Summary

This SE Ranking research study analyzed 129,000 domains and 216,524 pages across 20 niches to identify what actually drives ChatGPT citations. The findings challenge common assumptions about AI optimization.

Key Findings:

The research reveals that sites with over 32,000 referring domains are 3.5 times more likely to be cited by ChatGPT than those with up to 200 referring domains, and high-trust domains earn almost 4 times more citations than low-trust sites

What Works:

• Strong domain authority through diverse backlinks (the #1 factor)
• High organic traffic, especially to the homepage (190K+ monthly visitors optimal)
• Comprehensive content (1,900-2,900+ words)
• Regular content updates (quarterly refreshes nearly double citations)
• Brand presence on Quora and Reddit (millions of mentions correlate with 4x higher citations)
• Fast Core Web Vitals (pages under 0.4s FCP get 3x more citations than slower pages)
• Review platform presence (Trustpilot, G2, Capterra, etc.)
• Clear topical URLs and titles rather than keyword-stuffed ones
• FAQ sections and question-based headings (contextually beneficial)

What Doesn't Work:

• LLMs.txt files (negligible to negative impact)
• FAQ schema markup alone (pages without it actually averaged more citations)
• Keyword over-optimization in URLs and titles
• Domain extensions like .gov or .edu (surprisingly, averaged fewer citations)

Strategic Insight:

The study emphasizes that ChatGPT evaluates content similarly to how Google does—prioritizing authority, trust, and quality over technical tricks. Smaller sites can compete by offering comprehensive content, social engagement, and structural clarity, while established sites maintain advantages through diverse backlink profiles and sustained traffic.

The article can be found here.