 30 days with the 30 days before that, on a rolling basis rather than per calendar month. It looks only at the ad itself, not at other ads in the campaign.
For single image ads, a CTR drop of 20% or more triggers a warning. A drop of 30% or more triggers a stronger one: something has to change.
2. Judge each format on its own metric
CTR is the right signal for single image ads, but not for every format. The radar uses a different metric per format:
| Format | What the radar measures |
|---|---|
| Single image | CTR |
| Video and meme ads | How many viewers keep watching past the first seconds |
| Document ads | Cost per completion and/or cost per download |
| Bottom-of-funnel ads | Cost per result |
A video ad with a low CTR can be doing its job perfectly. Judging it on clicks would flag the wrong ads.
3. Set weekly frequency limits per funnel layer
Frequency is how often the same person sees your ad. The radar checks it per week and sets the limit by funnel layer, because people who already know you tolerate more exposure than a cold audience.
| Funnel layer | Warning | Stronger warning |
|---|---|---|
| Cold audience | 2.5 or more per week | 3.5 or more per week |
| Retargeting | 5 or more per week | 7 or more per week |
A small target account list hits these limits faster, because the same budget is spread over fewer people.
4. Treat age as context, not a trigger
Remember the client with 32% of its spend on ads older than 180 days? The radar doesn’t flag those ads for their age. An old ad only becomes a problem when its performance drops against its own earlier results.
“A good ad on a large audience can last a long time,” says Jasper. “Age has to go together with a drop in performance against the ad’s own performance.”
What does the radar catch that a manual check misses?
Slow trends. Campaign Manager shows you numbers, but it doesn’t make it easy to see how one ad develops over time, across hundreds of ads.
“It’s very hard to see trends developing by hand,” Jasper says. “Say an ad has a CTR of 0.5% one month, 0.4% the next, and then 0.35%. That development is hard to spot yourself, especially when you have hundreds of campaigns live. The radar helps spot the kind of trends a person doesn’t easily notice, partly because LinkedIn Campaign Manager’s reporting isn’t great.”
In that example, the drop from 0.5% to 0.4% is a 20% decline within 30 days, so the radar would flag it at the first step down.
What do you do with a flagged ad?
A flag starts a decision. After every run, we review the flagged ads and replace the ones that need to go with winning ad concepts.
How far you go depends on what the flag says:
| What you see | Likely cause | What to do |
|---|---|---|
| One ad drops against its own past, frequency normal | Creative fatigue | Refresh the hook or visual, keep the concept |
| One ad drops and frequency is high | The audience has seen it too often | Swap in a different concept, or rotate in more ads |
| All ads in a campaign drop at once | Audience saturation | Widen or refresh the audience before making new creative |
| An ad never had a good month | Weak creative | Replace the concept |
When should you ignore a fatigue flag?
When a person who knows the account can see something the rules can’t.
“You always want a human reviewer,” says Jasper. “An ad fatigue radar simply follows the rules. For example, the radar might say an ad should be swapped, but there’s a really good comment under it that strengthens the story.”
Automated detection also produces false alarms by design. In a synthetic test of an automated creative fatigue detector, a researcher at WPP Data Science found it caught 94% of simulated fatigue events, but also raised alerts on 31% of the controls where nothing had changed. That was simulated data, not real campaigns, but the lesson holds: a fatigue detector that never cries wolf will miss real fatigue.
That’s why our radar is alert-only, and why we keep adjusting it. “The radar gets it wrong now and then, but you can adjust the radar when that happens,” Jasper says. “We’ve done that a few times already.”
How often should you refresh LinkedIn ad creative?
LinkedIn’s own advice is to run 4-5 ads per campaign and, every 1-2 weeks, pause the ad with the lowest engagement and replace it with new creative.
That rhythm works for testing. The catch is that it runs on a calendar: it replaces your weakest ad even when every ad is still doing fine, and it says nothing about a winner that slowly wears out.
We prefer to let the numbers set the pace: keep an ad as long as it holds up against its own past and stays under the frequency limits for its funnel layer, and refresh it when it doesn’t. For some ads that means a few weeks. For others, many months.
How to build a lighter version yourself
You don’t need an AI agent to start. A monthly routine gets you most of the way:
- Export ad-level results for the last 60 days from Campaign Manager.
- Split them into the last 30 days and the 30 days before.
- Pick the right metric per format and calculate the change for each ad.
- Calculate weekly frequency and compare it with the limit for the ad’s funnel layer.
- List every ad that crosses a threshold, and review the list before you change anything.
An AI assistant can do steps 2 to 5 for you if you give it the export and these rules. The review in step 5 stays with a person.
FAQs about creative fatigue in LinkedIn ads
Short answers to the questions that come up most when teams start checking for fatigue.
How do I know if it’s creative fatigue or a bad ad?
Look at the ad’s history. Creative fatigue means the ad was performing and is now declining against its own earlier results. A bad ad never had a good period. The first needs a refresh, the second needs a new concept.
What frequency is too high on LinkedIn?
We start warning at 2.5 impressions per person per week for cold audiences and at 5 per week for retargeting. Above 3.5 for cold and 7 for retargeting, we expect something to change.
Should AI pause fatigued ads automatically?
We don’t let it. Our radar only flags. A rule can’t see context such as a strong comment thread under an ad, a campaign that is about to end, or a concept that sales relies on.
Do old LinkedIn ads need to be replaced?
Not because of their age alone. An ad that has run for months on a large audience can still be one of your best. Replace it when its performance drops against its own past.
Want this done for you?
Catching creative fatigue is one part of running LinkedIn ads well. We run LinkedIn ads as part of The Growth Program, with strategy, copy, design, video and paid media in one team, so the ads that replace a tired one come from the same people who know your positioning.
Still comparing options? Our overview of the best LinkedIn ads agencies for B2B SaaS matches each agency to the situation it fits best.