Ranking Factor
A ranking factor is a signal a search engine or AI system weighs to decide where a page appears, or whether a source gets cited at all. Google has confirmed and documented many of its ranking factors over two decades of public guidance. AI-citation "factors" are a newer, murkier idea: no major AI provider publishes a weighting formula for why one source gets cited over another. What you find in most GEO blog posts is inference from correlational studies, not a confirmed algorithm, and treating the two as the same thing is the most common mistake in this space.
Why ranking factors matter
Position determines whether anyone sees you. In traditional search that has always been true; in AI answers it is now more extreme. Pew Research Center tracked real browsing behavior across roughly 68,879 Google searches and found users clicked a traditional result in only 8% of visits when an AI summary appeared, versus 15% without one, roughly half the click rate. Ahrefs separately found AI Overviews correlate with a 58% lower average click-through rate for the page ranking #1 organically. Zero-click behavior is climbing too: SparkToro and Similarweb measured 68.01% of US Google searches ending with no click at all in early 2026, up from 60.45% two years earlier. Ranking factors matter because fewer clicks are up for grabs, and being the source an AI system actually names is increasingly the only exposure left.
How ranking factors work
For traditional search, Google has confirmed hundreds of signals over the years, grouped roughly into relevance, authority, technical health and user experience, and it periodically discusses categories of these in its own documentation. For AI citation, the picture is different and worth being honest about: no provider has published how ChatGPT, Gemini, or Perplexity actually pick and weight sources. What exists instead is correlational research. An Ahrefs analysis of roughly 75,000 brands found that how often a brand is mentioned across the web correlates with AI citation rate at about 0.664, nearly three times stronger than the correlation with backlinks alone (about 0.218). That is evidence a factor matters, not proof it is causally weighted by any model. Ahrefs also found only about 6-8% of URLs cited by ChatGPT overlap with Google's own top-10 for the same query, and roughly 80% of ChatGPT's cited URLs do not rank in Google's top 100 at all, which tells you these are two different selection processes, not one system with a different coat of paint.
Common mistakes
The biggest mistake is borrowing the vocabulary of SEO ranking factors and applying it uncritically to AI citation, as though someone has confirmed a formula when nobody has. A close second is chasing markup that does nothing: Google's own AI-features guidance states plainly that no special schema, structured data, or AI-specific files are required for AI Overviews or AI Mode, and explicitly warns against writing content "for AI" as a separate exercise. The llms.txt file is a sharp example, Google's John Mueller confirmed in 2025 that no Google Search system reads or acts on it, and an Ahrefs analysis of roughly 137,000 sites that published one found about 97% saw zero measurable referral traffic tied to it. A third mistake is skipping a baseline measurement before changing anything, so there is no way to know later whether an adjustment helped, hurt, or did nothing.
Relation to AI recommendations
The term "Generative Engine Optimization" comes from a specific paper, "GEO: Generative Engine Optimization" (arXiv:2311.09735), by Pranjal Aggarwal, Vishvak Murahari, Karthik Narasimhan, Ameet Deshpande, Tanmay Rajpurohit and Ashwin Kalyan, presented at ACM SIGKDD 2024. It proposed that visibility in generative answers depends on different levers than classic ranking, evidence and citations, clear structure, and topical authority rather than backlinks and keyword density alone. The scale of what's being optimized for is real: ChatGPT reported 900 million weekly active users as of late February 2026, Google's Gemini app passed 1 billion monthly active users in August 2026, and Google AI Overviews reaches over 2 billion monthly users across 200-plus countries. But scale doesn't mean the selection criteria are documented. Brand mentions correlate more strongly with citation than links do, and Wikipedia and Reddit reportedly account for over a quarter of ChatGPT's US citations in one report, yet source-share data like that is volatile and engine-specific, not a stable ranking factor you can bank on. It is directional evidence, treat it that way rather than as a checklist.
Example
Picture a small tax firm in Leipzig. For traditional ranking factors, it fixes the obvious things: the site loads fast, works on mobile, and its pages target clear terms like "freelance tax filing Leipzig" instead of vague ones. Separately, and without assuming the two efforts are the same project, it also gets mentioned by name in a regional business association's article and a local Reddit thread answering tax questions. Months later the firm can check whether its Google position moved and, independently, whether an AI assistant ever names it when someone asks about freelance tax help in Leipzig. Those are two different outcomes worth tracking separately, because they are not driven by the same mechanism.
Common questions
Are AI-citation ranking factors the same as Google's SEO ranking factors?
No. Google has confirmed and documented many of its search ranking factors over the years. No AI provider has published an equivalent formula for why sources get cited in AI answers. Lists of "AI ranking factors" you see in GEO content are mostly industry inference from correlational data, not confirmed algorithms.
Do schema markup and llms.txt count as ranking factors for AI answers?
Not on current evidence. Google states no special schema is required for AI Overviews or AI Mode, and Google has confirmed no Search system reads llms.txt files, a study of about 137,000 sites publishing one found roughly 97% got no measurable referral traffic from it. Structured data can still help machine-readability, but it isn't a proven independent ranking or citation factor.