Short answer: run a new ad long enough to get a fair read, about a week of steady delivery, then keep it for as long as it holds up. Good ads don’t expire. Some of the ads we track have been live for well over a year.
That’s my answer. Here’s the data behind it, and why “how many days” is the wrong way to frame the question.
How long real Facebook ads run
We track the public Meta Ad Library ads of 40 brands across beauty, supplements, software, fitness and more: 2,021 ads in all. For each one we know the day it started and the last day we saw it.
| How long the ad ran | Share of ads |
|---|---|
| At least 7 days | 77% |
| At least 30 days | 41% |
| At least 90 days | 17% |
| At least 180 days | 4% |
The median ad ran 20 days. The longest has been live for 482 days and counting.
That median hides two very different groups, though, and you need to see them apart.
Most ads get cut fast
Of the 807 ads these brands had already switched off, the median one lasted 8 days. 41% were gone within a week, and 66% within two weeks.
That’s on purpose. Brands that test a lot launch ads in batches, give each one a short window and cut the ones that don’t hold up. Liquid I.V. is the clearest case: its median ad ran 5 days across 219 public ads. If your ad gets cut in its first week, you’re in good company.
The ones they keep run for months
The 1,214 ads still live at our last check tell the other half of the story. Their median age was 39 days, and the top quarter had been running 89 days or more.
Some brands live almost entirely in that group:
- Firmoo: a median of 311 days, with videos live for 482 days.
- Huel: a median of 113 days, and 38 of its 66 public ads past 90 days.
- CrossFit: video ads live for 372 days.
- Saily: a median of 139 days, and a longest run of 204.
An ad that stays on for months isn’t proof it works. We can’t see what these brands paid or what they got back. But brands don’t keep paying for an ad month after month if they think it’s dead weight. A long run is the best public signal you’ll get that an ad’s holding up.
It depends on the brand more than the category
| Category | Brands | Ads | Median run (days) |
|---|---|---|---|
| DTC | 7 | 240 | 61 |
| Software | 5 | 174 | 35 |
| Local services | 3 | 100 | 34 |
| Beverages | 3 | 102 | 28 |
| Marketing services | 6 | 306 | 22 |
| Beauty | 6 | 469 | 16 |
| Fitness | 3 | 179 | 13 |
| Supplements | 3 | 316 | 7 |
Supplements looks like the fastest category, but that’s mostly Liquid I.V. testing at volume. Huel sits in the same category and keeps its ads for months. So don’t copy your category’s average. Look at the specific brands you’re up against and how they behave.
So how long should your ad run?
Here’s how I’d decide:
- Give a new ad about a week of steady delivery before you judge it. Early numbers swing while Meta works out who to show it to. A verdict on day two is mostly noise.
- Judge it against your other ads, not against the calendar. If it’s holding its cost per purchase as well as the rest, it stays.
- Don’t retire an ad just because it’s old. If it’s still holding up, age doesn’t matter. The brands above keep some ads for a year or more.
- Watch frequency, not the date. When the same people see it again and again and fewer of them click, it’s tired. I’ve covered the signs in why your Facebook ads stopped working.
- Have replacements in testing before you need them. Then cutting a tired ad never leaves you with nothing running.
Check what your rivals keep running
The quickest way to calibrate is to look at your own market. Open a competitor in the Meta Ad Library and look at the start dates on their active ads. The oldest ones are what they’ve decided to keep. The StealAds Ad Library shows how long each ad has run, so you don’t have to work it out from start dates, and you can see how to find a competitor’s ads if you haven’t done it before.
How we counted
Every number here comes from public Meta Ad Library ads that StealAds has collected, from the day each ad started to the last day we saw it. An ad we saw in the last week counts as still live, so its run length is a floor, not a final number. These are public ads, not ad accounts: we can see what ran and for how long, never what a brand paid or what an ad brought in. Each brand’s ads have their own page, like Huel’s, if you want to check them.