What 10,796 Long-Running Meta Ads Have in Common
We studied 10,796 Facebook and Instagram ads that ran for 60+ days in Europe: how long winners run, how long their copy is, and where they are placed.
adthief.ai · · 4 min read
Brands test many ads and keep paying only for the ones that sell. So an ad that keeps running for months is the closest thing to a public “this works” you can get from the Meta Ad Library.
We keep an archive of those ads at adthief.ai. This post looks at what they have in common: 10,796 Facebook and Instagram ads from 5,921 brands that each ran for at least 60 days in Europe.
How we built the data set
- Source: Meta’s public Ad Library, collected by adthief.ai every night.
- Markets: the European Union and the United Kingdom.
- Which ads: ads that had been running for 60 days or more, one per creative, found through searches for the product categories we cover (supplements and health, beauty and skincare, fashion, home, pets, food and drink).
- Snapshot: 29 October 2026.
Two honest caveats. First, this is a study of survivors: every ad here already ran for two months, so it tells you what winners look like, not what share of ads win. Second, Meta doesn’t publish spend or sales for commercial ads, so “long-running” stands in for “profitable”. It is a strong signal, not a guarantee.
Finding 1: Winners run for months, not weeks
| Measure | Value |
|---|---|
| Ads in the study | 10,796 |
| Median days running | 77 |
| Ran 180 days or more | 2,488 (23%) |
| Ran a full year or more | 135 (1.3%) |
| Top 10% ran at least | 245 days |
Nearly a quarter of these ads had passed six months. A creative that works is not used up in two weeks, as many advertisers fear. The brands behind these ads keep a winner running and add new ads next to it, rather than replacing it.
What to do with it: don’t switch off an ad that is still performing just because it is “old”. Watch the numbers, not the calendar. And when you research competitors, sort by start date: the oldest active ads are the ones worth studying.
Finding 2: There is no ideal copy length
We counted the words in each ad’s main text.
| Words in the post text | Ads | Share |
|---|---|---|
| 1 to 20 | 2,599 | 24% |
| 21 to 50 | 2,910 | 27% |
| 51 to 125 | 3,255 | 30% |
| 126 or more | 2,032 | 19% |
The median was 49 words, but the spread is wide: one ad in five had more than 125 words, and one in four had fewer than 20. Short and long copy both win. What matters is whether the first line earns the “See more” click.
What to do with it: test one short version (a hook and an offer) against one long version (a story, proof and objections) for every new angle. Let the results decide.
Finding 3: Most winners use numbers and emoji
| The post text contains | Share of ads |
|---|---|
| At least one emoji | 62% |
| A number | 55% |
| A question | 26% |
| A percentage | 17% |
Numbers make a claim specific (“25 g of protein”, “30-day guarantee”, “2,000 reviews”). Emoji mostly work as bullet points that make long text easy to scan. Questions are less common than you might expect: about one in four.
What to do with it: replace vague claims with a number wherever you honestly can, and use emoji as structure, not decoration.
Finding 4: Headlines are short
The median headline (the bold line under the image) was 5 words. It sits next to the button, so it does one job: say what you get. “Free shipping on your first order” or “Try it for 30 days”, not a second hook.
Finding 5: Winners run everywhere
| Placements | Ads |
|---|---|
| Facebook, Instagram, Messenger, Threads, Audience Network | 5,083 (47%) |
| Facebook, Instagram, Messenger, Threads | 998 (9%) |
| Facebook and Instagram only | 920 (9%) |
| Every placement including WhatsApp | 532 (5%) |
Almost half of the long-running ads run on all of Meta’s main placements, which is what Advantage+ placements does by default. Only about one in twelve is limited to Facebook and Instagram. Brands that keep an ad for months mostly let Meta decide where it shows.
What to do with it: unless you have a clear reason (for example, a creative that only works in Stories), start with automatic placements and make sure the creative works in square, vertical and feed sizes.
Finding 6: Most brands have one winner, not ten
The median brand in our archive had just one ad that ran for 60+ days. Long-running ads are rare even for brands that advertise all the time. That is why it pays to research across a whole market rather than one competitor: the winners are spread over thousands of brands.
What this means for your ads
- Judge ads by results, not age. Keep winners running; add new tests next to them.
- Research by start date. In the Meta Ad Library, the oldest active ads in your market are the best brief you will get.
- Test copy length. Short and long both win; make one of each.
- Be specific. Numbers beat adjectives.
- Make creative for every placement, and let Meta place it.
See the ads behind the numbers
Every ad in this study is in adthief.ai, ranked by how long it ran and how far it reached, with filters for country, industry and format. If you want help turning them into ads for your brand, see how we run Meta ads for our clients.