New Data Finds Gap Between Google Rankings And LLM Citations

Table of Contents

A new study reveals that Google search rankings don’t match the way AI tools like ChatGPT, Gemini, and Perplexity choose their cited sources.

Key Takeaways

Perplexity searches the web in real time, so its answers look closest to Google results.
It shared around 25–30% of domains with Google.
About 43% of the domains Perplexity cited were also found in Google search results.

These models depend more on their training data and less on current Google rankings.
ChatGPT had only 10–15% domain overlap with Google and very few URL-level matches.
Gemini was less stable—some results matched Google closely, while others barely matched at all.

Overall, Gemini shared only 4% of Google’s domains, even though they made up a good part of Gemini’s citations.

Just ranking on Google is not enough anymore.
Perplexity rewards websites that already do well in Google.
ChatGPT and Gemini cite fewer sites and rely more on trusted, authoritative sources from their training—not necessarily what’s ranking today.

So your website may rank high on Google but still not appear in AI answers.

Strengthen your topical authority – Publish deep, expert-level content that shows expertise.
Earn high-quality backlinks – Large Language Models prefer domains with strong authority and trusted signals.
Improve structured data and clean site architecture – Makes your content easier for AI systems to understand.
Update content regularly – Fresh, well-optimized pages are more likely to be retrieved by LLMs.
Focus on E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness)
Target long-form, fact-based content – LLMs often cite clearer, data-backed pages.
Create content that answers questions directly – Helps AI models pick up your page while forming responses.