In the largest study available on what correlates with being mentioned by AI platforms, branded web mentions scored between 0.66 and 0.71. Backlinks scored, in the researchers’ own words, very weak. So yes, the claim going around is directionally right: mentions appear to matter more than links for AI citations.

Two things nobody repeating that claim tells you. The study did not actually separate linked mentions from unlinked ones, so the specific advice to chase unlinked mentions is not what it measured. And the factor that beat branded web mentions was YouTube.

Here is what the data supports, what it does not, and what to do about it as of October 2026.

What did the study actually find?

Ahrefs published a correlation study in December 2025 covering 75,000 brands, measuring how often each appeared across millions of AI responses in ChatGPT, Google AI Mode and AI Overviews. They used Spearman rank correlation.

So the hierarchy is clear enough: being talked about correlates with being cited, and the raw link count barely registers.

What is the finding everyone is skipping?

YouTube mentions beat every other factor measured, on every platform tested.

This has barely been written about, and it is the most actionable thing in the study. The researchers counted a YouTube mention when the brand appeared in a video title, not in the transcript, which makes it a narrower and more achievable target than it sounds. You do not need a viral channel. You need your business name in the titles of videos that exist.

The likely mechanism is unglamorous. YouTube transcripts and metadata are in the training data of several major models, and YouTube is among ChatGPT’s most-cited domains. A video titled with your company name is a durable, machine-readable assertion that your company exists and does a particular thing.

For a service business, that means your own channel with plainly-titled videos, yes, but it also means being named in other people’s video titles. A podcast episode published to YouTube with your business in the title does more measurable work than a guest post.

What does the study not show?

Four limits, and they matter more than the headline numbers.

It does not separate linked from unlinked mentions. The metric counts web pages that mention the brand. Whether those pages also linked is not reported. So the popular version of this advice, which is that unlinked mentions are now as good as links, is not tested by the study everyone cites for it. The honest statement is that mentions correlate with citations whether or not they carry a link, because nobody measured the difference.

It is correlation, and the authors say so plainly. Their own words: correlation isn’t causation, and improving these metrics will not automatically boost your AI visibility. They go further in the conclusion, allowing that these factors may simply be markers of existing brand strength rather than causes of citation. That is the honest reading. Brands that get mentioned a lot are brands that are already notable, and notability is what the AI is detecting.

The sample is established companies. Brands were selected by filtering for domains with Domain Rating above 40 whose top keyword had at least 800 monthly searches. That is not a small US service business. Whether these relationships hold for a twelve-person clinic is unknown.

And there are no significance tests. No p-values, no confidence intervals, no per-factor sample sizes.

Why do Ahrefs’ own two studies disagree?

This is worth knowing because it undercuts the confident version of the advice.

Ahrefs ran a smaller version of this analysis in July 2025, looking at the top 50 most-mentioned domains per platform across about 76.7 million AI Overviews and roughly 950,000 prompts each on ChatGPT and Perplexity. For branded web mentions against AI visibility it found Google AI Overviews at 0.65, Perplexity at 0.30, and ChatGPT at 0.15.

Compare that ChatGPT figure, 0.15, with the 0.664 from the 75,000-brand study. Same relationship, same platform, wildly different answers.

The samples explain it. The second study looked only at the fifty biggest brands, where everyone has enormous mention volume and mention count stops discriminating between them. The larger study included the full spread.

Read together, the pattern suggests mentions help you get onto the radar and stop helping once you are already famous. That is an inference from comparing the two datasets rather than a finding either study states, but it is the reading that reconciles them, and it is good news for anyone who is not already famous.

What does this mean if you are not a big brand?

Three practical consequences.

Mention volume is probably a real lever for you specifically, more than it is for the household names. You are in the part of the distribution where going from barely mentioned to consistently mentioned is a meaningful change.

ChatGPT showed the weakest correlations with every traditional authority signal of the three platforms tested. The study’s authors suggest it may therefore be the better entry point for brands without established search authority. That is their interpretation rather than a measured finding, but it is a reasonable one, and it means a smaller business should not assume the AI platforms are uniformly closed to it.

One genuinely odd result: ChatGPT’s visibility correlated more closely with advertising metrics, ad traffic at 0.286 and ad cost at 0.273, than any other platform’s did. The authors found that surprising and were careful to say it does not mean ChatGPT rewards ad spend, only that heavy advertisers tend to dominate the content ChatGPT draws from. Worth knowing before someone sells you AI visibility through paid media.

What should you actually do?

Five things, ordered by evidence behind them.

Get your business name into video titles. Your own channel, customer explainers, podcast appearances published to YouTube. This is the highest-correlating factor in the best available study and the least contested space to compete in.

Pursue mentions in places AI systems demonstrably read. Wikipedia and Reddit are the two most-cited domains in ChatGPT’s responses by a wide margin. You cannot write your own Wikipedia page, but you can be a genuinely useful participant in the subreddits where your industry is discussed, and you can earn the kind of third-party coverage that eventually supports a Wikipedia citation.

Chase named mentions, not just links. Branded anchors correlated noticeably better than raw backlinks, which suggests the signal is someone deliberately naming you. A press mention that spells out your company name in the text is worth more than a link buried in a resource list.

Do real digital PR. Industry publications, podcast guesting, data you can give journalists, expert commentary through services like Qwoted. The point is not the link. The point is a durable page on a credible domain that says what your company is.

Be the clearest source about yourself. Everything above puts your name into the corpus. Your own site has to be the thing that explains, in plain machine-readable text, what you do and who for, so the model has something accurate to draw on.

The honest summary

Mentions beat links in the only large study on the question, and that is a real finding worth acting on. It is correlational, it was measured on established brands, and the specific claim about unlinked mentions has never been isolated and tested.

The strongest single result, YouTube mentions in video titles, is the one almost nobody is working on. If you want one thing from this article, take that one.

Frequently asked questions

Do brand mentions matter more than backlinks for AI search?

In the largest study available, yes. Ahrefs’ 75,000-brand analysis found branded web mentions correlating 0.656 to 0.709 with AI visibility, while number of backlinks and URL rating were described as very weak with no coefficient published. The authors stress these are correlations, not proven causes.

Do unlinked brand mentions help AI citations?

Probably, but it has not been directly tested. The study everyone cites for this counted web pages mentioning a brand without reporting whether those pages also linked. So mentions correlate with citations regardless of links — which is not the same as showing unlinked mentions work on their own.

What is the strongest factor for getting mentioned by AI platforms?

YouTube mentions, at roughly 0.737, ahead of branded web mentions. The study counted a brand appearing in a video title rather than a transcript, which makes it achievable without a large channel. It is the least-discussed finding in the research and the easiest to act on.

Can a small business get cited by ChatGPT?

Nothing in the research says otherwise, and ChatGPT showed the weakest correlations with traditional authority signals of the three platforms tested, which its researchers suggest makes it a better entry point for less-established brands. Note that the study sampled companies with Domain Rating above 40, so it does not directly measure small-business outcomes.

Want help working out where your mentions should come from?

A strategy call covers which platforms currently mention you, where your competitors are being talked about that you are not, and which three or four placements would actually move the needle for a business your size.

Book a strategy call. We will tell you if digital PR is the wrong priority for you right now, because sometimes it is.