Backlinks still matter for AI search in 2026, but far less than most people assume: an Ahrefs analysis of 75,000 brands found brand web mentions correlate with AI citation rates at about 0.66, roughly three times stronger than backlinks at about 0.22. The unit AI engines like ChatGPT, Perplexity, Claude, and Google Gemini count is no longer the link, it is the mention, with or without a hyperlink, sitting on a source the engine already trusts. Backlinks still help indexing, authority, and the citation-heavy engines, so this is not a “links are dead” post. It is an honest account of what the current data shows and where you should actually spend your effort.
Do backlinks matter for AI search?
Backlinks matter for AI search, but they are a secondary signal in 2026, not the primary one. The Ahrefs study of 75,000 brands put the correlation between backlinks and AI citation at roughly 0.22, while unlinked and linked brand mentions correlated at about 0.66, meaning being talked about predicts citations about three times better than being linked to. Links still help crawlability, indexing, and traditional authority that some engines lean on, so they retain real value, they are just no longer the lever they were in classic SEO.
The distinction that matters is between a link and a mention. A backlink is a hyperlink pointing to your site; a mention is any reference to your brand or name on another source, hyperlinked or not. AI engines build their understanding of who exists and who is authoritative from mentions across trusted sources, so a press article naming your company without linking to it still teaches the engine that you exist in that category and geography. That reframing is the heart of the shift, and we cover it in depth in backlinks vs citations for AEO.
Why do brand mentions beat backlinks for AI citations?
Brand mentions beat backlinks because AI engines model the world as entities and their associations, and mentions are how those associations form, whereas links are a narrower, gameable signal built for a different era. When a trusted source names your brand alongside a topic, the engine learns you belong to that topic; a raw link carries far less of that meaning.
The mechanism differs by engine, which explains the gap. ChatGPT constructs many answers from parametric knowledge baked into its training data, so its recommendations depend on brand entity strength built over time through third-party mentions and co-occurrence with category terms, not on page-level link optimization. Perplexity and Google AI Mode do live retrieval and cite sources more directly, so they respond somewhat more to fresh, well-sourced pages, but even there the sources they trust are the ones with strong mention footprints. The practical takeaway: mention frequency across authoritative sources is a roughly three-times stronger predictor of AI citation than backlinks, and it compounds. We explain the engine-by-engine differences in what sources do AI engines cite.
Not sure whether AI engines already associate your brand with your category, or skip you for a competitor? Get your free AI visibility audit and see exactly where your mentions earn citations and where they fall short.
Does link building still have any value for AI search?
Link building still has value for AI search, mainly through the sources that generate the links rather than the links themselves. A backlink from a respected publication usually comes packaged with a brand mention, editorial context, and co-occurrence with your topic, and those are the signals AI weights, so pursuing quality coverage still pays off even though the link is not the point.
Think of it as a shift in what you are optimizing for. Instead of chasing link volume or domain-rating math, chase presence on sources AI trusts: earn coverage, get named in roundups and comparisons, appear on high-authority publications and community platforms. The link that comes with a Forbes feature matters less than the fact that Forbes named you next to your category. Links also still support crawlability and traditional ranking, which feed the engines that blend classic search with generative answers. The point is not to abandon links but to stop treating them as the finish line, a reframing we cover in does domain authority matter for AI search.
What should you do instead of chasing backlinks?
Instead of chasing backlinks, build brand mentions across trusted sources through digital PR, earned coverage, expert commentary, and community presence, because mention frequency is the stronger citation predictor. The goal is to be named, consistently and credibly, wherever the engines look, not to accumulate links for their own sake.
Four moves do most of the work. First, invest in digital PR that lands your brand name in publications and industry outlets AI trusts, since a mention on a credible source is worth more than a link on a weak one. Second, seed co-occurrence, get your brand named alongside your category terms in comparisons, roundups, and “best of” lists, so engines associate you with the topic. Third, show up on community platforms like Reddit and Quora and in podcasts and expert roundups, which AI engines cite heavily. Fourth, publish original data and named frameworks that others reference, generating mentions passively over time. The PR engine behind this is detailed in digital PR for AI visibility and how to get your brand mentioned by AI.
How do you measure mention-based AI visibility?
You measure mention-based AI visibility by tracking how often your brand is cited in AI answers, how consistently you co-occur with your category, and whether that presence is growing, rather than by counting links in a backlink tool. The metric that matters is citation share, not link volume.
Use AI-visibility platforms like Profound, Otterly, or Semrush’s AI features to track share of voice and citation frequency across ChatGPT, Perplexity, Gemini, and Claude. Run manual checks monthly by asking each engine your category questions and logging whether you are named and what sources it references. Watch GA4 for referral traffic from AI domains, and monitor brand-mention tools to see where new unlinked references appear. Together these show whether your mention footprint is translating into citations, which is the outcome that actually drives AI-sourced traffic, tracked the way we describe in AI share of voice.
What are the common misconceptions about backlinks and AI search?
The common misconceptions about backlinks and AI search are that links are dead, that domain rating drives citations, and that a strong backlink profile alone earns AI recommendations. All three are wrong in the same direction: they overweight links and underweight the mentions and entity signals the 2026 data shows actually predict citation.
Three myths deserve correcting. First, “links are dead” overstates it, the Ahrefs study still found a 0.22 correlation between backlinks and citation, so links help, they are just not the primary lever. Second, “high domain rating means AI will cite me” confuses a link-based authority score with entity strength, when a site can have strong domain metrics and still be invisible to ChatGPT because it lacks category co-occurrence and third-party mentions. Third, “if I build enough links, citations follow” ignores that a link without a meaningful mention teaches the engine little, while an unlinked mention on a trusted source teaches it your brand exists in a category. The correction in each case is the same: optimize for being named on sources AI trusts, not for link math, a reframing we extend in does domain authority matter for AI search.
Frequently asked questions
Are backlinks completely irrelevant to AI search now? No. The Ahrefs study of 75,000 brands found backlinks still correlate with AI citation at about 0.22, so they retain value, especially for crawlability and the citation-heavy engines like Perplexity. They are simply a weaker signal than brand mentions, which correlate at about 0.66, so links should be a byproduct of earning coverage, not the primary goal.
What exactly counts as a brand mention? A brand mention is any reference to your brand or name on another source, whether or not it includes a hyperlink. A news article that names your company, a Reddit thread that recommends you, or a roundup that lists you next to competitors all count. AI engines learn your existence and associations from these references regardless of linking.
Why does ChatGPT respond less to links than Perplexity? ChatGPT builds many answers from parametric knowledge in its training data, so it depends on entity strength accumulated through third-party mentions and category co-occurrence rather than live page-level links. Perplexity and Google AI Mode do more live retrieval and cite sources directly, so they respond somewhat more to fresh, well-sourced pages, though they still favor sources with strong mention footprints.
Should I stop doing link building entirely? No, but reframe it. Pursue quality coverage on trusted sources, which usually delivers a brand mention, editorial context, and topic co-occurrence alongside any link. Stop chasing raw link volume or domain-rating math as an end in itself, and judge outreach by whether it gets you named on sources AI trusts.
How do brand mentions actually get into AI answers? Through training data and live retrieval. Mentions on trusted sources become part of what the model learns during training and part of what retrieval engines surface at query time. When enough credible sources name you alongside a topic, the engine treats you as a relevant entity and includes you in the answer.
How long does it take to build mention-based visibility? Typically a few months of consistent digital PR and earned coverage, since entity associations build through repetition across sources and update as engines retrain and re-crawl. Brands that sustain a steady mention cadence usually see citation share improve over two to four months, with gains compounding as the footprint grows.
The honest answer for 2026 is that backlinks still count, but they have been demoted: the Ahrefs data on 75,000 brands is clear that being named across trusted sources predicts AI citations about three times better than being linked to, because engines like ChatGPT and Perplexity count entities and associations, not just hyperlinks. Teams still pouring their budget into link volume are optimizing for the last era of search while the mention footprint, the thing that actually drives citations, goes unbuilt. Chase coverage, co-occurrence, and community presence, let the links come along for the ride, and measure citation share instead of link count. The teams that internalize this reallocate budget from raw link acquisition to digital PR and earned mentions, and they judge every campaign by whether it got them named on a source AI trusts, not by how many domains linked back. That single change in scoreboard is what separates brands that show up in AI answers from brands with strong link profiles and no citations to show for them. Want to see whether your brand’s mention footprint is earning AI citations or leaving them on the table? Claim your free AI visibility audit and get a clear read on where you stand.
Tagged