By Phil Helsel · Not-Just-AI Advisors · San Antonio, TX

If you’ve spent time working on your website’s Google ranking, you probably have a good sense of how to boost your site’s visibility. But new research from Cyrus Shepard (2026), along with data from Ahrefs, BuzzStream, and Seer Interactive — covering millions of AI citations — reveals that search engines powered by AI (like ChatGPT, Perplexity, and Google’s AI Overviews) play by a different set of rules than classic Google search results.

Some traditional SEO tactics still help, but others aren’t as important as they once were. And surprisingly, a few of the biggest factors for getting cited by AI have very little to do with your website at all. That’s why we’re breaking down these key differences for you here.

Here’s what the latest data tells us.

Key factors for getting cited by AI

Cyrus Shepard’s meta-analysis examined 23 factors, rating the strength of the evidence for each across 54 studies, patents, and case studies focused on ChatGPT, Gemini, and Perplexity. To get you oriented, here are the factors with the strongest evidence — ranked from most to least important:

Factor	Score (0–10)	What it means
URL Accessibility	9.5	Your page needs to be easy to access — no paywalls, errors, or blocks, or it won’t get cited.
Search Rank	9.4	Being in Google’s top 10 still really helps your chances of being cited by AI.
Fan-out Rank	9.3	It’s not just about the main keyword — ranking for lots of related sub-questions matters, too.
Preview Control	9.2	If you block snippets (like using “nosnippet”), you can prevent AI from citing you, since they rely on snippet data.
Query-Answer Match	9.2	Directly answering the exact question users ask will beat content that’s only loosely related.
Topic Cluster Ranking	8.9	Covering a topic in depth means your page gets cited for more related questions.
Answer Near the Top	8.8	Put clear answers at the beginning of sections — AI is more likely to pick them up.
AI-Ready Structure	8.6	Use clear headings and tables so AI can easily extract answers.
Self-Contained Passages	8.0	Make sure each section can stand on its own, without needing the rest of the page for context.
Cites Sources Internally	8.0	Referencing your sources makes your content look more reliable to AI.
Freshness	7.0	Newer content is slightly more likely to be cited, but it’s not the most important factor.
Brand Web Mentions	High*	If your brand is mentioned around the web, you’re about 3x more likely to be cited by AI than with just backlinks.
LLMs.txt	2.0	There’s little evidence that the LLMs.txt file impacts citations at all.

*Brand Web Mentions wasn’t scored on Shepard’s 0–10 scale, but a separate Ahrefs study found they’re a very strong signal in practice.

What still matters from traditional SEO

Your page must be accessible. This is the top factor. If your content is blocked, paywalled, or broken, it simply won’t show up in AI citations — no matter how good it is.
Ranking well helps. Being in the top 10 of Google’s results remains a big advantage. Nearly 38% of AI Overview citations come from these top spots. So, classic SEO still counts, but it’s not the only thing that matters anymore.

Where AI citation patterns break from classic SEO

You might be surprised by the biggest change in the world of AI citations: brand mentions now matter more than traditional backlinks.

Research from Ahrefs (covering 75,000 brands) found that just having your brand name mentioned — without even a link — correlates with AI Overview visibility much more strongly (0.664 correlation) than the number of backlinks you have (0.218 correlation). In other words, being talked about across the web is almost three times more important than racking up raw links.

The difference is dramatic: brands in the top quartile averaged 169 AI Overview mentions, while the next group down only managed 14. That’s not a gradual decline — it’s a classic “winner takes most” scenario.

Third-party coverage is crucial. Separate analysis from Seer Interactive of large language model (LLM) training data found:

Earned media (news, forums, reviews, directories) makes up 48% of AI citations
Content published on a brand’s own site only accounts for 23%
Brands featured in editorial sources are 6.5x more likely to be cited than those relying only on their own websites

So, what does this mean in practice? Digital PR is more important than ever. Getting quoted in industry articles, appearing in “best of” lists, and having real visibility on forums or review sites can move the needle more than just chasing backlinks.

Why does this happen?

AI engines don’t just “rank” pages — they retrieve answers from a cluster of related questions. That’s why only 38% of AI Overview citations come from the top 10 Google pages — down from 76% in 2025. Pages ranked 11–100 (and even those beyond 100) now each make up about 31% of citations. It’s not about being #1 for a single keyword but about being a trusted source across many related topics.

Depth beats shallow wins. Websites that cover topics thoroughly are more likely to get cited repeatedly, even if they never hit the top spot for any one keyword.

Content has to be easy for AI to extract

AI systems don’t cite whole pages — they pull out specific passages. The research shows they prefer content that:

Answers the exact question up front (not buried halfway down)
Uses clear headings and tables for easy scanning
Makes each section self-contained
Sticks to facts and specifics over marketing fluff

Overly promotional content gets ignored; AI models want information they can quote or summarize with confidence.

Editorial content dominates. A BuzzStream study (4 million citations, 3,600 prompts, 10 industries) found:

53% of citations come from blogs and editorial content
When you ignore branded searches, earned editorial accounts for about 80% of citations
Press releases almost never get cited

Also, YouTube is now the most-cited domain in Google AI Overviews, making up 5.6% of all cited URLs — a sign that “content” now includes much more than just text.

Freshness: helpful, but not everything

AI-cited content is generally newer than top 10 organic results — about 26% fresher on average (1,064 days old vs. 1,432 days for organic, according to Ahrefs). ChatGPT favors the newest content; Google’s AI Overviews are closer to traditional results.

Be cautious about viral stats. That claim that “fresh content gets cited 4.3x more often”? There’s no real study behind it. Take similar multipliers with a grain of salt.

Two factors that are more hype than reality

Schema markup: It does help Google AI Overviews and Bing Copilot (65–71% of cited pages use it), but for models like ChatGPT and Claude, which answer from training data rather than live retrieval, schema isn’t a direct factor. Use it as good practice, but don’t expect miracles.
llms.txt: This file scored last in the 23-factor meta-analysis. There’s no credible evidence it boosts AI citations, so it shouldn’t be a top priority.

The real shift: from rankings to trust and coverage

Traditional SEO is about ranking on a list people might click. AI visibility is about becoming the answer AI trusts enough to show or paraphrase.

That shifts the focus from pure links to:

Brand presence beyond your own site
Thorough topical coverage
Clear, extractable structure

It’s about being credible, visible, and well-covered in the places AI looks for answers.

What should you prioritize?

Here’s the order the evidence supports:

Fix accessibility — if AI can’t reach your pages, nothing else matters.
Invest in classic SEO — ranking still helps, but less than before.
Build brand mentions and digital PR — editorial coverage is now more powerful than backlinks.
Cover your topic broadly — think clusters of questions, not single keywords.
Structure content for extraction — put the answer first, use headings/tables, and make sections stand-alone.
Keep content updated; freshness is a modest but real factor.
Treat schema markup and llms.txt as maintenance — not game changers.

Not sure where your own site stands on any of this? That’s exactly what an AI search visibility audit is built to answer — a look at how accessible, structured, and well-represented your business is across the sources AI engines are already pulling from. Get in touch if you’d like a second pair of eyes on it.

Phil Helsel is the founder of Not-Just-AI Advisors, a San Antonio-based local SEO and search visibility consultancy. He spent 25+ years in enterprise technology, including a focus on search at Fast Search and Transfer, Oracle, and Microsoft.

What Actually Gets You Cited by AI