The 27-Day Shelf Life of a Meta Ad

← All posts · Updated August 26, 2026 · 7 min read · by the Adshelf team

"How long do Facebook ads run?" is one of those questions every media buyer asks and every answer dodges — because Meta deletes an ad from its Ad Library the day it stops, so the historical record needed to answer it doesn't exist. Unless someone was archiving.

We were. Adshelf tracks 35,510 Meta ads across 258 named brands, keeps every one after it stops, and records the run dates. That makes the question answerable with data instead of folklore. Every number on this page recomputes from the live shelf — last updated August 26, 2026.

The median ad is 27 days old

Take every ad running right now and sort by age: the middle one launched 27 days ago. Half of everything you see in your feed — polished, confident, expensive-looking — is younger than that.

That number should recalibrate how you read a competitor's account. The ads are not a stable catalog; they're a conveyor belt. What looks like a strategy is mostly a stream of short-lived tests flowing past a small, old, load-bearing core. The skill is telling the two apart — and age is the tell.

Death comes in the first ten days

Stopped ads cluster hard at the young end: the modal outcome for a Meta ad is to be killed inside its first ten days. In the last seven days alone, 12 ads across our tracked brands were launched and pulled that fast.

This is the most honest signal advertisers produce. Nobody announces a failed test — they stop paying for it, and Meta erases the evidence the same day. Which is why we archive it: the graveyard is where the real lessons live, because it shows what a competitor tried and abandoned, not just what they show.

Past 30 days, the odds invert

An ad that survives a month has crossed a real line — 44% of live ads are past it, and only 9% reach 90 days. It has been through at least two budget reviews and kept its seat; someone with access to the numbers keeps choosing it. That's why 30 days is the survivor threshold across the product: past it, an ad stops being an experiment and starts being evidence.

The far end of the curve is genuinely far: the longest-running ad on the shelf has been live for 1076 days. Years of someone paying, every day, because it converts. When you're hunting for what works in a category, those outliers are the entire prize — and run duration is the only public signal that finds them.

Staying power belongs to angles, not just ads

Group the survivors by persuasion angle — the argument the ad makes, which our classifier tags on every ad — and the medians spread apart in a way that should change what you brief next:

Read the two columns against each other. Founder Story ads hold a 75-day median on just 1% of the live set: the strongest survival on the least crowded ground. Meanwhile the angles with the biggest shares — where most of the shelf fights — sit further down the table.

The pattern generalizes: the crowded angles die faster. Partly saturation, partly that easy angles attract low-conviction tests. Either way, the opening is usually where proof is high and traffic is low — that's the logic behind The Gap view in the product, and you can check the current spread on the live benchmarks.

What to do with this on Monday

Sort your own account by run duration and look at the top five — that's your real strategy, whatever your deck says. Then sort a competitor the same way and read their oldest ads' angles; you now know what their spend believes. And when you brief the next round of creative, steal from the survival table, not the feed: the feed shows what's new, which this data says is mostly what's about to die.

Every figure here refreshes as the shelf grows, and the underlying numbers are published as an open JSON feed if you'd rather cite them directly. Want the same read on your own category? Put your competitors on a shelf and the curve builds itself.

Straight answers