Long-running ads are the cheapest competitive intel you'll ever get. Skip the feed and pull from the Meta Ads Library instead. Run time is a signal, not proof of profitability, but an ad still live after months kept passing somebody's kill test.
Log the video, launch date, hook, customer problem, promise, proof, offer, CTA, creator format, and the first three seconds. Group by angle and hook, not just competitor, then review the batch together and write an one-line hypothesis for why each survived.
The step that's moved numbers for me: test the pattern in your own customer language. Copying the surface of a long-runner usually misses the real reason it worked, and that mistake torched plenty of budgets
