A model proposes places worth looking at. It never gets to decide whether they are real — that is settled by fetching each one from its own live source.
Your website, a couple of competitors, and one line about the customer. Under a minute of typing.
Hundreds of candidates, each fetched from its own live source. About a third turn out to be dead and never reach you.
Grouped by what it takes — pitch, buy, list, build, partner — with a contact route on every row we could find one for.
A podcast and a directory cannot be proved the same way, so each channel type has its own test — and every row tells you which test it passed.
The feed has to load, parse, and carry an episode from the last six months. Around two thirds of what a search returns fails one of those.
The site has to resolve, a feed has to be findable, and it has to have published inside four months. A post's comment feed does not count.
The site has to be live and the submission route has to be findable. Ten of the twenty-seven big ones block automated checks — we mark those as unconfirmed rather than pretending.
Two independent signals that a firm sells services to your buyer. One is not enough: plenty of product sites have a services link.
The date has to be ahead, not archived, and the page has to offer a way in — tickets, a sponsor form or a call for speakers. A blog post about an event is not one.
A price is shown only when a page states it, with the words beside it saying what it buys. Each one keeps its quote.
On a real run: 142 opportunities proposed, 87 survived, 55 dropped. One podcast in that list had not published an episode in 3,358 days and was still indexed everywhere as though it were live. That is the show a top-100 list would have told you to pitch.