AuthoritySource.aiAuthoritySource.ai
Back to journal

AuthoritySource.ai journal

Web Mentions Are the New Backlink — What the Data Says

Ahrefs analysed 75,000 brands and found web mentions correlate with AI visibility three times more strongly than backlinks. Here is what the data means for GEO and SEO buyers.

Web Mentions Are the New Backlink — What the Data Says

For two decades, "backlinks" has been the dominant vocabulary in search marketing. The link graph was the signal that separated authoritative pages from invisible ones.

That is changing — not because links stopped mattering, but because a new layer of search has appeared above them. AI assistants answer questions directly, and the evidence suggests they are reading the web differently from classic crawlers.

The short version: web mentions are now a stronger predictor of AI visibility than backlinks are.

What the studies found

1. Google AI Overviews: mentions 0.66, backlinks 0.22

In an Ahrefs study of 75,000 brands appearing in Google AI Overviews, the correlation between web mentions and AI visibility was 0.664. The correlation between backlinks and AI visibility was 0.218.

That is roughly a 3x stronger relationship for mentions over links inside Google's AI-generated answers.

Source: Ahrefs AI Overview brand correlation study

2. Cross-platform: 0.66–0.71 for mentions

A follow-up analysis across ChatGPT, Google AI Mode and Google AI Overviews put the correlation between web mentions and visibility at 0.66–0.71, consistently above the backlink signal.

The pattern is not limited to one model or one Google surface. Wherever AI answers are being generated, the entity-mention signal is showing up.

Source: Ahrefs AI brand visibility correlations

What 0.66 vs 0.22 actually means

Correlation is not a score out of one. It is a measure of how reliably two things move together.

  • 0.66 means that if you know how often a brand is mentioned across the web, you can predict its AI visibility fairly well.
  • 0.22 means backlink count still tells you something, but far less.

For a marketer, the practical takeaway is: getting your brand named inside relevant editorial content is now a higher-leverage activity than chasing raw link volume.

This does not mean backlinks are worthless. It means the reason they worked — trusted sources talking about a brand — is being measured more directly by AI systems.

Why AI engines weight mentions over links

Classic search engines used links as a proxy for reputation. A link was a vote, and votes could be counted.

AI answer engines do something closer to entity association:

  • They read the text around a brand name, not just the anchor tag.
  • They associate brands with topics based on co-occurrence across trusted sources.
  • They are trained on content where a brand is discussed as an authority, whether or not the discussion includes a hyperlink.

A linked mention still carries both signals — entity association plus a navigable citation. But an unlinked mention on a topically relevant, trusted site is no longer invisible to the system. That is why brand mentions have become a distinct placement type.

Mapping this to the two placement types

AuthoritySource.ai offers two placement products, and the data explains why both exist:

Brand mentions (GEO) — an editorial reference to your brand, usually without a link. This is the pure mentions signal the studies highlight. It targets AI answer engines first.

Link placements (SEO) — a linked editorial reference. This carries the mention signal plus classic Google ranking authority. It targets both AI answers and blue-link search.

The research suggests the brand-mention component is doing more of the AI work than the link itself. The link still matters for Google rankings, referral traffic, and click-through — but the entity mention is what teaches the models to name you.

Important caveats

Correlation is not causation. A brand that is frequently mentioned is usually also well known, well funded, and active in PR. Mentions may simply track fame, not cause visibility.

Topic fit still matters. A mention on a site about car insurance will not help a SaaS analytics brand, no matter how authoritative the domain.

Links are not dead. The 0.22 correlation is still positive, and links drive direct traffic, rankings, and trust signals that mentions do not replace. The claim is that mentions have become the primary AI-visibility signal, not the only signal.

What to do with this

If you are buying placements:

  1. Prioritise sites AI assistants already cite. A mention there is more likely to be pulled into future answers.
  2. Treat brand mentions as a deliberate product, not a consolation prize when you cannot get a link.
  3. Keep link placements in the mix — they still move Google rankings and send referral traffic.

If you are a publisher:

  1. Your existing unlinked brand mentions are now more valuable than many link sellers admit.
  2. Editorial context matters more than anchor text. The sentence around the brand is the asset.
  3. A site that AI models already quote can charge a premium for both mentions and links.

The bottom line

Backlinks built the old search layer. Web mentions are building the new one.

The data is early, the models will change, and correlation is not a strategy. But the direction is clear: AI answer engines are learning which brands to name by reading where they are already being named. Placements that create those mentions are no longer a side effect of link building. They are the point.

More from the journalGet started