The Shift to Off-Page AI Visibility
Traditional search optimization focused heavily on exact-match keywords, on-page optimization, and direct backlink profiles. But as artificial intelligence reshapes how people discover products and services, the rules of visibility are changing rapidly. Data from Ahrefs' Brand Radar indicates that brand mentions across third-party websites are now one of the single most influential factors in determining whether a company gets cited in AI-generated answers.
In many ways, we have entered a brand-new era of off-page SEO. When large language models and conversational search engines compile answers, they rely on training data and live web citations that reflect broader digital consensus. If your brand is frequently discussed in context by reputable sources, AI search engines take notice.
Why Citations Matter More Than Ever
According to insights shared by Ahrefs executives, off-page activity that increases how often your brand is mentioned on other sites improves visibility in AI search results across the board — both in answers drawn from a model's training data and in answers assembled from live web search. The benefits of solid off-page work apply to both surfaces; the practical difference is timing. Ahrefs CMO Tim Soulo points out that AI companies do not ingest fresh web data continuously, so there is a lag of months between when your mentions start accumulating and when they are reflected in models that get retrained on a newer snapshot of the web.
That lag is exactly why mentions are a compounding asset. Soulo's guidance is straightforward: find where your industry, your competitors, and your own brand already get mentioned, and get your brand into those places. If an AI chatbot searches and finds those pages, it builds its answer from what appears on them. If an AI provider later retrains its model on a recent crawl, it draws on essentially the same set of pages. Earning mentions early means you are positioned for both the live-answer and the retrained-model scenarios.
The 0.67 Correlation: What the Data Actually Shows
The strongest evidence came from testing. Ryan Law, Ahrefs' Director of Content Marketing, described research that examined many factors correlating with how often a brand appears in AI Overviews. The standout result was branded web mentions, with a correlation coefficient of nearly 0.67 — described as a very strong relationship.
To put that number in context, correlation coefficients run from -1.0 (perfect negative relationship) through 0.0 (no relationship) to 1.0 (perfect positive relationship). A figure near 0.67 is unusually strong for behavioral search data, meaning brands that are mentioned across many places on the open web also tend to appear in a large share of AI conversations. Crucially, these are mentions rather than links: a brand can benefit from a citation even when the referring page never passes a traditional backlink.
This is the core of what practitioners are calling the new era of off-page SEO. As Law frames it, in some ways the content on your own site is no longer as valuable as the content about you that lives on other pages across the web.
Topicality: Context Is Part of the Signal
Volume alone is not the whole story. Ahrefs' team emphasized the topicality of mentions — the context in which your brand is discussed and the topics and other brands it gets mentioned alongside. That context shapes how an LLM understands what your brand is about and when it should recommend it.
You cannot always control the narrative, but this is where disciplined outreach earns its keep. Including the right quotes, framings, and supporting detail in the content you contribute helps build the context you want the models to associate with your name. The goal is not just to be mentioned, but to be mentioned next to the right ideas.
Tactics That Earn AI-Friendly Mentions
Law recommends a practical sequence for building the kind of off-site mentions that move the needle in AI search. First, identify the domains that frequently get cited in AI answers for your topics — the sites AI assistants already trust — and pursue getting your brand mentioned on those pages. Next, prioritize user-generated content platforms such as Reddit and Quora, where conversational recommendations carry weight. Review sites matter for similar reasons, capturing real evaluations that assistants quote verbatim.
YouTube deserves specific attention because its videos are highly cited by AI search; getting your brand referenced in video content and transcripts puts you inside a source type assistants lean on heavily. Throughout, the unifying principle is authoritativeness with relevance. Appearing on credible, go-to websites for a particular kind of information is what counts — and in this context, a site's own authority can be judged partly by whether AI assistants already cite it.
How Brand Radar Measures AI Visibility
The tool underpinning much of this data is Ahrefs Brand Radar, which calculates AI Share of Voice by tracking how often brands are mentioned or cited across ChatGPT, Perplexity, Gemini, Microsoft Copilot, and Google's AI Overviews and AI Mode. Understanding how it works makes its numbers easier to trust and use.
Brand Radar starts from real search behavior rather than fabricated prompts. Queries are collected from Google's "People Also Ask" corpus and Ahrefs' keyword database, then expanded into related sub-questions using two systems: PAA, which reflects how real people actually refine their searches, and Fanout, which expands a query by semantic relationship to ensure logical topic coverage even for questions people rarely type. Running both together gives a fuller view of your AI visibility funnel. Each query is executed against the public web interfaces of the supported assistants, and the raw responses are stored so you can search the corpus for citations (linked URLs) and mentions (string matches) for any term. Monthly query volume is substantial across the board — for example, AI Overviews are sampled at roughly 282 million prompts per month, with each chatbot in the range of 13 to 15 million.
Reading the Numbers Responsibly
Brand Radar's metrics are directional visibility signals, not exact traffic counts, and the team is explicit about the limitations. Estimated Impressions weight mentions by Google search volume to model potential exposure — a modeling choice, not a validated causal link. Strongest coverage is in English, with non-English markets represented proportionally. LLMs occasionally generate hallucinated or malformed links, and these are intentionally left in the data because they reflect genuine model output. Most chatbot data refreshes monthly on a 90-day reporting window for stability, while AI Overviews and AI Mode update continuously. Keeping these caveats in mind turns the data into a benchmarking and comparison tool — useful for tracking your Share of Voice and spotting competitor co-citations — rather than a literal traffic report.
What This Means For Your Off-Page Strategy
Taken together, the signal is clear: brand mentions are the new off-page SEO for AI search. The tactics need not be exotic — appear on credible, relevant websites; earn mentions in the places your industry already talks, including UGC forums, review sites, and YouTube; and shape the context around those mentions. Because models retrain on lagged snapshots of the web, the work compounds over time, rewarding brands that start building citation surface area now. Measure progress with tools like Brand Radar's Share of Voice, interpret the metrics as directional, and let the 0.67 correlation guide where you invest your off-page budget next.
