Unlinked mentions SEO data: Ahrefs' 75,000-brand study found mentions beat backlinks 3x for AI citations. How to earn them without a link-building budget.
Backlinks are what Google's crawler follows to find and rank a page. Mentions are what a language model reads to decide who to trust when it answers a question. Those are two different jobs, and for years SEO teams have optimized almost entirely for the first one. Ahrefs' study of 75,000 brands is the first hard data showing the gap between them, and it's the case for treating unlinked mentions SEO as its own discipline: unlinked brand mentions correlate with AI visibility roughly three times more strongly than backlinks do (Ahrefs).
That means the digital PR pickup with no link, the podcast guest slot, the "best tools for X" roundup that names you without a hyperlink, the tactics most teams file under "nice to have but no SEO value," are quietly one of the highest-leverage plays available for getting cited by ChatGPT, AI Mode, and AI Overviews. This isn't a rebrand of link building. It's a distinct discipline with its own data, and this post covers what that data actually says and how to act on it.
This sits next to what we've already covered on G2 and Capterra reviews, where ChatGPT mentions a brand about 3.2x more often than it links or cites it, and 44% of prompts return zero brand mentions at all, per BrightEdge's analysis of ChatGPT citation triggers. That earlier post looked at the mention-versus-citation gap on one channel. This one looks at what predicts the mention happening at all, and the Ahrefs data says it isn't your backlink profile.
A backlink and a mention look similar on a page, but they're read by two different kinds of system, built for two different jobs, and that difference is the whole reason the correlation data below looks the way it does.
A crawler follows a backlink to discover a URL and treats the link as a vote of confidence from the linking domain, weighted by that domain's own authority. It's a structural signal: the link exists, it points somewhere, and PageRank-style math propagates trust through the graph it forms. The crawler doesn't need to understand what the linking page said about you. It just needs the edge in the graph.
A language model works differently. When it's deciding whether to name your brand in an answer, it's drawing on two things: what it learned about you during training, and what a live retrieval pass turns up. Both of those are built from text, not from a link graph. A sentence that says "Ahrefs found unlinked mentions predict AI citation" teaches the model an association between your name and a topic whether or not that sentence happens to carry a hyperlink. The link graph rewards concentration of authority. The mention signal rewards being talked about, in plain text, in a lot of different places. Ahrefs' own analysis of the data found volume alone doesn't move the needle either: neither link volume nor content volume correlated meaningfully with AI visibility, which is why Ryan Law, Ahrefs' Director of Content, summed up the finding as, "It's not just a content creation arms race."
This is a different mechanism from why Wikipedia dominates ChatGPT's citations: that's about training-data weight for one specific, heavily-cited reference domain. This is about your own brand's name showing up, unlinked, across the ordinary web the model reads everywhere else.
If mentions and links were doing the same job, a linked mention should out-correlate an unlinked one. Ahrefs' own numbers say the opposite. Branded anchors, the clickable, hyperlinked form of a mention, correlated with AI visibility at 0.511 on ChatGPT, 0.628 on AI Mode, and 0.527 on AI Overviews. General branded web mentions, most of them unlinked, correlated higher across all three: 0.664, 0.709, and 0.656 (Ahrefs). Adding a hyperlink to a mention didn't strengthen its relationship with AI visibility. It weakened it, on every surface Ahrefs measured.
That's the closest thing in the study to direct proof that the unlinked portion of a mention is doing real, independent work, not just riding along with whatever a link happened to add. A roundup post that names you in a sentence with no link is not a weaker version of a roundup post that names you with one. By this data, it may be a stronger one.
Read the numbers in one place before the breakdown: across 75,000 brands, branded web mentions correlated with AI visibility at 0.656 to 0.709 depending on the platform, while number of backlinks, the most basic link metric, correlated at roughly 0.218 for AI Overview visibility specifically, about a threefold gap (Ahrefs). That gap is the headline. The methodology and the platform-by-platform detail are what make it trustworthy.
Ahrefs pulled 75,000 brands with a Domain Rating above 40 and at least one top keyword drawing 800 or more monthly searches, a filter meant to exclude brand-new or tiny sites where the data would be noisy. For each brand, it measured a Spearman correlation between a set of on-site and off-site metrics and the brand's visibility across ChatGPT, Google's AI Mode, and AI Overviews, using Brand Radar's tracking of AI responses (Ahrefs). Spearman correlation measures how consistently two variables move together in rank order, not whether one causes the other, which matters here: this is a correlational study, not a controlled experiment, and Ahrefs is explicit that it's reporting relationships, not proof that a mention directly causes a citation.
The gap holds on every platform Ahrefs tested, though the exact size shifts. Branded web mentions correlated with AI visibility at 0.664 on ChatGPT, 0.709 on AI Mode, and 0.656 on AI Overviews. For AI Overview visibility specifically, Ahrefs' companion breakdown of the same 75,000-brand dataset puts number of backlinks at just 0.218, with referring domains a little higher at 0.295 and URL Rating lower still at 0.18 (Ahrefs), all of them far below the mention numbers above. AI Mode shows the strongest mention correlation of the three surfaces and also the highest (still weak) backlink correlation, around 0.2, which tracks with AI Mode sitting closest to classic Google Search of the three engines Ahrefs measured. One channel stood out further still: YouTube mentions correlated with AI visibility at roughly 0.737, the single strongest individual signal Ahrefs measured, ahead of general web mentions on every platform.
Ahrefs didn't hedge this finding. Its own post states there are "very weak correlations between link metrics ('number of backlinks' and 'URL rating') and brand mentions across all AI systems" it tested. That's a direct statement from the company that built its entire product category on link metrics, saying those same metrics barely predict the thing AI answer engines actually reward. It doesn't mean backlinks stopped mattering. Google's crawler still needs links to find and rank your pages, and referring domains still correlate with classic organic visibility. It means backlinks and AI citation are now two separate outcomes with two different sets of predictors, and treating them as the same problem with the same fix is where most teams' AEO effort quietly goes to waste.
None of this requires a link-building retainer. It requires being the source a journalist, podcast host, or roundup writer reaches for, which is a content and outreach problem, not a link-negotiation one.
The fastest way to earn an unlinked mention is to give a writer something they can't get anywhere else: a specific, sourced number or a direct quote they can attribute to a named person at your company. Pitch journalists and industry newsletters the same way you'd pitch a source for a comparison post, with the number up front, not buried in a press release. A quote from your founder or head of product, tied to a dated, verifiable stat, is exactly the kind of material a writer keeps in their notes for the next three articles on the topic, whether or not the first one links back to you. Original research earns an 82% AI citation rate, the highest of any content type measured, per NP Digital's 2026 survey of 500 marketers and business owners: the number is the pitch, and the mention follows the number.
Podcast transcripts and roundup posts are two of the most common unlinked-mention surfaces, and neither requires a link to work. A guest slot on a podcast in your category gets your name and your product spoken aloud, transcribed, indexed, and read by a model that treats the transcript as ordinary web text. A "best tools for X" roundup that names you in a sentence, with or without a link, does the same job. Community discussion works the same way: Reddit shows up in roughly 21% of Google AI Overviews, and a thread where your product gets named unprompted by someone who isn't you is a mention no outreach campaign can fully manufacture, but one you can earn by being worth recommending when the question comes up.
The most durable way to generate unlinked mentions at scale is to be the source other people cite when they write about your category, which means publishing a number nobody else has. A churn breakdown, an adoption-rate study, a survey of your own customer base, reported with a real sample size and date, is the kind of material that gets referenced in someone else's roundup, podcast, or newsletter for months after you publish it, sometimes with a link and often without one. Our guide to turning first-party data into a citable post covers the process: pick a number your product already tracks, write a methodology block, and lead with the finding instead of burying it.
Yes, ask for the link every time, but don't make the ask a condition of the pitch. Nothing in Ahrefs' data says a backlink hurts you, and a link still does real work outside AI citation: it sends referral traffic, it feeds classic Google ranking through the mechanism backlinks have always used, and it costs you nothing extra to request once a writer has already agreed to mention you. What the data changes is your response when the answer is no. A journalist who won't link but will name you, a podcast host who mentions your product on air with no show-notes link, a roundup that lists you without a hyperlink: none of that used to register as a result worth pursuing under a link-first playbook. Under the correlation data above, it's the stronger signal of the two. Treat the link as a bonus you always ask for, not a gate that decides whether the mention was worth earning.
A mention campaign you can't measure is a campaign you're guessing about. Run the same manual prompt-testing discipline you'd use for any AEO work: build a list of the buyer-style questions your prospects actually ask, run them weekly across ChatGPT, Perplexity, Gemini, Claude, and Copilot, and log whether your brand gets named, whether a competitor gets named instead, and whether the mention carries a citation link or not. Pair that with a Google alert or a media-monitoring tool on your brand name so you catch unlinked mentions as they publish, not months later. Our AI citation tracking guide covers the free GA4 setup for the traffic layer this does produce, plus the proxy signals (branded-search lift, Direct-traffic trends, a "how did you hear about us" field) worth watching for the no-click mentions that never show up as a session at all.
The honest limitation is the same one every correlation study carries: Ahrefs measured a relationship, not a lever you can pull and watch move in isolation. A mention doesn't guarantee a citation any more than a backlink guarantees a top-10 ranking. What the data supports is a reallocation, not a guarantee: if your team has been spending a digital PR budget on link placements and treating unlinked pickups as a consolation prize, the 75,000-brand correlation gap is the argument for reversing that priority.
Earning unlinked mentions takes quotable, data-backed content, the same fact-checked writing Lyra ships as a pull request you review and merge.
FAQ
A linked mention is your brand name wrapped in a clickable URL back to your site, the kind link builders chase. An unlinked mention is your brand name in plain text, no hyperlink attached, the kind a podcast transcript, a roundup post, or a Reddit thread produces constantly. Ahrefs' 75,000-brand study found the unlinked kind correlates more strongly with AI visibility than the linked kind does, which flips decades of SEO priority.
The strongest evidence is for AI citations specifically. Ahrefs measured branded web mentions against AI visibility on ChatGPT, AI Mode, and AI Overviews and found correlations of 0.656 to 0.709, versus 0.218 for number of backlinks, the classic link metric. Google's own crawler still relies on links to discover and rank pages, so unlinked mentions aren't a backlink replacement for organic ranking. They're a separate, and currently stronger, signal for what a language model decides to cite.
About three times more, per Ahrefs. Branded web mentions correlated with AI visibility at 0.664 on ChatGPT, 0.709 on AI Mode, and 0.656 on AI Overviews, all Spearman correlations across the 75,000 brands in the study. Number of backlinks correlated at roughly 0.218, and Ahrefs itself described the link metrics it tested as having 'very weak correlations' with brand mentions across every AI system it measured.
Yes, ask for it, but don't let the answer decide whether you pitch the story. A link still helps referral traffic, domain authority, and Google's classic ranking, and it costs nothing to request. The data just says don't skip the pitch when a writer says no to the link, because Ahrefs found the unlinked mention itself, not the link riding along with it, is doing most of the correlation with AI visibility.
Built by the tool you're reading about
Lyra finds the topics worth ranking for, writes them in your repo's voice, fact-checks every claim, and opens a pull request scored and ready to merge. You review and hit merge. Want to see what she'd write for you? Start free with three posts, no card.
Keep reading
ChatGPT citation drop 2026, explained: the March-April volatility, why ads pushed citations to product sites, what rebounded by May, and what to do now.
Cloudflare's content signals policy gets enforcement on September 15, 2026, blocking Training and Agent crawlers by default on ad-monetized pages.
Conversational discovery ads put Sponsored results inside Google AI Mode answers. What they are, why now, and what they mean for your organic click share.