There is a difference between a link and a reputation, and for most of the search era, that difference did not really matter.
A link was a vote. The more votes, the higher the ranking. The system was crude but legible. You could buy links, count them, score them, and report on them. The whole industry of guest posting grew up around that arithmetic, and brand marketers learned to treat link acquisition as a parallel discipline to actual brand-building, with separate budgets, separate vendors, and separate KPIs.
That separation is starting to collapse.
The systems now mediating how brands get discovered — Google’s AI Overviews, ChatGPT, Perplexity, Gemini — are not reading the web the way the previous generation of search engines did. They are reading it more like a person reads a magazine. Who keeps showing up? Who keeps getting mentioned? Whose name appears alongside the topic, in publications that look like real publications, often enough that the recommendation feels safe to make?
A link can still help with that. A reputation does more.
What the data is now saying out loud
This is no longer a hypothesis. Ahrefs studied 75,000 brands and found that unlinked brand web mentions correlate with AI Overview visibility at 0.664, while backlinks correlate at just 0.218 — roughly a 3× gap in favor of mentions. Ahrefs
That number deserves a moment. For two decades, link counts were the closest thing search marketing had to a universal currency. They are still useful. They are no longer the dominant signal. AI search engines don’t evaluate authority solely through link graphs; they pattern-match on co-occurrence — how frequently and contextually a brand name appears alongside relevant topics, trusted sources, and authoritative domains. Digital Journal
Translated for a brand marketer: the systems recommending products and services to your customers are reading the web for reputation, not just for votes. And reputation, as it turns out, looks a lot like what an editor would have recognized as reputation thirty years ago. Consistent presence in places that take editing seriously. A name that turns up in real conversations. A footprint that reads as authentic because it is.
What this changes about how brands buy placements
The implication is uncomfortable for any program that has been optimizing exclusively for link count.
A campaign that produces 200 placements on thin, anything-goes sites generates 200 links and very little reputation. The links register. The mentions, in the AI-search sense, do not — because the surrounding context doesn’t read as a credible source, and the systems doing the reading have learned to discount mentions from environments that don’t look like real publications.
The reverse is also now true, and this is the part more brands are starting to notice. A smaller number of placements in publications that have an actual editorial identity, an actual audience, and an actual history can outperform much larger volume programs in the metrics that increasingly matter for AI discovery. SearchAtlas research found that 60% of new Knowledge Panels were triggered by unlinked brand mentions from trusted sources. theStacc
The link still matters. The link in a credible context matters more. The mention in a credible context, even unlinked, is now doing measurable work it was not doing five years ago.
What “credible context” actually requires
There is no metric that fully captures credible context, but there is a working description.
A credible context is a publication that has been on the same beat long enough for the surrounding content to reinforce rather than dilute the placement. It has a recognizable editorial voice. It has an audience that returns. It has, to use a phrase that has quietly come back into circulation, a point of view. The signals that AI systems are now trained to detect — coherent topic profile, consistent identity, signs of real human curation — are precisely the signals that editors used to produce as a matter of course, before that part of the publishing economy got hollowed out.
The Good Men Project is one of the publications that has kept that layer intact. Founded as a book in 2009 and launched as a media platform shortly after, GMP has spent fifteen years on a defined beat: modern masculinity, men’s mental health, relationships, fatherhood, identity, wellness, and social change. The point is not the topic list. The point is the continuity. The beat has not drifted. The archive reinforces itself. A brand mention inside that environment lands in a context the new generation of systems can actually read.
The shift from footprint to signal
For a long time, brand placement strategy lived in the language of footprint. How many sites, how many links, how wide the spread. Footprint was the metric. Footprint was the deliverable.
The new question is closer to signal. Not how many places does the brand appear, but how legibly does it appear in places that count. The marginal dollar of SEO budget now buys more AI visibility when it is spent on earned conversations than on link placements. The reframe is not subtle. It changes which placements get prioritized, which publishers get longer-term relationships, and which line items get cut when budgets compress. Soar Agency
For in-house brand marketers, the practical move is to treat a portion of the program — the portion responsible for reputation rather than raw coverage — as digital PR work that happens to take the form of paid editorial placement. The work belongs in publications that read like publications. The deliverable is not a link in a spreadsheet. It is a presence the next generation of discovery systems can detect.
What to do with this, practically
None of this requires abandoning what already works. Backlinks still pass direct authority. Backlinks still pass direct ranking authority in Google — brand mentions do not replace them for traditional SEO. The volume program can continue doing its job. Brandmentions
What changes is what sits on top. The authority layer — the placements doing the reputation work, the ones that will still be cited in three years, the ones that AI systems will read as legitimate signal — needs to live in environments that earn that read. That is a smaller list than the volume list. It deserves a different budget and a different vendor conversation.
For brands, authors, and organizations that need content to appear somewhere AI systems will read as a credible source, GMP offers a publishing home with history, readership, and editorial identity. Email [email protected].
—
Photo credit: iStock.com
