Short answer: the ads still running after six months say more. In 1,987 active US Meta ads with copy, the ads live for 180 days or more had a median of 31 words of primary text, against 19 words for ads under 30 days old. Long runners had longer copy in 10 of the 10 categories we looked at. The opening line itself barely changed: about 52 characters either way.
Copy length: new ads vs long runners
| Ads under 30 days | Ads live 180+ days | |
|---|---|---|
| Median words | 19 | 31 |
| Share with 20 words or fewer | 51% | 35% |
| Share broken into 4+ lines | 25% | 41% |
| Share with any emoji | 39% | 48% |
| Ads | 293 | 557 |
Median words of copy, by category
| Category | Under 30 days | 180+ days |
|---|---|---|
| Consumer apps | 15 | 21 |
| Beauty | 18 | 39 |
| Education | 29 | 42 |
| Fashion | 18 | 20 |
| Food and drink | 21 | 29 |
| Health and supplements | 29 | 36 |
| Home | 15 | 47 |
| Travel and local | 18 | 30 |
| Business software | 32 | 56 |
| Financial and home services | 16 | 20 |
The first line: what the hook does
We looked at the first sentence of each ad’s text, the part people read before “See more”. The differences between new ads and long runners are small, so treat these as tendencies, not rules.
| The first line… | Ads under 30 days | Ads live 180+ days |
|---|---|---|
| Has a number | 18% | 23% |
| Asks a question | 10% | 11% |
| Says “you” or “your” | 30% | 28% |
| Opens with an emoji | 13% | 9% |
| A word in capitals | 6% | 10% |
| Mentions an offer | 11% | 11% |
| Social proof | 3% | 5% |
- Numbers show up more in long runners (18% to 23%): a price, a count, a result, a timeframe.
- Opening with an emoji is less common among long runners (13% to 9%), even though emoji anywhere in the copy is more common (39% to 48%). Emoji work as bullets, not as the hook.
- Questions, offers and “you” barely move. They are common in ads of every age, so they are not what separates the keepers.
The button
“Learn more” is on 31% of long runners against 24% of new ads; “Shop now” is on 40% against 46%. The ads that last more often send people to read first and buy second.
What to do with this
Write the short version and the long version of every ad. The short one is easy to approve; the long one, broken into lines, with a specific number near the top, is what this data says advertisers keep paying for. When you study a competitor, read the copy of their oldest ads, not their newest. Hermoso’s competitor teardown does that for you, and our ad copy length checker shows where each platform cuts text off.
How we collected the data
On October 26, 2026 we searched Meta’s public Ad Library for 80 product and service keywords, eight in each of ten categories (beauty, health and supplements, fashion, home, food and drink, consumer apps, business software, financial and home services, education, travel and local), with the country set to the United States and only active ads. We kept the first page of results Meta returned for each keyword: 2,360 ads. Ads that Meta groups as versions of the same ad were counted once, which left 2,073 ads from 1,179 advertisers. “Days live” is the number of days from the start date Meta shows to October 26, 2026.
What this data can and cannot say. Every ad here was still running on the day we looked, so the numbers describe which ads survive, not which ones sell. An advertiser usually keeps paying for an ad that works, which is why run length is the most useful public signal there is, but some long runners are simply forgotten. Meta decides which ads appear first for a keyword, so this is a large sample, not a census. Nothing here was extrapolated or modelled: every figure is a count or a median of the ads in the file.
Download the data (CSV): one row per ad with its category, format, days live, number of versions, copy length, first-line features, button and, for videos, length in seconds. It holds no advertiser names and no ad copy, so it can be shared freely; please cite “Hermoso Research, October 26, 2026”.
More from this dataset
Which ad formats run longest · Video ad length vs time live · All research
Frequently asked questions
Do long-running Facebook ads have long or short copy?
Longer. In our sample, Meta ads live for 180 days or more had a median of 31 words of primary text, against 19 words for ads under 30 days old, and long runners said more in 10 of 10 categories.
What makes a good ad hook?
In our data the first lines of long-running ads more often contain a number and less often open with an emoji. The differences are small, so test hooks rather than follow a rule.
Which call to action do long-running ads use?
Shop now and Learn more lead both groups. Learn more is on 31% of ads live 180+ days against 24% of new ads.
Can I use this data?
Yes. The CSV is free to download and use with credit to Hermoso Research, 2026-10-26. It holds no advertiser names and no ad copy.
Start free, no card needed. See which of your competitors’ ads have run the longest.
Start free → See pricing