Meta publishes every active ad on the platform in the Ad Library, free. Most founders open it once, search their category, see a wall of noise and close the tab. The difference between noise and intelligence is knowing what to search and which two numbers to read.
Searching so results mean something
- Search competitor names first. The library matches keywords against ad text, so name searches return both their ads and everyone who mentions them. Check the advertiser name on each result before drawing conclusions.
- Then search buyer language. Not your category's jargon, the words a buyer would type: what the product does for them. Category jargon returns other marketers, buyer language returns the ads your buyers actually see.
- Filter to active ads. Dead creative teaches history, active creative teaches what is paying for itself right now.
The two numbers that matter
- Days live. An ad running 100+ days is paying its way. Nobody funds a loser for a quarter. Longevity is the single most honest signal in the library.
- Version count. An advertiser running several near-identical variants of one creative found a winner and is scaling it. The clone count tells you more than the ad copy does: the run is an absence of a decision, the clones are a decision.
The noise to ignore
In hiring-adjacent and consumer categories especially, keyword searches surface recruitment ads and job listings that share your words without selling to your buyer. An ad addressing your audience without selling them anything is a destination, not a competitor. Skip it, whatever its days-live number says.
The weekly ritual: one search per rival plus two buyer-language searches, fifteen minutes. Write down anything that crossed 90 days since you last looked. Those are the structures worth borrowing, covered further in how to spy on competitor ads properly.
found does this work as a product.
Run the free scan and get the measured version for your own site: who the AI models recommend for your category, where your SEO stands page by page and a fortnight of work with every claim verified. $30 a month after, cancel in two clicks.
Run the free scan →