Web Mentions vs. Backlinks: What Correlates More Strongly With AI Visibility?

The headline number: 0.664 vs. 0.218

Ahrefs' original 75,000-brand study compared a range of conventional search and brand metrics with brand visibility inside Google's AI Overviews.

The strongest correlation in the study was not backlink count, referring domains, Domain Rating, traffic, or paid search activity. It was branded web mentions.

Domain Rating correlated at 0.326. Branded anchors—the visible link text that contains the brand name—correlated at 0.527. Branded search volume correlated at 0.392.

That creates an interesting hierarchy. Pure link volume sat toward the bottom. Brand references across the web sat at the top. Branded anchors, which contain both a mention and a link, sat between them.

What Ahrefs meant by a web mention

This distinction matters because a web mention is broader than a backlink.

Ahrefs defined branded web mentions as occurrences of the brand name across the web. Those mentions could be linked or unlinked. A journalist naming the company without linking to it can count as a mention. So can a comparison article, review, directory, Best Of list, industry guide, forum discussion, or other indexed page that explicitly references the brand.

A backlink requires something narrower: one page links to another.

That means the two metrics answer different questions:

Why might mentions matter more to AI visibility?

The Ahrefs study does not prove the mechanism. But there is a reasonable strategic interpretation.

A link is fundamentally a relationship between documents. A mention can carry a much richer linguistic relationship around an entity.

Consider the difference between these examples:

• “Visit Acme here.”
• “Acme is an enterprise payroll platform.”
• “Acme is commonly evaluated alongside Gusto and Rippling by mid-market companies that need multi-state payroll.”

The first may contain a link while conveying almost no market context. The third can connect the brand to a category, competitors, audience, and use case even if it contains no link at all.

That distinction is especially relevant to AI systems built to synthesize language and entities. The words surrounding the brand can help describe what the company is, where it fits, and the situations in which it may be relevant.

The later Ahrefs study found the pattern held across platforms

Ahrefs later repeated and expanded the research across ChatGPT, Google AI Mode, and AI Overviews. Branded web mentions continued to correlate strongly with AI visibility, roughly between 0.66 and 0.71 depending on the platform.

Link metrics remained weak across the three systems. Ahrefs summarized the finding by noting that simply building large quantities of links was not enough and that brands appearing across a broader scope of sites showed the stronger pattern.

The later study also found YouTube mentions correlated even more strongly—approximately 0.737 across the platforms studied. That strengthens the broader interpretation: AI visibility appears associated with brand presence across many kinds of public web content, not merely the link graph pointing to the company's domain.

Does this mean backlinks are dead?

No. That conclusion would go well beyond what the research supports.

Google's current guidance for generative AI Search says conventional SEO best practices remain relevant because AI Overviews and AI Mode rely on Google's core Search ranking and quality systems to retrieve web pages. A page still needs to be crawlable, indexable, relevant, useful, and competitive enough to become part of that retrieval environment.

Google's Search Essentials likewise continues to emphasize crawlable links as part of how Google discovers and understands sites.

Backlinks can still support:

• traditional organic rankings;
• page and domain discovery;
• referral traffic;
• publisher relationships;
• authority for content that may itself become an AI source; and
• branded anchors, which Ahrefs found correlated more strongly with AI visibility than raw backlink volume.

The smarter interpretation is that links are now one part of a larger off-site authority model.

Branded anchors may be the bridge between links and mentions

One of the most interesting results in the original study was the 0.527 correlation for branded anchors.

A branded anchor is not merely a hyperlink. It explicitly names the entity inside the hyperlink—for example, linking the words “LSEO” rather than “click here.”

That matters because a branded anchor combines two kinds of information:

1. the publisher chose to create a link relationship with the brand; and
2. the publisher explicitly named the brand in the surrounding content.

Ahrefs' later study found branded anchors were particularly correlated with visibility in AI Mode, reaching 0.628 in that platform's dataset.

Again, that does not establish a direct ranking factor. It does suggest marketers should think less about raw backlink accumulation and more about earning legitimate editorial references in which the brand is clearly identified and contextually relevant.

The quartile data makes the mention gap hard to ignore

Ahrefs also grouped brands by how frequently they were mentioned across the web. Brands in the top quartile for web mentions averaged 169 AI Overview mentions. Brands in the next quartile averaged 14. The lower two quartiles averaged roughly zero to three.

Those numbers should not be interpreted as a formula. Earning enough mentions to move into a particular quartile does not mean a brand will suddenly receive 169 AI Overview appearances. Larger, more established brands often have stronger search demand, PR coverage, customer bases, advertising footprints, links, and market awareness at the same time.

The useful insight is relative: brands that are barely discussed across the wider web were also rarely visible in the AI Overview dataset.

Correlation is not causation—and the sample matters

Ahrefs used Spearman correlations, which measure whether two variables tend to move together. They do not prove that one causes the other.

The study also used specific filters to identify legitimate brands at scale. Domains needed a Domain Rating above 40, and Ahrefs used the highest-volume associated keyword with at least 800 monthly searches to identify brands. Roughly 26% of the original sample had no AI Overview mentions at all.

Those choices make the research useful, but they also mean marketers should not treat the coefficients as universal benchmarks for every startup, local company, B2B niche, or market.

A stronger reading is:

Among the brands Ahrefs studied, broader branded web presence was much more closely associated with AI visibility than raw link count.

Do not replace link spam with mention spam

There is an obvious bad response to this research: stop buying links and start buying huge volumes of low-quality unlinked mentions.

Google explicitly warns against that interpretation. Its generative Search guidance says seeking inauthentic mentions across the web is not as useful as marketers may assume because Google's core ranking and spam systems still focus on quality.

The same principle applies outside Google. Fifty irrelevant mentions created on weak websites do not necessarily create the same market evidence as five credible articles that accurately connect the brand to the right category and buyer context.

Quantity is easy to measure. Context is harder—but context is where much of the strategic value lives.

What should marketers optimize for instead?

The research suggests a broader off-site portfolio.

Link building is becoming brand building

For years, SEO teams could describe off-site authority primarily through links. The Ahrefs research suggests that model now needs to expand.

A strong digital PR story may earn a followed backlink, an unlinked brand mention, executive visibility, branded search demand, social discussion, and future inclusion in recommendation content. A high-quality comparison article can create a link while also connecting the brand to competitors and buyer criteria. A credible Best Of placement can introduce the brand into a shortlist even when the commercial link is appropriately qualified.

The best campaigns increasingly produce several kinds of value at once.

That does not make SEO less important. It makes the boundary between SEO, digital PR, brand marketing, and GEO less clean than it once was.

What this means for a modern GEO strategy

If AI systems are increasingly part of how buyers research companies, marketers need to ask a broader question than “How many backlinks do we have?”

Ask instead:

• How frequently is the brand discussed across the web?
• Which categories and problems are associated with those mentions?
• Which publications and communities mention competitors but not us?
• Where are buyers being shown third-party recommendations?
• Are we present in Best Of, comparison, alternatives, review, and buyer-guide content?
• Are our mentions distributed across credible sources or concentrated in a small network?
• Are those changes followed by stronger AI visibility and business outcomes?

This is where Mention Engine fits into the broader strategy. It gives brands a systematic way to build third-party recommendation coverage around specific buyer questions and content types instead of treating every off-site opportunity as a generic backlink.

The Ahrefs data does not give marketers a magic mention quota. It gives us something more useful: evidence that the web's conversation about a brand may matter at least as much as—and in this dataset considerably more than—the raw number of links pointing to its website.