What's still running after 90 days is the signal, not today's

What a public ad library actually tells you

Every advertising platform of size publishes what's currently running, and almost nobody looks. For a commercial advertiser, the public library gives you the creative, platform, start date, active variant count and destination. It gives you no spend, no impressions and no conversions for anyone outside the account.

The inference is longevity, not performance, and it's still worth having: a campaign that survives three months is being paid for by something. One $4.99 product in a comparable library ran live for 482 days, which is the only kind of signal this exercise gives you.

Table what you find, the same columns every time

Build one table, the same columns on every pull: advertiser, days running, who the first line is spoken to, the promise, the price on the destination, the proof type, the awareness stage, and the opening words verbatim rather than paraphrased.

Keep one row at the bottom naming what none of it tells you, so you never quote any of this to a client as performance data. Longevity is a reasonable inference about budget commitment. It is not a conversion rate.

Read the longest-running ad line by line

Read the longest-running advertisement in your category line by line, rather than skimming past it. What actually does the work is rarely the copy itself. It's the format and the opening move, both of which are reproducible once you've seen them clearly stated.

Where this breaks

The common failure is reading performance into longevity you can't actually measure. A swipe-file quote with no screenshot and no run-duration behind it is not evidence, however often it gets forwarded around as if it were.