The top sources ChatGPT cites for its answers
When ChatGPT answers with web search, who does it actually quote? We analysed its citations across a buyer-intent prompt panel. Wikipedia leads, Reddit is close behind, and only ~13% of citations point to brand-owned sites — here's the full breakdown and how to get cited.
Wikipedia
ChatGPT's #1 cited domain (~9% of citations)
#2 — community content is heavily cited
13%
of citations point to brand-owned sites
9
source types analysed across the prompt panel
Why it matters
ChatGPT's answer is only as good as the sources it trusts
With web search, ChatGPT retrieves a set of pages, then synthesises an answer and cites a handful of them. The domains it pulls from are remarkably consistent — and they're mostly not brand websites. Knowing which sources it favours tells you exactly where to earn presence to become part of the answer.
The breakdown
Where ChatGPT's citations come from, by source type
Share of all ChatGPT citations in our sample, grouped by the kind of source. Reference and community content lead; brand-owned sites sit fourth.
Share of ChatGPT citations · by source type
Source: Indexly ChatGPT citation panel, Apr–Jun 2026
Wikipedia dominates this band
Reddit, Quora, Stack Overflow
Editorial outlets ChatGPT trusts
Your own site — only ~1 in 8 citations
G2, Capterra, NerdWallet, listicles
GitHub, product docs, MDN
.gov / .edu / research
YouTube transcripts & pages
LinkedIn, X
The leaderboard
The 10 domains ChatGPT cites most
Individual domains by share of all ChatGPT citations. Two sources — Wikipedia and Reddit — account for roughly one in six citations between them.
- 1
wikipedia.org
Reference
9.4% - 2
reddit.com
Community
7.1% - 3
youtube.com
Video
4% - 4
github.com
Docs & dev
3.3% - 5
g2.com
Reviews
2.6% - 6
forbes.com
News & media
2.4% - 7
medium.com
Editorial
2.1% - 8
linkedin.com
Social
1.9% - 9
stackoverflow.com
Docs & dev
1.8% - 10
quora.com
Community
1.5%
Shares are of total observed citations. The long tail (everything outside the top 10) makes up the remainder — ChatGPT still cites thousands of distinct domains, but concentrates on a familiar core.
Headline findings
Six things the data tells you
Wikipedia is ChatGPT's #1 source
A single encyclopedic domain accounts for ~9% of all citations. ChatGPT leans on neutral, well-structured reference content to anchor answers and entities.
Community content punches above its weight
Reddit, Quora and Stack Overflow together form the largest category after reference. For opinion, comparison and 'is X worth it' questions, UGC is cited heavily.
Third-party beats your own site
Only ~13% of citations point to brand-owned domains. Most of the time ChatGPT describes you using someone else's page — so off-domain presence matters as much as your site.
Authority and editorial trust win
News, reviews and high-DR editorial sources are over-represented. ChatGPT favors sources with established authority and clear, factual structure.
Freshness matters once search runs
When ChatGPT triggers web search, recently updated pages are cited more often. Stale content quietly drops out of the answer.
Crawlable + extractable or invisible
Pages that block OAI-SearchBot/GPTBot, or bury facts in scripts and images, rarely get cited. Clear lists, tables, definitions and schema do.
What to do about it
How to get your brand cited by ChatGPT
- 1
Earn presence on the sources ChatGPT trusts
Get accurate, current coverage on Wikipedia (where eligible), Reddit, review sites (G2, Capterra) and the editorial outlets in your category — these are the pages ChatGPT quotes.
- 2
Make your own pages extractable
Lead with clear definitions, add comparison tables, FAQs and schema markup. ChatGPT cites pages it can lift a clean, factual passage from.
- 3
Build third-party citations and mentions
Consistent, factual mentions across trusted domains raise your entity authority — the signal that gets you named even when your own site isn't cited.
- 4
Keep content fresh
Update stats, dates and product details regularly. Once ChatGPT search runs, freshness is a tiebreaker between competing sources.
- 5
Confirm AI crawlers can reach you
Allow OAI-SearchBot and GPTBot in robots.txt, add an llms.txt, and avoid blanket AI-bot blocks — citation starts at crawl access.
You don't win ChatGPT citations from your homepage alone. You win them on the sources ChatGPT already trusts — then by making your own pages clean enough to quote.
Methodology & sources
How we measured this
Indexly ran a fixed panel of buyer-intent prompts through ChatGPT with web search enabled, captured the cited sources from each answer, normalised domains and classified every citation by source type, between April and June 2026. Shares are reported as a percentage of total observed citations. Figures are directional benchmarks to guide GEO strategy, not audited measurements — ChatGPT's sources shift as its index and ranking change, and only answers that trigger web search produce visible citations.
FAQ
What ChatGPT cites, answered
What sources does ChatGPT cite most?
Across our prompt panel, ChatGPT's citations skew toward reference and encyclopedic content (led by Wikipedia), community and forums (Reddit, Quora, Stack Overflow), news and media, and review/comparison sites. Brand-owned official sites account for only about one in eight citations — most of the time ChatGPT describes a brand using a third-party page.
Does ChatGPT cite Wikipedia?
Yes — heavily. Wikipedia is consistently the single most-cited domain, making up roughly 9% of all citations in our sample. Its neutral tone, structured facts and strong entity coverage make it an easy, safe source for ChatGPT to anchor answers on.
Does ChatGPT cite Reddit?
Yes. Reddit is the second most-cited domain and the backbone of the 'community & forums' category. For subjective, comparison and recommendation questions, ChatGPT frequently pulls from Reddit threads — which is why an unmanaged Reddit presence can shape how AI describes your brand.
Why doesn't ChatGPT cite my brand's website?
Brand-owned domains receive only about 13% of citations. ChatGPT tends to prefer neutral, third-party and high-authority sources for factual claims. To be cited from your own site, make pages factual and extractable (definitions, tables, schema), keep them fresh, and ensure AI crawlers can access them.
How do I get my content cited by ChatGPT?
Win on two fronts: earn presence on the third-party sources ChatGPT already trusts (Wikipedia, Reddit, review sites, editorial outlets), and make your own pages easy to cite — clear definitions, comparison tables, FAQs, schema, fresh data, and crawlable to OAI-SearchBot and GPTBot.
Is this based on ChatGPT's training data or live web search?
This analysis is about visible citations, which appear when ChatGPT runs live web search (ChatGPT search / browsing). Training-data 'parametric' knowledge isn't cited with links, so it isn't measurable here. As more answers trigger search, these citation patterns increasingly determine who gets named.
How was this measured?
Indexly ran a fixed panel of buyer-intent prompts through ChatGPT with web search enabled, captured the cited sources from each answer, normalised domains and classified each citation by source type, between April and June 2026. Shares are directional benchmarks to guide GEO strategy, not audited measurements — ChatGPT's sources shift as its index and ranking change.
See whether ChatGPT cites you — and who it cites instead
Indexly tracks your citations across ChatGPT, Perplexity, Gemini, Grok and Google AI Overviews — which prompts name you, which cite a competitor, and exactly where to earn the citations you're missing.
More like this? See all Indexly Insights.