Your customers are asking ChatGPT which product to buy, Google's AI Overviews are answering their questions before they ever reach your website, and Perplexity is summarizing your category without mentioning you. If your brand isn't showing up in those answers, you're invisible at the exact moment buying decisions get made.
The good news: this isn't a black box anymore. Ahrefs has spent the last two years running some of the largest studies in the industry — correlating AI visibility against dozens of factors across 75,000 brands, analyzing 56 million AI Overviews, and tracing millions of AI citations back to their sources. The data points to a clear playbook, and some of it will surprise you.
Why AI search visibility is now a board-level problem
A few numbers set the stakes:
- ChatGPT reached 900 million weekly active users in early 2026.
- AI Overviews now appear on roughly 16% of U.S. Google searches, and coverage more than doubled in a matter of months after Google's March 2025 core update.
- When an AI Overview is present, Ahrefs found the top organic position loses about 58% of its clicks — up from 34.5% when they first measured it in early 2025. Pew Research confirms the pattern: users click a result on only 8% of searches with an AI summary, versus 15% without.
Here's the counterweight, though: the traffic AI search does send is extraordinary. Ahrefs found that AI-referred visitors made up just 0.5% of its visits but drove 12.1% of signups — roughly 23x the conversion rate of organic traffic. Similarweb data shows ChatGPT visitors converting at 7% versus Google's 5%, with nearly double the session time. Fewer clicks, but far warmer ones. The brands that win AI visibility capture disproportionate value.
What actually drives AI visibility (it's not what most SEOs think)
In December 2025, Ahrefs published a correlation study of 75,000 brands across ChatGPT, Google AI Mode, and AI Overviews. The ranked factors:
- YouTube mentions — the single strongest correlate of AI visibility on every platform (0.74 for AI Overviews, 0.737 for ChatGPT)
- Branded web mentions — how often your brand is named across the web (0.656–0.709)
- Branded anchors and branded search volume — moderate correlations
- Domain Rating and backlinks — surprisingly weak (roughly 0.2–0.33)
Read that again: the metrics SEOs have optimized for a decade — backlink counts, Domain Rating, publishing volume — barely move the needle. What matters is how often and where your brand gets talked about.
The earlier AI Overviews edition of the study quantified how winner-takes-all this is: brands in the top quartile of web mentions averaged 169 AI Overview mentions, versus just 14 for the next quartile down — and 26% of brands had zero AI Overview presence at all.
With that foundation, here's the playbook.
1. Make brand mentions your primary KPI
Since branded web mentions are the strongest controllable signal, digital PR graduates from nice-to-have to core AI search strategy. Ahrefs' LLM optimization guide points to Herman Miller, which generated 273 pages of press coverage associating the brand with "ergonomic" — exactly the kind of repeated topical association LLMs learn from.
Quality beats volume here. In Ahrefs' analysis of what we actually know about LLM search, a single data-driven study from background-check company Checkr, picked up by a handful of authority publications, translated into consistent ChatGPT presence. Original research, expert commentary, and data journalism earn the kind of mentions that compound.
2. Stop treating YouTube as a side channel
The most underpriced finding in the entire dataset: YouTube mentions correlate with AI visibility more strongly than any other factor Ahrefs measured. YouTube is also one of the most-cited domains in AI Overviews, alongside Reddit, Wikipedia, and Quora. If your brand has no YouTube footprint — your own channel, creator reviews, mentions in comparison videos — you're missing the strongest signal in the model.
3. Show up on the pages AI already trusts
Across 56 million AI Overviews, the top 50 cited domains account for nearly 29% of all mentions — led by Reddit (~2.97M mentions), Wikipedia (~2.92M), and Quora (~2.28M). You can't outrank those sites, but you can be present on them: an accurate, well-sourced Wikipedia presence, authentic community participation on Reddit, and inclusion in the "best X" listicles AI leans on heavily. Ahrefs found "best X" lists make up 43.8% of page types ChatGPT cites for top-of-funnel queries — and yes, publishing your own comparison list works; 67.6% of "best software" SERPs feature a company ranking itself first.
One caveat from Ahrefs' study of 1.4 million ChatGPT prompts: ChatGPT reads Reddit heavily but rarely credits it, so treat community presence as influence-building, not citation-building.
4. Structure content for machines that skim
The same 1.4M-prompt study found the biggest on-page differentiator between cited and uncited pages is semantic similarity between your title and the "fan-out" queries AI generates behind the scenes. Practical translation:
- Lead with the answer (a clear, quotable summary up top), then elaborate.
- Match titles and headings to how people actually phrase questions.
- Use descriptive, natural-language URL slugs — pages with readable slugs were cited at 89.8% versus 81.1% for opaque ones.
- Keep content fresh. Ahrefs' freshness study of 17 million citations found AI assistants cite content roughly 26% newer than what organic Google surfaces.
- Include quotes, statistics, and cited sources — a 10,000-query study referenced by Ahrefs found these elements lift AI visibility 30–40%.
5. Keep doing SEO — but optimize for topics, not keywords
Rankings still matter, just less exclusively. As of Ahrefs' March 2026 citation analysis, only 37.9% of URLs cited in AI Overviews rank in the organic top 10 — down from 76% a year earlier. Query fan-out means AI systems break one question into many sub-queries and pull from pages ranking for the long tail, sometimes from beyond the top 100. Deep topic clusters that cover adjacent questions now earn citations that a single optimized page can't.
And don't neglect your bottom-funnel pages. Over 80% of Ahrefs' own AI search traffic lands on its free tools, product pages, and homepage — not the blog. AI prompts average 42 words and are mostly commands ("recommend a tool that does X"), so pages that clearly explain what you do, for whom, at what price are what AI retrieves when buying-intent prompts come in.
6. Skip the snake oil
Two "AEO hacks" the data says to deprioritize. Ahrefs' controlled schema experiment across nearly 2,000 pages found adding JSON-LD produced no meaningful citation uplift — major AI systems parse visible HTML, not structured data, at retrieval time. (Keep schema for classic rich results; just don't expect it to win AI citations.) And llms.txt? Across 137,000 domains, 97% of llms.txt files were never requested by anyone — no AI bot proactively looks for it.
One thing you should check: that you're not blocking AI crawlers like GPTBot in robots.txt. Roughly 6% of sites do, forfeiting visibility entirely.
7. Measure it, or it didn't happen
You can't manage what you can't see. Three layers of measurement:
- AI share of voice. Tools like Ahrefs Brand Radar track your brand's mentions, citations, and share of impressions versus competitors across AI Overviews, AI Mode, ChatGPT, Perplexity, and Gemini.
- AI referral traffic. GA4 has no default AI channel — build a custom channel group with regex on AI referrers (chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com). Expect undercounting; some platforms strip referrers.
- The decoupling. Watch for impressions rising while clicks fall in Search Console — the signature of AI Overviews absorbing your clicks — and track cited-versus-uncited CTR. Seer Interactive found brands cited in an AI Overview earn about 120% more clicks per impression than uncited brands on the same SERPs.
The bottom line
AI search visibility is brand visibility. The models recommend what the web talks about — on YouTube, in the press, on Reddit and Wikipedia, in comparison lists — and they retrieve pages that answer questions directly, stay fresh, and speak plainly about what a product does. That's less a new discipline than a forcing function for marketing fundamentals: be genuinely notable, be talked about in the right places, and make your best pages effortless for a machine to quote.
The brands treating this as a 2027 problem are already losing answers today. Start with an audit of where you appear (and don't) in AI responses, put brand mentions and YouTube on the roadmap, and measure share of voice monthly. One in four brands still has zero AI Overview presence — which means the answer box in your category is still up for grabs.
Sources: All statistics link to their original studies above, primarily Ahrefs' research library (2025–2026), Pew Research Center, Seer Interactive, Similarweb, and TechCrunch.