Backlinks vs Brand Mentions for AI Search: What the Data Actually Says (2026)

If you have read any SEO commentary in the last year, you have met the headline: "Brand mentions are the new backlinks." It usually comes with a number, most often "3x more important," and the conclusion that link building is dead. For a SaaS team deciding where next quarter's off-page budget goes, that is not an academic question. It is the choice between paying for link placements and paying for PR.

So let's look at the actual numbers: where they come from, what they measured, and what they do and don't prove. The short version is that the "3x" figure is real, it comes from a single study of a single search feature, and the honest reading of it is not "stop building links." It is more useful than that.

Key takeaways

• In Ahrefs' study of 75,000 brands, unlinked brand mentions correlated with Google AI Overview visibility at 0.664, versus 0.218 for raw backlink count, roughly a 3x gap. That single study is where the "3x" claim comes from.
• That figure is a correlation, not proof of causation, and it is specific to Google AI Overviews, not "all AI search." Anyone applying it to ChatGPT or Perplexity is guessing.
• Brand mentions and backlinks do different jobs. Mentions teach AI systems that your brand is a known name worth citing. Links carry authority, get pages discovered and indexed, and put you on the credible sites where mentions count.
• They are not rivals. They work as a funnel: relevant links get you onto trusted publications, and being on those publications is what creates the brand mentions AI rewards.
• For SaaS, the winning move is to earn both, in the right order: be a source worth citing, on credible and topically relevant sites. That is a tighter, more focused job than the old "get as many links as possible" approach.

The short answer

For Google AI Overviews specifically, the data suggests brand mentions are the stronger signal, about three times more strongly correlated with visibility than how many backlinks you have. But "stronger correlation in one study of one search feature" is a long way from "links are dead." Backlinks are still positively correlated with the same outcomes, they are still how Google finds and trusts pages, and they are how your brand gets mentioned on authoritative sites in the first place. So treat mentions and links as two parts of the same system, not as a choice you have to make. For the wider picture of how AI answers change the value of a link, see generative engine optimization for SaaS.

What the numbers actually say (and what people get wrong)

The whole "brand mentions beat backlinks" idea traces back to one piece of research: Ahrefs' analysis of AI Overview brand visibility factors, published May 26, 2025. It studied 75,000 brands and measured how strongly different signals correlate with a brand showing up in Google AI Overviews. Using Spearman rank correlation, here is what they found:

Signal	Correlation with AI Overview visibility
Branded web mentions	0.664
Branded anchors	0.527
Branded search volume	0.392
Domain Rating (DR)	0.326
Referring domains	0.295
Branded traffic	0.274
Number of backlinks	0.218
URL Rating (UR)	0.180

Divide 0.664 by 0.218 and you get roughly 3.05. That is the "3x." It is a real figure from a real study, and most articles get that part right.

Here is the nuance most summaries skip, and it changes what you should take away:

• It is a correlation, not causation. Ahrefs say so directly, and they call these relationships "moderate to very weak." Brands that get mentioned a lot also tend to be bigger, better known, and better funded, and all of those things drive AI visibility on their own. The mentions may be a symptom of authority as much as a cause of it.
• It is about Google AI Overviews only. It does not measure ChatGPT, Perplexity, Gemini, or Google's AI Mode. Treating one AI Overview result as a universal law of "AI search" is the most common mistake people make on this topic.
• Backlinks did not score zero. A 0.218 correlation is still positive. And referring domains (0.295) and Domain Rating (0.326) both scored higher than raw backlink count. The real lesson is the one good link buyers have always repeated: the quality and variety of your links matter more than how many you have.

Ahrefs also found that the brands with the most web mentions earn up to roughly 10x more AI mentions than the tier below them. So mentions clearly matter. The real question is how you get them, and that is where links come back in.

Backlinks vs brand mentions: what each one actually does

These two are not interchangeable. They sit at different points in how a brand becomes visible, and they tell both classic search and AI systems different things.

	Backlink	Brand mention
What it is	A hyperlink from another site pointing to yours	Any reference to your brand or product, linked or not (e.g. "tools like Acme")
Primary job	Passes authority, helps Google find and index pages, sends referral clicks	Tells search and AI systems that your brand is a known name in a topic
Classic SEO role	Direct, well-established ranking factor	Indirect "implied" trust signal; reinforces that your brand is a real entity
AI search role	Helps the source page get crawled, trusted, and retrieved	Strongly correlated with being named in AI answers; builds the entity the model recognizes
How you earn it	Outreach, digital PR, guest placements, niche edits, link insertions	Digital PR, podcasts, community presence, being genuinely notable in your category
Can it exist without the other?	A link is always also a mention	Yes, an unlinked mention is just text

The key difference is linked vs unlinked mentions. A link is, by definition, also a mention. An unlinked mention (your brand named in text with no hyperlink) passes none of the traditional "link equity," but it still feeds the systems that Google and large language models use to decide who is a credible name in a category. When people say "brand mentions are the new backlinks," what they really mean is that unlinked mentions now do real work that SEOs used to dismiss.

Why AI search reweights the two (the mechanism nobody explains)

Most articles claim that "LLMs trust brands" without explaining how. Here is the actual mechanism, because once you understand it, you know what to do.

An AI answer engine does not rank ten blue links. It writes an answer, and it builds that answer from two things: what it learned during training (a huge amount of web text) and, more and more, what it pulls in live when you ask the question. In both, your brand appears as text sitting next to a topic. The more often credible pages say "Acme is a leading SaaS analytics tool," the more strongly the model links "Acme" to "SaaS analytics," with or without a hyperlink in those sentences. That is why an unlinked mention on a trusted publication can shape an AI answer even though it passes no PageRank.

The clearest proof of how loosely AI citations relate to brand naming is Semrush's Ghost Citations study, published June 9, 2026. It looked at 3,981 domain appearances across 115 prompts, 14 countries, and four engines (ChatGPT, Google AI Overviews, Gemini, and Google AI Mode). It found that 61.7% of AI citations are "ghost citations": the page is used as a source, but the brand is never named in the answer. Only 38.3% of those source appearances actually named a brand. Being used as a source and being named in the answer are two separate things, so optimizing for only one of them costs you visibility.

One accuracy note, because you will see it everywhere: Google's "implied links" patent is not about unlinked mentions on web pages. The patent is about brand and URL references inside search queries, used as a ranking input. That is a different thing. People conflate the two constantly; Search Engine Journal's coverage explains what it actually covers. For a real official line on mentions, the better source is Google's Gary Illyes, who said at Brighton SEO in 2017 that Google looks at content "highly cited on the internet, and I'm not talking about just links, but also mentions on social networks and people talking about your branding."

But links are how mentions get made

This is the part the "links are dead" argument skips, and it matters most for where you actually spend.

Ask the practical question: how does a SaaS brand build up the credible, on-topic mentions the Ahrefs data rewards? You do not get mentioned on a respected industry publication by hoping. You get mentioned because you ran a digital PR campaign, published original research that journalists cited, earned a guest placement, or made it into a "best tools" roundup. Every one of those activities produces a mention, and usually a link at the same time. The link and the mention come from the same piece of outreach.

So the budget debate is mostly a false choice. The work that earns mentions (getting onto trusted sites) is the same work that earns links. When you place a link on a relevant, real-traffic publication, you are also putting a brand mention on a site that AI systems may pull from later. The "3x" data does not tell you to stop doing the thing that creates mentions. It tells you to put your placements on credible, relevant sites, because that is where mentions count. That is the same case for quality over quantity we make in why organic traffic beats DR when buying links.

It also helps to remember that backlinks still do jobs mentions cannot. They get new pages found and indexed, and a page that never gets indexed cannot rank or be pulled into any answer. They pass authority that lifts your whole domain. And they still correlate positively with rankings; for the full evidence, see do backlinks still matter for SaaS SEO. Google's Gary Illyes said in 2024 that Google now needs "very few links" to rank a page well, which is an argument for fewer, better links, not for none.

Why this is urgent now: AI answers are eating the click

This debate matters now because AI answers are no longer a small, side feature. Semrush's AI Overviews study (updated December 15, 2025, across 10M+ keywords) tracked the AI Overview trigger rate rising from 6.49% of searches in January 2025 to 24.61% by July, then settling around 15.69% in November as Google adjusted where it shows them. Either way, a large and growing share of searches now show an AI answer above the normal results.

And those answers take the click. Pew Research Center studied 68,879 real searches from 900 US adults in July 2025. When an AI summary appeared, people clicked a normal result just 8% of the time, versus 15% when there was no summary, and only 1% clicked a link inside the summary. When the answer is right there on the page, fewer people travel to yours.

That is what is at stake. If a buyer researches your category through an AI answer and your brand is not one of the names it knows, you are invisible at the moment they are deciding, no matter how well you rank in the normal results. Being cited is becoming its own goal, alongside ranking.

What a source AI actually trusts looks like

If mentions only count when they sit on credible sources, the next question is obvious: what makes a source credible enough that AI systems pull from it and trust it? We can answer that with real data, from the sites we vet for SaaS link placements.

Across the 100-plus SaaS publishers currently in the Saaslinks catalog, the sites that pass vetting share a clear profile:

• An average Domain Rating of 50, with 15 sites at DR 70 or above and most of the rest in the healthy DR 30 to 69 range. These are established publications, not inflated metrics.
• 82% have a spam score of 5 or lower. Low spam is not a vanity number here. It is the kind of clean, trusted signal that keeps a site in the sources AI engines draw from.
• Together, these sites pull more than 600,000 organic visits a month and rank for over 200,000 keywords. That is real audience and real topical relevance, not parked domains dressed up with metrics.
• The catalog sits exactly where SaaS buyers need mentions: the biggest niches are SaaS, B2B, tech, martech, AI, e-commerce, and devtools.

That data points to the same conclusion as the Ahrefs study: relevance and credibility beat raw volume. A DR 45 publication that genuinely covers your category, has real traffic, and carries a low spam score is a better home for both a link and a mention than a DR 70 generalist that AI systems have no reason to connect with your topic. Topical fit is exactly what a SaaS-only catalog is built to deliver, which is why we show every metric with its source and refresh date instead of asking you to trust a number. (For how to read those metrics yourself, see how to judge a link before buying.)

The SaaS playbook: earn both, in the right order

Here is the practical sequence that uses both signals instead of pitting them against each other.

1. Make your own pages worth citing first. AI engines reward content they can quote. The Princeton-led GEO study (tested on 10,000 queries) found that adding statistics, direct quotations, and cited sources raised visibility in generative engines by roughly 25% to 28% per technique, and up to around 40% overall. Original data, clear structure, and clear sourcing are the cheapest AI-visibility wins you have.
2. Earn placements on relevant, trusted publications. This is the move that gets you a link and a mention at once. Put topical fit and real traffic ahead of headline DR. Digital PR and original research are the strongest version, and we walk through it in digital PR for SaaS.
3. Build named, unlinked presence on purpose. Podcasts, expert roundups, community answers, and "best X tools" lists all connect your brand to a topic in the model's view, even without a link. Use branded anchor text on the links you do place, so you reinforce your brand name and not just an exact-match keyword.
4. Keep the basics running. Get pages indexed, fix internal linking, and keep a clean, varied link profile. None of the AI data argues against this. It argues for doing it well.
5. Measure the right thing. Track rankings and referral traffic, but also track whether your brand shows up in AI answers for your core questions, and which sites get cited when it does. Then work to be mentioned on those sites.

The bottom line

The "3x" is real, but the conclusion people draw from it usually is not. Brand mentions correlate more strongly with Google AI Overview visibility than backlink volume does, and that should change how you build authority, not whether you do. Mentions and links come from the same work: showing up, credibly and on-topic, on sources that both people and AI models trust.

That is the bet Saaslinks is built on. We do not compete on catalog size, because size was never the signal. We list SaaS-relevant publications with real, source-dated organic traffic and low spam scores (the kind of credible sites AI answer engines actually pull from), and we back every placement with a 30-day indexation guarantee, because a link or a mention a model never sees cannot help you. To see how that works in practice, here is how a link-building marketplace works.

In the AI-answer era, the goal is the same as it always was, only sharper: be the obvious, credible name your category keeps citing.