Finding link buyers in a publisher's outgoing links
Everyone selling links has the same problem: finding people who buy them. The usual answers are cold outreach to anyone with a blog, scraping job ads for "link building", or buying a list. All of them start from people who might spend money.
There is a better list, and it is already public.
The idea
A publisher that sells guest posts links out to the businesses that paid for those posts. Every one of those links is a company that has already spent money on a placement — recently, at a price, with someone.
So instead of exporting a site's backlinks, export its outgoing links. In Ahrefs that is Site Explorer, then Outgoing links. In SEMrush it is Backlink Analytics, then Outbound Domains.
What comes back is a list of every domain that publisher links to. Buried in it are its customers.
Separating customers from citations
Most of that export is not customers. A tech blog links to a platform's documentation, to a news story, to a reference site, to its own social profiles. Those are ordinary editorial citations and they are noise here.
The difference shows up in the shape of the data.
Follow status. A paid placement is almost always dofollow — it is what was bought. A sponsored or nofollow attribute is the publisher telling you directly that this was not an editorial link, and for our purpose it is out.
How many links. A business that appears once or twice on a site looks like a placement. A domain appearing eighty times is a template link, a footer, a CDN, or a platform the site uses. Volume is the clearest sign that something is infrastructure rather than a customer.
Domain rating. A very strong domain being cited is a citation. Nobody pays for a link to a household name, and household names do not buy guest posts on small blogs. The obscure ones are the interesting ones — this is the opposite of the instinct you built up doing outreach.
The anchor. A company's own brand name as the anchor is the shape of a paid placement. So is commercial phrasing. An anchor that is the title of an article is a citation.
Why one publisher is not enough
Everything above is inference. Any one of those signals can be wrong about any one link.
The signal that is hard to be wrong about is repetition across unrelated publishers. One tech blog linking to a small SaaS company could be genuine editorial interest. Three unrelated blogs, in different niches, all linking to the same small company, is a company buying placements. There is no innocent explanation that produces that pattern.
This is why a single export gives you candidates and two or three give you a list. It is also why the count has to be worked out across everything you have imported rather than stored with one import — the moment you add another publisher, every previous judgement should change.
The mistakes worth naming
Counting links across publishers as if they were one publisher's. Three links each from thirty different sites is ninety links, and ninety links from one site would mean a footer. Add them up and you throw away exactly the businesses you were looking for. The volume test only means anything per-publisher.
Trusting an export whose filters you have forgotten. If you exported with "dofollow only" and "one link per domain" already applied, everything left will look like a placement, because you removed everything that did not. Write down what filters were on.
Treating affiliate destinations as buyers. A review site links to VPN and hosting brands because it earns commission, not because they paid for a placement. Those brands do buy links, but not from that site, and not through that relationship.
What to do with the list
Sort by how many publishers each business appeared on. Start at the top. Those are companies actively spending money on exactly the thing you sell, and you know it because you watched them do it.