Do AI search engines cite the same websites?
No. Across 5.31 million citations, the five major AI engines agree on only 9% of their most-cited domains. Vercite captured every citation from ChatGPT, Perplexity, Gemini, Google AI Overviews and Google AI Mode, then classified 158,847 domains by source type and publisher. The result: there is no single “AI consensus” on who to trust. Each engine pulls from a different slice of the web, with its own bias toward social platforms, retail, local sites, or, in Google’s case, Google’s own products.
Which websites does each AI engine cite most?
Each engine leans on a different source. ChatGPT cites Reddit more than anything else, Perplexity and AI Mode cite YouTube, and AI Overview leans on Google itself. Below, every engine profiled by citation volume, how many distinct sites it reached, the single site it leans on hardest, and how often it cites Google’s own products.
Do ChatGPT, Gemini and Perplexity cite the same sources?
Barely. To measure agreement we take each engine’s 100 most-cited domains, pool all five lists, and get 253 distinct domains. Only 23 of those (9%) appear in all five engines’ top 100. More than half show up for a single engine and no other. (The 9% is overlap among each engine’s top sources — not of the 158,847 total domains, most of which are long-tail pages cited once.)
Only 23 of the 253 domains (9%) are cited by all five engines — the closest thing to a shared AI “trust list.” It isn’t made of big media brands, but of a narrow set of local institutional sites — banks, property portals, government services — plus a handful of global anchors like Reddit and Wikipedia.
Do AI engines rely on a few big sites or a long tail?
It depends entirely on the engine. Google AI Mode draws half its citations from just 71 sites; ChatGPT spreads the same half across 712. To read it: rank every site an engine cites from most-cited to least, then count down until you have accounted for half of all its citations. The fewer sites it takes, the more winner-takes-all that engine is.
How much do AI engines rely on Reddit, YouTube and Wikipedia?
It varies sharply, and each engine leans on a different one. Google AI Mode leans hardest, sending 5.7% of its citations to the three (mostly YouTube). ChatGPT’s pick is Reddit, Perplexity’s is YouTube, and AI Overview leans on both. Gemini barely touches any of them (1.4%). Together these three user-generated and reference platforms are a citation channel in their own right, and which one matters depends entirely on the engine.
What kind of sources does each AI engine trust?
Set aside what a site sells and look at its character: is it a forum, an official body, a news outlet, a brand’s own page? Each engine has a signature. AI Overview leans hardest on tech and SaaS platforms; AI Mode on social video, official sources and brand-owned pages; ChatGPT on tech and news; Perplexity on government and academic sources; Gemini on reviews and comparisons. The grid shows each engine’s share of citations by source type, after setting aside the generic blog/niche bucket. The brightest cell in each column is that engine’s tell.
| Publisher type | Perplexity | ChatGPT | AI Overview | Gemini | AI Mode |
|---|---|---|---|---|---|
| Social / Video / UGC | 2.7 | 2.4 | 5.8 | 1.2 | 7.8 |
| SaaS / Tech platform | 1.9 | 3.8 | 10.0 | 2.7 | 5.6 |
| Government / Official | 3.3 | 2.0 | 2.2 | 3.1 | 4.9 |
| News / Media | 1.6 | 2.1 | 1.4 | 1.1 | 1.5 |
| Review / Comparison | 2.0 | 1.2 | 1.5 | 2.9 | 1.0 |
| Corporate / Brand-owned | 0.3 | 0.3 | 0.7 | 0.1 | 2.1 |
| Academic / Research | 1.1 | 0.5 | 0.2 | 0.5 | 0.2 |
| Reference / Wiki | 0.3 | 0.9 | 0.3 | 0.4 | 0.3 |
When Google’s AI cites Google, which part is it citing?
It depends which Google engine. AI Overview overwhelmingly cites its own Search results: 79% of its Google citations, and 6.7% of everything it cites, point back to a google.com search page. AI Mode is different: its Google citations are mostly Translate and Maps, not Search. And Gemini, Google’s chatbot, barely cites Google at all (0.2%). The two Google search products lean on Google itself; the one built as an assistant doesn’t.
How do you get your brand cited across every AI engine?
By treating each engine as its own channel: the same brand has to win five different source landscapes at once. ChatGPT rewards forums and tech sources, Gemini rewards online retailers, AI Mode rewards a narrow club of recognised authorities. Here is what each engine over-indexes on, and four ways to act on it.
- National news outlets
- Academic & research
- Local listings & reviews
- Aggregators & directories
- Community forums (Reddit)
- Developer & mapping tools
- Tech review media
- Reference (Wikipedia)
- Google Search & Translate
- Tech & SaaS platforms
- High-authority commercial sites
- Reddit & YouTube
- Online retailers & shops
- Product & comparison sites
- Consumer & lifestyle media
- Brand-owned sites
- A small club of trusted sites
- Official & institutional sources
- Established national brands
- Recognised authorities, not volume
How did we collect and count this?
A dataset is only as useful as its definitions, so here is exactly what we counted and how.
Every source an engine attaches to its answer. Some are cited directly in the response; others come from the web search the engine runs to build that answer. We count each source once and do not weight by how prominently it is shown, so the figures measure whether a domain is surfaced as a source, not how visibly. Redirect and bot artifacts are removed before counting.
These come from real Vercite user data, so they span a broad mix of languages, industries and query types rather than a controlled sample. Because the mix is user-driven, we emphasise structural patterns that hold across topics — cross-engine overlap, source concentration and platform reliance — over industry-by-industry breakdowns.
Collected over four months (March–June 2026) across all five engines.
Try Vercite for free today
Monitor your brand across ChatGPT, Gemini, Perplexity & Google AI