Public ad libraries show which ads a page is running. They never show spend, results or campaign structure. So when you are looking at a competitor's ads and trying to work out which ones are worth learning from, you are reading a list with no scores on it.
There is one signal the libraries leak anyway: how long an ad has been running.
Why run length is evidence
An advertiser keeps paying for an ad as long as it keeps paying them back. Nobody leaves a losing ad on for a quarter by accident, at least not at any budget that matters. So an ad that has been live for months has survived whatever test its owner put it through, and an ad that appeared on Monday has survived nothing yet.
Creative strategists already work this way. The working rule in performance teams is roughly: thirty days or more is a proven winner, a few days is a hypothesis. The number is not magic. It is the point at which an ad has outlived the learning phase, a couple of optimisation passes, and the first round of fatigue, and is still being funded.
That makes run length the closest thing to a performance signal you can get from a public library. It is not a result. It is the shadow a result casts.
Where it breaks
Run length lies in four ways, and you should know them before you build a reference deck on it.
A long run can be a small budget. An ad can sit live for a year on ten dollars a day because nobody turned it off. Run length tells you the ad was not killed; it does not tell you it was scaled. The fix is to read it next to how many ads the same advertiser is running and how many of those are also old. One ancient ad in a sea of fresh ones is a leftover. A page where most ads are old is a page that has found something.
A short run can be a winner that was replaced. Advertisers refresh winners with new cuts of the same idea. The ad that ran for sixty days gets retired, and its successor is three days old. If you only look at today's library you miss the lineage. The fix is to look at the idea, not the ad: when three ads share an angle and the oldest of them is months old, the angle is proven even if the newest cut is not.
Start dates are not always what you think. On Meta the library gives the date the ad started delivering. On Google the Transparency Center shows a first-shown date, and the advertiser's history can be long but thin. On LinkedIn the Ad Library shows the ad, and that is about it: no start date you can rely on, no history, no spend. If you want run length on LinkedIn you have to record the library yourself, repeatedly, and compute it from when you first and last saw the ad.
Absence is not evidence. An ad you cannot find might be targeted away from you, running in another country, or simply not in the library's index yet. Never conclude that a competitor stopped doing something because a search came back empty. Conclude it when you have been checking on a schedule and the ad dropped out.
How to read it, platform by platform
On Meta, open the Ad Library, search the page, filter to active ads and sort by the start date. Every ad shows when it started. Write the dates down. The ones past thirty days go in the proven pile. Then group the proven pile by the message, not by the format. Five versions of one promise are one angle.
On Google, open the Ads Transparency Center and search the advertiser. You will see first-shown dates and the formats, but no hook to speak of, because search ads answer a query rather than interrupt a feed. Read the offer instead: what is promised, what is discounted, where the ad sends people. The offers that have been shown for months are the ones that convert.
On LinkedIn, open the Ad Library and look up the company. Take a screenshot with today's date on it. Come back next week and do it again. After a month you have the only run-length data that exists for LinkedIn, because the platform does not publish any.
What this costs you to do by hand
Done honestly, for a competitor set of ten pages across three platforms, this is a recurring half day a week, and the LinkedIn part never gets done because nobody keeps the diary.
This is the part MessCube does for you. It reads every ad reaching your audience on Meta, Google and LinkedIn, records when each ad was first and last seen, and shows the run length next to the ad and next to the angle, labelled by where the number came from. For LinkedIn it keeps the history the platform does not. The market read is $39 a month and you can start free with a preview of your own market first.
Either way, the rule holds: pretty is not proven. Look at the dates.