Why unlinked mentions now carry weight

Lesson 45 of 50Module 9 · The future of link building5 min read

By Bill Hartzer · Hartzer Consulting · 25+ years in search, domains and expert witness work

What an unlinked mention actually is

An unlinked mention is any place your brand, product, publication or personal name appears in someone else's content without an href pointing at you. A journalist quoting you and naming your company in the attribution line. A podcast host saying your name in an episode that gets transcribed. A forum thread where three people recommend your tool. A trade article listing you among five vendors. A research paper citing your data in the text without a hyperlink in the web version.

SEOs have historically treated these as failures — near-misses to be converted into real links by emailing the editor. Module 5 teaches you how to do that conversion, and it is still worth doing when the piece is genuinely important. But the framing is wrong. The mention is not a failed link. It is a different kind of object, and it does some things a link does not do.

Here is the test I use. Ask what the sentence tells a machine reading it. A bare link tells you that one document points at another, and gives you anchor text if you are lucky. A sentence like “Hartzer, who has worked in domain-name disputes for two decades, says the opposite” tells a reader and a machine who I am, what I am associated with, what I claim, and that a publication thought it worth printing. That is more information, not less.

Co-occurrence: what repetition builds

Co-occurrence is the plain fact of two things appearing near each other in text, repeatedly, across many documents written by people who did not coordinate. Search engines have used co-occurrence signals for a long time — it is one of the oldest ideas in information retrieval, and it underpins how systems learn that Nike and running shoes belong together without anyone declaring it.

The important property is that co-occurrence is cumulative and statistical, where a link is discrete. One link either exists or it does not. Association builds:

• One mention tells a system almost nothing. It is a coincidence.
• Twenty mentions across twenty independent publications, all placing your name near the same subject, is a pattern.
• Two hundred over three years, in sources that get retrieved, is the thing people loosely call “being known for something.”

You cannot buy that in a week, which is exactly why it is durable. It is also why one prestigious link, on its own, does not produce it. A single link from a national title is a strong ranking signal and a weak association signal. Forty mentions across the trade press is the reverse. A good program produces both, from the same work.

Independence matters more than volume. Two hundred mentions generated by one syndicated press release are one source repeated, and any system worth its salt will treat them that way. Twenty mentions from twenty newsrooms that each decided independently to name you is a far stronger pattern than two hundred copies of your own words.

What we know, what we infer, and what we do not know

Be careful here, because this is where the industry starts overclaiming.

Reasonably settled. Retrieval systems work over text. Text that names you in connection with a topic is retrievable and quotable. Publications with real readerships are more likely to be in whatever corpus a system draws on than a blog nobody reads. None of that is controversial.

A reasonable inference. Google holds patents describing the idea of counting implied links — references to a source that are not expressed as hyperlinks — alongside express links when scoring a resource. I have seen this cited as proof that mentions rank you. It is not proof. A patent describes what a company thought about at a point in time, not what shipped, and there are patents on plenty of things nobody ever used. Treat it as evidence that the idea is taken seriously by serious people, and no more.

Not known. Whether any specific system counts unlinked mentions as a ranking input, and if so how heavily, is not public. Whether being named more often makes a model more likely to name you in a generated answer is a plausible mechanism, but I have never seen a controlled study that isolates it, and I am not going to invent one for you. If someone shows you a percentage on this, ask what the control group was.

Which mentions are worth having

Not all mentions do work. The ones that do share properties, and they are the same properties that make a link good — which should tell you something about how consistent the underlying logic is.

• The source gets read. A mention in a publication with an audience is a mention in a document that gets crawled, cited, quoted onward and used as a source by other writers. A mention on a dead site is a string in a file.
• The context is topical. Your name near the subject you want to be known for. Being named in a list of “brands that sponsored a golf day” builds association with golf days.
• The naming is consistent. Lesson 9.3 is entirely about this. If half your coverage says Acme, a quarter says Acme Inc. and a quarter says The Acme Group, you are building three weak associations instead of one strong one.
• The sources are independent of each other. Different owners, different newsrooms, different writers.
• You are the subject, not scenery. A quote attributed to you by name outperforms an appearance in a supplier list.

And the ones that are worthless: scraped aggregators, AI-spun listicles that name forty vendors, low-quality directory pages, and anything produced by a service selling “brand mentions” at volume. Manufactured mentions fail for precisely the same reason manufactured links fail. Cheap to produce means easy to discount.

So do you still reclaim them?

Yes — selectively, and with the reason updated.

The old reason for unlinked mention reclamation was that the link was the point and you had been denied it. The better reason is that a link on a strong, relevant page is still a real ranking asset, and asking politely costs one email. Reclamation is one of the highest-yield outreach tactics on this site because the hard part — someone deciding you were worth writing about — already happened.

What changes is the decision rule when the answer is no. Under the old framing, an editor declining to add the link meant the coverage was a write-off. Under this one, it means you got association, corroboration and readers, and did not get a ranking signal. That is a partial win recorded as a partial win, not a zero.

Two practical cautions. First, do not chase reclamation on every mention — prioritize pages with real traffic and topical fit, and ignore the aggregator noise. Second, do not badger newsrooms. Many publications have a standing policy against adding outbound links after publication, some for good editorial reasons. Asking once is professional. Asking three times gets you remembered for the wrong reason, and in digital PR your reputation with a desk is worth more than any single link.