We measured 57,643 LinkedIn ads.
Here's what actually survives.
Three years of European LinkedIn advertising — 7,773 advertisers, all five formats — measured to answer one question: why do some ads run for years while most die in two weeks? The findings surprised us.
Exactly what this study covers
| 57,643 | ads — every ad in our European index |
| 7,773 | distinct advertisers across 64 countries |
| Jun 2023 – Jul 2026 | observation period, by ad start date |
| All 5 formats | image 32,141 · video 12,845 · InMail 8,200 · carousel 3,086 · influencer 1,371 |
| Source | LinkedIn's own public EU ad transparency library |
42,901 ads had a measurable start and end and carry every survival figure below. Durations are a floor: ads still running are measured to the last observation.
1 · Most LinkedIn ads die almost immediately
For each measurable ad we recorded how long it stayed live. Watch how fast the field thins:
The median is 13 days. Three quarters of European LinkedIn advertising is switched off inside a month. When people say LinkedIn ads "don't work", this is usually what happened: they ran a few, saw nothing, and stopped.
Or as a picture — start 100 ads today, and this is what's still alive in three months:
6 survive. 94 are switched off, out of budget, or replaced.
2 · Check your own industry
The averages hide big differences between markets. Pick yours — these numbers come from the same live index, not a survey:
Want the full picture for your market?
The Ad Playbook shows every surviving ad in your category, your named competitors decoded, and three ready-to-run ads written for your company — built from this exact index.
See what's in the playbook →The survivors — straight from the index
Real ads that outlasted the field — led by the underdogs: small companies out-reaching giants — then the longest runners in every format. This is the raw material your playbook is built from.
3 · Winning ads and losing ads are written the same way
We split the corpus into ads that survived 90 days (2,763) and ads that did not (40,138), then compared the traits marketers are told decide outcomes:
| Trait | Survived 90d | Did not |
|---|---|---|
| Average headline length | 28 chars | 26 chars |
| Average body length | 379 chars | 360 chars |
| Headline is a question | 3% | 4% |
| Headline contains a number | 9% | 11% |
| Addresses the reader as "you" | 43% | 42% |
4 · What separates them is how many you run
We grouped 5,862 advertisers by how many distinct ads they ran, and asked one question: did any of their ads reach 90 days?
| Ads the company ran | Companies | Ever produced a 90-day ad |
|---|---|---|
| 1 | 1,009 | 1% |
| 2–3 | 868 | 3% |
| 4–9 | 2,293 | 9% |
| 10–24 | 1,649 | 23% |
| 25 or more | 43 | 63% |
Run 1 ad
Run 25 ads
One ad gave a 1% chance of ever producing something that lasted a quarter. Twenty-five or more gave 63%. That is the largest effect in the dataset, and it dwarfs anything found in the copy.
5 · Format shifts the odds more than wording does
| Format | Share of ads | Median life | Pass 90 days |
|---|---|---|---|
| Single image | 55.8% | 13d | 6% |
| Video | 22.3% | 13d | 5% |
| Message ads (InMail) | 14.2% | 16d | 11% |
| Carousel | 5.4% | 16d | 5% |
| Person-fronted / influencer | 2.4% | 4d | 0% |
Share of each format's ads that pass 90 days — bars scaled to the leader.
Message ads survive at roughly double the rate of everything else — and they are the format most companies never try. The influencer row runs against current fashion: median life four days, none in this sample passed 90.
6 · What we would do with this
Budget for iteration, not for a launch. Three ads and a verdict means you stop at the 3% mark and conclude LinkedIn doesn't work.
Stop optimising the headline first. Cheapest thing to change, least decisive on this evidence.
Try a message ad. Underused, survives at twice the rate.
Study what already survived. Every ad here is public. Ones running six months are working for someone with a budget, in your market, right now.
7 · The conclusion
- LinkedIn ads are a survival game, not a writing contest. The median ad lives 13 days, only 6% see a quarter — and the survivors' copy is indistinguishable from the failures'. Nobody can tell you which ad will win. The data can only tell you how winners are found.
- They are found by volume. One ad gives you a 1% chance of a long-runner; twenty-five give you 63%. The companies that "win LinkedIn" are not writing better — they are still at the table when their winner shows up.
- The shortcut is that the table is public. Every surviving ad in your market is visible in LinkedIn's own library — what it says, how long it has run, who is paying for it. Iterating from what already survived is cheaper than iterating from a blank page. Almost nobody does it.
The winners aren't writing better ads. They're finding theirs faster.
The ads in your market are public too
This study is the market as a whole. The useful question is narrower: which ads run in your category, which survived, and what are your competitors doing right now? The Ad Playbook answers that for your company — built from this same index, delivered in minutes.
Get your company's playbook →See which LinkedIn ads are working in Europe — then copy them.