TL;DR: OtterlyAI just analyzed over 1 million citations across ChatGPT, Perplexity, and Google AI Overviews.
Study Scope: Analysis of 1+ million AI citations across ChatGPT, Perplexity, and Google AI Overviews from January-September 2025.
Key Findings:
- Community platforms (Reddit, Quora) capture 52.5% of citations vs. 47.5% for brand domains
- Platform behaviors differ significantly: ChatGPT provides clickable links, Perplexity emphasizes domains, Google prioritizes brand visibility
- 73% of sites have technical barriers blocking AI crawler access
Primary Actions: Fix robots.txt blocks → Audit crawler access → Create reference-grade content
📌 Key Takeaways
The findings reveal a fundamental shift in how discovery works online. We are shifting focus from search engine rankings to being cited by AI systems.
- AI citations determine visibility: Community sites and Wikipedia dominate AI citations across all platforms
- Platform differences matter: Treat ChatGPT, Perplexity, and Google AI Overviews as separate citation environments with distinct behaviors
- Fix crawlability first: robots.txt, CDN rules, and JavaScript (JS) rendering prevent 73% of sites from being crawled
- Create reference-grade content: Chunked, quotable, schema-tagged pages receive 3-5x more citations
- Operationalize measurement: Add citation and mention metrics to analytics dashboards; monitor server logs for AI crawler access
Check out our interactive AI Citations Website here or download the full PDF on The State of AI Citations here:
When you ask ChatGPT a question, have you noticed it frequently cites Reddit and Wikipedia? This isn’t coincidental. A new study by OtterlyAI analyzed over one million website citations across major AI search platforms during September 2025. The findings reveal distinct citation patterns that could reshape how businesses approach their digital presence. Understanding these patterns unlocks the secret to getting your content noticed in the AI-driven search landscape.
The stakes are higher than traffic loss. When AI engines ignore your brand, you lose trust, authority, and market position.
Here’s what the data reveals, and more critically, what you need to do about it.
The AI Citation Gap: Which Websites Get Cited on AI Search?
Run this test right now: Search your brand plus a key topic in ChatGPT and Perplexity. Are you there? If you see competitors, Reddit threads, or generic advice where your brand should be, you have a citation problem.
The study found that brands represent 52.5% of all citations across AI search engines. That sounds good until you realize the other 47.5% goes to news sites (20.3%), community forums (5.9%), and other sources. Your expensive content is competing with free Reddit threads, and often losing.
This matters because AI-generated responses are becoming the first and last stop for users. They don’t click through to verify. They trust the AI’s synthesis. If not cited, prospective users may not encounter your brand through AI-generated answers.
Key Finding #1: The Platform Paradox: Why Each AI Engine Sees Your Brand Differently
Stop treating “AI optimization” as a single strategy. Each platform has distinct citation preferences, and understanding these differences changes everything.
ChatGPT favors Reddit, Wikipedia, and news sites. The study shows brands get mentioned frequently but receive weak link citations. ChatGPT will talk about your brand; it just won’t send users to you. This creates awareness without conversion, a modern challenge for conversion tracking.
Perplexity leans even harder on Reddit and community forums (16.9% of citations). The platform offers balanced mention and citation ratios, making it potentially more valuable for driving traffic. When Perplexity cites you, users can actually find you.
Google AI Overviews show the strongest brand preference at 59.8% of citations (compared to 44.7% for ChatGPT and 28.9% for Perplexity). Google also provides the highest number of clickable link citations. The catch: AI Overviews only appear about 33% of the time. Traditional domain authority still matters here; elsewhere, not so much.
The strategic implication is clear. You need three separate content approaches mapped to three different citation environments. A monolithic “AI SEO strategy” will fail.
Key Finding #2: The Winners Aren’t Always the Usual Suspects
We ran a domain-level analysis of the most cited websites across all platforms. While some results were predictable (Wikipedia, YouTube), others signal a major shift in authority dynamics.
Here’s a snapshot of the top domains by AI platform:
Reddit dominated every platform. It’s the #1 most cited domain overall. And it’s not even close.
The Community Advantage: Reddit
Community-driven platforms dominate AI citations across the board. Reddit claims the top position despite experiencing “a significant drop on Sept 11” in the data.
Wikipedia maintains strong performance particularly within ChatGPT results. News and media sites round out the top citation sources, suggesting AI systems value frequently updated, discussion-rich content.
User-generated content often provides diverse perspectives, real-world experiences, and timely information that AI engines find valuable. Editorial content consistently outperforms commercial pages, with AI search engines showing a clear preference for informational over transactional content.
Key Finding #3: Technical Barriers to AI Visibility: Why AI Can’t See Your Website
Overview
Some websites are invisible to AI crawlers, and it has nothing to do with content quality. The study reveals three technical barriers killing your citability before your content even gets evaluated.
Barrier 1: Robots.txt Blocks
The Problem: AI crawlers like GPTBot and ClaudeBot are blocked by default robots.txt configurations.
How to Check:
- Open your robots.txt file:
yoursite.com/robots.txt - Search for these user-agents:
- GPTBot
- ChatGPT-User
- ClaudeBot
- PerplexityBot
- OAI-Searchbot (OpenAI Searchbot)
- Look for “Disallow: /” next to any of these
How to Fix:
# Allow AI crawlers
User-agent: GPTBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: ClaudeBot
Allow: /Barrier 2: Content Delivery Network (CDN) Restrictions
The Problem: CDN security rules often block non-browser user-agents.
How to Check:
- Review CDN security rules (Cloudflare/AWS/Akamai dashboard)
- Check for user-agent blocking rules
- Test with:
curl -A "GPTBot" https://yoursite.com/article
How to Fix:
- Add AI crawler user-agents to CDN allowlist
- Configure rate limiting (not blocking) for AI bots
Test: Use AI Crawler Simulation Tool to verify access
Barrier 3: JavaScript (JS) Rendering Issues
The Problem: Content requiring JavaScript execution may not be accessible to AI crawlers.
How to Check:
- Use Google’s Rich Results Test
- Compare rendered vs source HTML
How to Fix:
- Implement server-side rendering (SSR) for critical content
- Use progressive enhancement (content visible without JS)
- Add static HTML fallbacks for dynamic sections
Test: View page source (Ctrl+U) – main content should be visible
Key Finding #4: Content That Gets Cited vs. Content That Gets Ignored
Editorial content dominates citations across all platforms. The data breaks down like this: news and media represent 20-30% of citations depending on platform, blog and personal sites capture 3.5-10%, community forums take 5.9-16.9%, and brand sites get mentioned often but linked rarely (except in Google AI Overviews).
This tells us something profound about what AI considers “reference-grade content.” Instead of creating “SEO content,” you need to create reference-grade, easily quotable content with clear sources. Can your content be quoted without additional context? Does it answer questions cleanly, or does it dance around them to hit keyword targets?
Structure matters more than ever. Content needs to be chunked properly, tagged with schema markup, and formatted for easy extraction. Ask yourself: would Wikipedia link to this page? If the answer is no, AI probably won’t cite it either.
Key Finding #5: The Crawler Persona: Treating AI as Your Most Important User
You have a new primary audience: AI crawlers. The study lists them explicitly. GPTBot crawls for training data. OAI-Searchbot responds to real-time search queries. ChatGPT-User handles custom GPT interactions. PerplexityBot and Perplexity-User work together for search and user actions. ClaudeBot, Claude-Searchbot, and Claude-User follow the same pattern.
Each crawler has specific needs. They require clean, crawlable HTML without JavaScript dependencies. They need structured data and schema markup to understand context. They look for explicit permissions in robots.txt files. They favor chunked, quotable content over dense paragraphs. They value clear source attribution.
Check your server logs for these crawler names. You cannot manage what you do not measure. If you see low traffic from AI crawlers, you have a technical problem to fix before worrying about content optimization.
The study emphasizes this repeatedly: crawlability comes first. A perfectly written article that AI cannot access helps nobody. An adequately written article that AI can easily parse and cite will outperform your masterpiece every time.
The Strategic Implications: What This Means for Your Business?
Marketing leaders face a budget reallocation question. Does community presence matter more than other marketing activities now? The citation data suggests yes for awareness, maybe for conversion. You need an GEO / SEO manager, someone who tracks where your brand gets mentioned and works to increase quality citations. Metrics must shift from rankings to brand mentions and website citations.
Content teams now face a 10x higher quality bar. The study makes this clear through the citation gap between brand content and community content. Distribution matters as much as creation. Every content piece needs a citability check before publication. Will AI reference this, or will it find better answers elsewhere?
Teams must recognize that static site generation is now business-critical. Schema markup is no longer optional. Crawler access is a feature, not a security concern. These changes require cross-functional coordination. Your marketing, content, and technical teams need to work together on citation strategy. Siloed approaches will fail.
Interactive AI Citations Report
Great news: You can also explore our AI Citations Report as an interactive website. And you can even break down results per different countries and AI Search Engines.
View AI Citations Report here.
Want to See the Full Results as a PDF?
We’ve packed the full study with:
- Domain rankings across platforms
- Search intent breakdowns
- Technical recommendations
- Benchmarks for brand mentions and link citations
Don’t rely on assumptions. See exactly how your category is performing.
👉 Download the Full 2025 AI Citation Study here:
📚 Resources & Downloads
Free Tools
- AI Crawler Simulation – Test crawler behavior on your URLs
- GEO Content Checker – Analyze content for AI optimization
- GEO Audit Tool – Sign up for free here
Interactive Report
- 🔍 Explore Interactive Dashboard – Filter by industry, platform, time period
Need help optimizing for AI search?
[Start Free Trial] [Book Demo] [View Pricing]