How to Tell How Long an Ad Has Been Running (Every Major Library)

Every major ad library exposes some version of run time, and it is the single most useful number in each of them. The Meta Ad Library shows a "Started running on" date for each active ad. Google's Ads Transparency Center shows the date range in which an ad was shown. TikTok's Commercial Content Library lists first-shown and last-shown dates. And for native networks — Taboola, Outbrain, MGID, Revcontent — which publish no official libraries at all, run time comes from independent observation: an index like OpenAdLibrary records the first-seen and last-seen date of every captured creative. Here is exactly where to find the number in each library, and the caveats that stop it from lying to you.

Why run time is the number that matters#

Run time is the closest thing public ad data has to a profitability signal. Advertisers kill losing ads quickly — every day a paid placement stays live is a day someone with full visibility into its conversion data decided it was worth funding. A creative on day 3 tells you an advertiser is testing; a creative on day 30+ tells you the test succeeded. That inference, and its boundaries, are unpacked in why a native ad running 30+ days is probably profitable; the short version is that ad longevity plus variation count outperforms every other public signal for separating winners from noise.

Which is why knowing where each library hides its dates is worth ten minutes of your time.

Meta Ad Library: "Started running on"#

Search a brand's page in the Meta Ad Library at facebook.com/ads/library, and each ad card carries a "Started running on" date. Read it with three caveats:

• The date belongs to the ad, not the angle. Advertisers routinely duplicate a working ad into new campaigns or refresh it after a pause; each relaunch gets a new ID and a fresh date. A winning angle can look brand-new because the current copy of it is.
• Cumulative run time isn't shown. An ad that ran six months, paused, and relaunched last week shows last week.
• For most ad categories, only active ads are searchable. Historical archives exist mainly for special categories (politics and issues) and for coverage mandated in the EU, so "not found" usually means "not running right now," not "never ran."

Practical adjustment: judge Meta longevity at the angle level. If the same hook appears across several ads with staggered start dates, the angle has been alive across all of them — that continuity is the real signal.

Google Ads Transparency Center: date ranges#

Google's Transparency Center shows ads from verified advertisers with the period in which each ad was shown, filterable by time range and region. Two reading notes: the ranges reflect when Google served the ad in the region you selected, so a geo filter can truncate what you see; and formats rotate through auction eligibility, so a gap in the range does not necessarily mean the campaign paused. The Transparency Center glossary entry covers what the surface does and does not disclose.

TikTok and the EU-mandated libraries#

TikTok's Commercial Content Library reports first-shown and last-shown dates for ads delivered in the European Economic Area — coverage that exists because the EU's Digital Services Act obliges very large platforms to run ad repositories. The practical trick: because many advertisers run the same creatives globally, EEA-mandated libraries often reveal run-time data for campaigns you are actually competing against elsewhere. The same DSA logic gives you ad archives on several other large platforms; when in doubt, check whether the platform has a "commercial content" or "ad repository" page before assuming no data exists.

Native networks: observation is the only source#

Taboola, Outbrain, MGID, Revcontent, MediaGo, and the MSN feed have no official libraries — no start dates, no archives, nothing. The only way anyone can tell you how long a native ad has been running is to have been watching: capturing ads continuously from live placements and recording when each creative was first and last observed.

This changes what the number means. A first-seen date is a capture event, not a launch event: the ad existed from some unknown moment before the observer first met it, so every native days-running figure is a floor. It also means observation breadth is the quality bar for the data — an index that watches more publishers, geos, and devices meets new ads sooner and holds last-seen dates closer to reality.

That is how OpenAdLibrary computes its days-running figures. The index holds 725,000+ live native creatives across 49 networks, built from 6.8 million+ ad observations (July 2026); each creative carries a first-seen and last-seen date from that continuous capture. Two real examples of what the data looks like in practice: a Taboola home-and-garden creative ("My garden had no butterflies for years — then I hung one of these up") observed for 37 consecutive days, and a hearing-aid ad from Audika at the same 37-day mark — both patterns of an ad someone is very deliberately keeping alive. At the top end, the longest-observed creatives in the index show 38 days of continuous observation, which is currently the cap of the observation window itself — meaning every one of them is at least that old. You can browse the current holders in the longest-running native ads, and run the same lookup for any advertiser with the native ad spy tool.

Reading run time correctly: the caveats table#

Whatever library the number comes from, adjust for these before acting on it:

From run time to decisions#

The reason to care about any of this is what it does to your testing queue:

• The 30-day screen. When researching a vertical, filter to creatives running 30+ days and study only those. That single filter removes the testing noise that dominates any ad feed — it is step one in how to find winning ads.
• Cross-check with variation count. A long-running ad with many active variations is a scaled winner; long-running with one lonely variation may just be a forgotten campaign. The patterns that repeat across the index are documented in what the best-performing native ads have in common.
• Adapt the angle, not the asset. A 30-day runner proves the angle converts for that offer; cloning the creative pixel-for-pixel buys you saturation and none of their funnel. Long-runners also eventually die — creative fatigue comes for everything — so the goal is to be testing your adaptation while the original is still proving the market.

A concrete weekly workflow that uses all of this: pick your vertical, filter the library to 30+ day runners, cluster them by angle, and write down the two angles you have not yet tested. Check each angle's oldest observable date across ad copies — Meta start dates, Transparency Center ranges, native first-seen — to rank them by proven durability. Then adapt the most durable untested angle into your own creative and launch it. That loop takes under an hour and replaces guessing with the market's own survivorship data.

Run time is public, free to read, and quietly encodes the conversion data every advertiser keeps private. Learn where each library stores it, apply the caveats, and you can rank any market's ads by the only vote that counts: money that kept showing up.