231,885 of the Meta ads we hold started in the last 180 days
That is about 94% of the 245,937 ads in our library, though the two figures come from counters built differently. Part of the recency is the market and part may be our own collection, and this snapshot cannot separate them.
Ad start dates, cumulative windows to 14 August 2026
- 94.3%
- Started within 180 days
- 4,009
- Started over a year ago
- 245,937
- Meta ads we hold
245,937 Meta ads · Snapshot of 14 August 2026, 04:46 UTC
231,885 of the 245,937 Meta ads in our library started within the last 180 days. That is 94.3% of them. 241,928 started within the last year, which leaves 4,009 that did not, 1.6% of the corpus.
So a competitor's ad tab is a picture of now. It is not a history. Open one, see thirty creatives, and close to all of them are things the company started this half-year. What it ran in 2025 is not on that page, and no filter on that page will bring it back.
The scarcity cuts the other way too. An ad that has survived a long time is unusual in this data, and that is the reason to open it.
Part of this skew may be our own collection
The ads feed drops creatives we could not copy into our own storage before it counts them. Whether the library counter used here does the same is not established by this snapshot. The links those creatives sit behind expire with age, so any such exclusion would fall harder on old ads. Read the age skew below as partly market and partly collection, with the split unmeasured.
177,697 ads started between 30 and 180 days ago
Meta ads banded by start date, derived from the cumulative window counts in the snapshot.
Bands are subtractions between adjacent windows, because each window in the snapshot nests the one before it. The oldest band is the library total of 245,937 minus the 241,928 counted inside a year. Those two figures come from counters computed differently, so it is a residual.
The figures behind this chart
| Under 7 days | 3,024 |
| 7 to 14 days | 10,689 |
| 14 to 30 days | 40,475 |
| 30 to 180 days | 177,697 |
| 180 to 365 days | 10,043 |
| Over 365 days | 4,009 |
245,937 Meta ads · Snapshot of 14 August 2026, 04:46 UTC
The same ads counted cumulatively, the way the snapshot reports them. Each window contains the one above it.
| Started within | Ads | Share of 245,937 |
|---|---|---|
| 7 days | 3,024 | 1.2% |
| 14 days | 13,713 | 5.6% |
| 30 days | 54,188 | 22.0% |
| 180 days | 231,885 | 94.3% |
| 365 days | 241,928 | 98.4% |
The window counts and the library total are two counters in the same snapshot, computed differently. Subtracting one from the other leaves 4,009 ads outside the 365-day window. We print that as a residual and reconcile the two counters nowhere.
245,937 Meta ads · Snapshot of 14 August 2026, 04:46 UTC
The steep stretch runs from 30 to 180 days: 177,697 ads, 72.3% of the corpus, started in that band.
The near-term bands are a different thing and should not be read as advertiser behavior. 3,024 ads started in the last seven days against 40,475 between 14 and 30 days ago, and that gap is mostly our own fetch cadence. An ad reaches our library when a scrape run next reaches the page it sits on, not when it goes live. The start dates themselves are recorded exactly. What is uneven is when we arrived.
The far end is where the finding sits. 14,052 ads, 5.7% of the corpus, started more than 180 days ago. 4,009 started more than a year ago.
An ad past 180 days sits in the oldest 5.7% of what we hold
What run length is worth on a performance budget, and what it is worth on a brand budget.
We publish no ad spend. Meta discloses spend only for political and issue advertising, so it does not exist for ordinary commercial ads, for us or for anyone else. What is left to read is how long a creative has run and how many variants sit behind it.
Run length is a strong prior when the advertiser is buying installs. A performance team watches cost per install daily and cuts creative that misses its target within days. A creative live for six months has cleared that bar over and over.
It is a weaker prior when the campaign is a brand buy on a fixed budget. That creative runs because the flight was booked to run, and its age describes the media plan more than the results. The ad itself usually tells you which one you are looking at: the destination, the call to action, and whether the copy asks for an install at all.
A long run does not prove an ad is profitable. It says the ad kept passing whatever test its owner applies.
The threshold worth acting on
No ad can have run longer than the time since it started, so the start-date distribution caps run length. An ad still live that started more than 180 days ago is inside the oldest 5.7% of the 245,937 ads we hold. Past a year it is inside the oldest 1.6%. Take a competitor's ads, order them by days running, and open the ones past 180 days first. Under 30 days you are looking at 22.0% of the corpus, where run length carries close to no information.
Every figure here counts the ads in our library, not the ads that ran
Four things this distribution cannot see: the creative copy step, our fetch cadence, run length, and every network that is not Meta.
The ads feed drops creatives we could not copy. The feed excludes them before it counts them. Whether the library counter behind every figure on this page does the same is not established by this snapshot. The links those creatives sit behind expire with age, so any such exclusion would fall harder on old ads. That mechanism would produce the exact skew this article reports, and these counts cannot show whether it did. Read the 94.3% as partly what advertisers do and partly what we hold.
A start date is not a run length. The snapshot bands ads by when they started, not by how long they ran. Take two ads that started 200 days ago. One stopped after a week and one is still live, and both sit in the same band. Run length has to be read off each ad's own start and end dates.
Our fetch cadence shapes the near-term bands. We reach a page when a scrape run reaches it. So the 7-day and 14-day counts move with that schedule as much as with anything advertisers did.
We hold Meta ads and no others. An app buying on any other network contributes nothing here. These figures describe Apptonic's own library as of 14 August 2026. The library is a growing sample of the app market, not a census.
- 94.3%
- Meta ads started within 180 days
- 61.1%
- Organic videos published within 180 days
- 11.9%
- Organic videos older than a year
245,937 Meta ads and 565,898 organic videos · Snapshot of 14 August 2026, 04:46 UTC
The organic half of the library is older. 345,917 of the 565,898 videos we hold were published within 180 days, 61.1% of them, against 94.3% for ads. 67,162 videos, 11.9%, predate the last year, against 1.6% of ads.
The gap is worth printing and it settles nothing. A creative-copy exclusion would apply to ads and not to video. That difference alone could produce a gap of this shape. The two corpora also come from different scrape jobs on different schedules. Read the contrast as consistent with a faster ad treadmill, and not as a measurement of one.
How we counted
Every figure comes from one frozen snapshot of the library, harvested at 04:46 UTC on 14 August 2026. Ad ages come from the cumulative start-date windows in global/periods-ads.json, video ages from global/periods-organic.json, and both library totals from global/dataset-totals.json. The 3,996 apps holding at least one ad come from global/coverage-stats.json, generated at 04:29 UTC the same morning.
The windows nest. The 30-day count contains the 7-day count, and the 180-day count contains both, so the bands in the chart are subtractions between adjacent windows. Ads are banded by the start date the public ad library reports for them and videos by publish date. Both dates are recorded exactly, unlike the app estimates elsewhere in our data, which are third-party modeled figures and appear nowhere on this page.
The oldest band is a residual. The window counts and the library total are counters computed differently. Subtracting one from the other leaves 4,009 ads outside the 365-day window. That does not establish that each of those 4,009 carries a start date older than a year. We print it as a residual and reconcile the two counters nowhere.
The ads feed drops creatives we could not copy into our own storage before it counts them. Whether the library counter used on this page does the same is not established by this snapshot. Any such exclusion is age-dependent, because the links creatives sit behind expire and an older ad is likelier to have lost its link. So read the age skew as partly market and partly collection, with the split unmeasured. The near-term windows also carry our own fetch schedule: an ad enters the library when a scrape run reaches its page.
We publish no ad spend, and no run length is derived from these counts. Nothing here is modeled.
- Meta ads held in the Apptonic library
- 245,937 rows · Snapshot of 14 August 2026, 04:46 UTC
- Meta ads with a start date inside the last 365 days
- 241,928 rows · 365 days to 14 August 2026
- Organic videos held in the Apptonic library
- 565,898 rows · Snapshot of 14 August 2026, 04:46 UTC
- Apps holding at least one Meta ad
- 3,996 rows · Snapshot of 14 August 2026, 04:46 UTC
Frequently asked
- Can I see what a competitor was advertising two years ago?
- Not from this data. 4,009 of the 245,937 Meta ads we hold sit outside the last year, 1.6% of the corpus. An ad tab reads as a picture of current advertising. Part of that is how fast app creative turns over. Part may be our own collection. The ads feed drops creatives we could not copy into our own storage before it counts them. Whether the library counter used here does the same is not established by this snapshot.
- How long does an ad have to run before the run length means something?
- Use 180 days as the first cut. 14,052 ads, 5.7% of the 245,937 we hold, started more than 180 days ago. An ad still live past that point is scarce in our corpus. Past a year it sits inside the oldest 1.6%. Under 30 days the run tells you close to nothing: 54,188 ads, 22.0% of the corpus, started inside that window.
- Does a long run mean the ad is profitable?
- No. We hold no spend and no revenue for any ad, and Meta discloses spend only for political and issue advertising. A long run means the creative kept passing whatever test its owner applies. For a team buying installs against a cost target that test is strict and the signal is strong. For a brand campaign on a booked flight, the ad runs because the flight runs.
- How many apps do you hold Meta ads for?
- 3,996 apps in the library hold at least one Meta ad. That figure and the 245,937 ads on this page come from two different counters in the same snapshot. An app holding one ad and an app holding 400 count the same in the first and differently in the second.
- Is the run-length sort in the product an exact ranking?
- It groups ads into seven-day run bands and rotates between advertisers inside a band, so it orders bands and not individual ads. To rank single ads by days running, read each ad's own start and end dates.
- Does any of this include TikTok ads?
- No. Every figure here is Meta. TikTok ads are not in the product, so an app that advertises only on TikTok contributes nothing to these counts.