The Problem Is Real (And Measurable)
Your competitor appears when someone asks ChatGPT "best project management software" or "top marketing agencies in [your city]". You don't. That's not bad luck. It's a technical problem with specific causes.
A study analyzing 7,785 ChatGPT queries and 485,000+ citations across 38,000+ unique domains reveals exactly how ChatGPT selects sources. This data comes from Wellows' ChatGPT Citations Report and Profound's AI Platform Citation Patterns analysis.
Here's what the research shows.
How Does ChatGPT Select Sources to Cite?
ChatGPT doesn't randomly pick websites. It follows predictable patterns:
What Are the Most Cited Sources?
| Source Type | Citation Share | Why | |-------------|---------------|-----| | Wikipedia | 7.8% | Encyclopedic, factual, well-structured | | Reddit | High (licensed) | OpenAI content partnership | | Major Publishers | High | TechCrunch, TechRadar, news agencies | | Commercial (.com) | 80%+ | Dominant domain type | | Non-profit (.org) | 11.29% | Authoritative for specific topics |
Key insight: ChatGPT primarily cites sources from companies that OpenAI has licensed through content partnerships. Wikipedia, Reddit, and major news publishers dominate.
Does Content Freshness Matter?
Yes, but it depends on the query type:
For time-sensitive queries (containing "latest", "2024", "2025", "best X of 2026"):
- ChatGPT consistently cites recent sources
- Pages titled "Best X of 2025" from TechCrunch or TechRadar are frequently selected
- Fresh content wins when the prompt implies recency
For evergreen queries (conceptual, factual, "how does X work"):
- ChatGPT prefers authoritative, in-depth sources regardless of date
- A detailed 2023 guide on "how GPT-4 works" was cited over newer but shallower posts
- Academic research and technical documentation remain common citations
Why Your Competitors Get Cited (And You Don't)
Based on the research, here are the 5 most likely reasons:
Reason 1: They Have Content Partnerships
OpenAI has licensing deals with major publishers. If your competitor is a large media company or has content syndicated through licensed platforms, they have an unfair advantage.
What you can do: You can't buy a content deal, but you can be cited by sources that are cited. Create content that authoritative sites will reference.
Reason 2: They Have Topical Authority
ChatGPT looks for sources that cover topics comprehensively. If your competitor has 20 articles about "project management" and you have 2, they win.
| Your Competitor | You | |-----------------|-----| | 20 articles on topic | 2 articles on topic | | Internal linking between them | Isolated content | | FAQ sections | No FAQs | | Updated in last 30 days | Updated 6 months ago |
What you can do: Build topic clusters. Create 5-10 interlinked articles covering different angles of your expertise.
Reason 3: Their Content Is "Extractable"
AI systems prefer content that's easy to extract and quote:
Easy to extract:
- Clear headers as questions ("How much does X cost?")
- Tables with specific data
- Numbered lists with actionable steps
- First paragraph directly answers the main question
Hard to extract:
- Long paragraphs without structure
- Marketing fluff without data
- Content that requires reading the whole page to understand
Reason 4: They Allow AI Crawlers
Check your robots.txt. Are you blocking GPTBot, OAI-SearchBot, or PerplexityBot?
# Check if you're blocking AI crawlers
User-agent: GPTBot
Disallow: / # This blocks ChatGPT from training
User-agent: OAI-SearchBot
Disallow: / # This blocks ChatGPT Search citations
If you block OAI-SearchBot, you won't appear in ChatGPT Search results. See our robots.txt guide for AI.
Reason 5: They Have E-E-A-T Signals
Google's E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) matters for AI citations too:
| E-E-A-T Signal | Your Competitor | You | |----------------|-----------------|-----| | Named author with bio | Yes | "Admin" or no author | | Author credentials visible | LinkedIn, credentials | Nothing | | Schema markup (Person, Author) | Implemented | Missing | | First-person experience | "We tested X" | "X is said to be..." |
What Does the Data Actually Show?
Here's the breakdown from 485,000+ ChatGPT citations:
Citation Distribution by Domain Type
| Domain Type | % of Citations | |-------------|---------------| | .com (commercial) | 80%+ | | .org (non-profit) | 11.29% | | .edu (educational) | ~3% | | .gov (government) | ~2% | | Other | ~4% |
Takeaway: Commercial websites dominate. You don't need to be a .edu or .gov to get cited. But you need to match the quality signals they have.
Time Sensitivity Analysis
| Query Type | Preferred Source Age | |------------|---------------------| | "Best X of 2025" | Last 3-6 months | | "Latest X news" | Last 30 days | | "How does X work" | Any age if authoritative | | "X vs Y comparison" | Last 12 months |
How Do I Fix This? (Actionable Checklist)
Week 1: Technical Fixes
[ ] Update robots.txt to allow OAI-SearchBot and GPTBot
[ ] Add llms.txt file (see /blog/implementar-llms-txt-guia-practica)
[ ] Verify Schema.org markup for articles and authors
[ ] Check site speed (AI crawlers have short timeouts)
Week 2: Content Structure
[ ] Reformat H2/H3 as questions ("How much does X cost?")
[ ] Add a table with specific data to every key page
[ ] Write first paragraph to directly answer the main question
[ ] Add FAQ section with 4-6 conversational questions
Week 3: Build Topical Authority
[ ] Identify 3 topics where you want to be THE source
[ ] Create 5+ interlinked articles per topic
[ ] Update all content with current year and fresh data
[ ] Add unique data (your own benchmarks, case studies, stats)
Week 4: E-E-A-T Signals
[ ] Add real author name and photo to every article
[ ] Create author bio page with credentials
[ ] Implement Person schema for authors
[ ] Add first-person experience ("We tested this...")
How Can I Test If It's Working?
You can't directly query ChatGPT's index, but you can:
- Ask ChatGPT directly: "What does [your brand] do?" or "Who is the best [your service] provider?"
- Use Perplexity: It shows sources explicitly. Search for your topics.
- Monitor brand mentions: Set up alerts for your brand in AI platforms.
- Track referral traffic: Check analytics for traffic from "chatgpt.com" or "perplexity.ai".
Conclusion: It's Not Magic, It's Engineering
Your competitors appear in ChatGPT because they:
- Have content that's easy to extract (structured, factual, specific)
- Cover topics comprehensively (topical authority)
- Allow AI crawlers (robots.txt configured correctly)
- Show E-E-A-T signals (real authors, credentials, experience)
- Keep content fresh (updated with current data)
This isn't SEO voodoo. It's technical optimization with measurable causes and effects.
How Can Nandark Help?
We help businesses become visible to AI systems. Our GEO (Generative Engine Optimization) service includes:
- Technical audit: robots.txt, llms.txt, Schema markup, crawlability
- Content restructuring: Headers as questions, tables, FAQs, extractable format
- Topical authority building: Content clusters, internal linking, freshness strategy
- Monitoring: Track your visibility in ChatGPT, Perplexity, and Google AI Overviews
Related Services
- SEO & GEO Optimization: Technical optimization for Google and AI visibility
- Web Development: Build sites optimized for AI from day one
Book a free consultation: We'll analyze why your competitors appear in ChatGPT and you don't.
