The Gist 100Analysis

Each AI assistant cites a different part of the web

ChatGPT cites 3.3 different websites per answer and Claude cites 12.7. Brand sites are 55 percent of ChatGPT’s citations and a far smaller share for the others. The work of getting named looks different for each assistant.

websites cited per answer, ChatGPT and Claude
3.3 vs 12.7
of ChatGPT’s citations are brand sites
55%
of Google AI Overviews answers cite YouTube
49%
ByGist GEO team
PublishedOct 6, 2026
UpdatedOct 6, 2026
Reading time4 min
TopicsThe Gist 100, Sources, Assistants

Published by Gist GEO · About our content

Ask ChatGPT, Perplexity, Claude and Google AI Overviews the same buying question and each cites a different slice of the web. Gist counted the websites cited in 9,539 AI answers from October 2026. ChatGPT cites 3.3 different websites per answer, Google AI Overviews 5.7, Perplexity 8.9 and Claude 12.7.

The kind of site differs as much as the count. Brand sites are 55 percent of ChatGPT’s citations. Publishers are 36 percent of Perplexity’s. Google AI Overviews cites YouTube in 49 percent of its answers, while Perplexity and Claude cite it in 0 percent. Getting named means different work on each assistant.

Websites cited in a typical answer

Average number of different websites cited per answer, all 32 categories

  1. Claude12.7
  2. Perplexity8.9
  3. Google AI Overviews5.7
  4. ChatGPT3.3

Source: The Gist 100, October 2026.

Claude cites the widest range of sites

Almost every answer cites at least one source. The share is 100 percent for ChatGPT, Perplexity and Claude, and 98 percent for Google AI Overviews. So the first difference is how many sites an answer cites. ChatGPT’s answers cite the fewest and Claude’s the most.

The pool differs too. Across all its answers, Claude cites 2,271 different websites, Google AI Overviews 1,693, ChatGPT 755 and Perplexity 513. Perplexity cites 8.9 sites per answer from the smallest pool, and one reason is visible in the counts. In 251 of the 256 questions, Perplexity returned the same list of sources for all ten of its answers, though it worded each answer differently. Claude pairs the most sites per answer with the largest pool.

Each assistant leans on its own sites

Brand sites are 55 percent of ChatGPT’s citations but 9 percent of those in Google AI Overviews, a gap of 46 points. Perplexity leans on publishers, at 36 percent against 6 percent for ChatGPT. Google AI Overviews cites forums and video at 10 and 9 percent of its citations, against 1 percent or less for ChatGPT and Claude. Review sites are the common ground, at 15 to 23 percent of citations for every assistant.

Claude’s citations spread across kinds. None of its classified kinds tops 17 percent, and 53 percent of its citations are Not labelled. That share is the long tail of sites Gist did not classify, so each mix describes only the classified part of an assistant’s citations. The caveat matters most for Claude and, at 37 percent, Google AI Overviews.

The most-cited sites follow the same split. ChatGPT cites TechRadar most, in 13 percent of its answers, then Tom’s Guide in 8. No assistant has a lower top-site share. Perplexity cites Forbes in 42 percent of its answers, The New York Times in 31 and G2 in 27.

Google AI Overviews cites YouTube in 49 percent and Reddit in 35, then Forbes and The New York Times in 13 each. Claude cites Forbes in 27 percent, Tom’s Guide in 22 and TechRadar in 21. No site is among the eight most-cited for all four assistants, and only Forbes is among them for three.

A site’s score on the full list of most-cited sites averages the four assistants, which hides these splits. YouTube and TechRadar both score 13. Google AI Overviews cites YouTube in 49 percent of its answers, and the other three cite it in 1 percent or less. Every assistant cites TechRadar, in 4 to 21 percent of its answers.

Percent of each assistant’s answers that cite the site at least once

SiteChatGPTPerplexityClaudeGoogle AI Overviews
Forbes1422713
TechRadar1315214
YouTube10049
G2227144
The New York Times031013
Tom’s Guide812220
Reddit32035

Business software and consumer cite different sites

The 32 categories fall into 16 business software and 16 consumer, and the lists diverge. G2, a review site, leads business software at 23 percent of answers. Forbes, a publisher, leads consumer at 25, ahead of The New York Times at 20 and Tom’s Guide at 19. Of each group’s 15 most-cited sites, only four are on both lists: Forbes, TechRadar, YouTube and Reddit.

Four of the 15 business software sites are brand sites: Guideflow, Shopify, Airwallex and Rework. None of the 15 consumer sites is a brand site. The assistants still disagree inside each group. Perplexity cites G2 in 54 percent of its business software answers and ChatGPT cites it in 4. On the consumer side, Perplexity cites The New York Times in 57 percent of its answers and ChatGPT and Claude cite it in 0.

What to do with this

  1. Make your own site the clearest source for ChatGPT. Brand sites are 55 percent of its citations, and it cites only 3.3 different websites per answer.
  2. Check how publishers and review sites cover you for Perplexity and Claude. Forbes, G2 and NBC News are among the eight most-cited sites for both, and Perplexity cites Forbes in 42 percent of its answers, Claude in 27.
  3. Check what YouTube and Reddit show about you for Google AI Overviews. It cites them in 49 and 35 percent of its answers. Perplexity cites them in 0 and 2.

Counted from the answers behind the October 2026 Gist 100. How it is made.

Sources

  1. The Gist 100: how it is made. Gist GEO

Questions marketers ask

Which assistant cites the widest range of websites?

Claude. It cites 12.7 different websites per answer and 2,271 in total, the most of the four. Perplexity cites 8.9 per answer but only 513 in total.

What does the “Not labelled” share mean?

It is the long tail of cited sites that Gist did not classify, so each assistant’s mix describes only the classified part of its citations. The share runs from 17 percent for ChatGPT to 53 percent for Claude.

Keep reading

All analysis
  • Case study · 4 min read

    Ramp is in almost every AI answer on spend management

    Ramp is in 281 of 285 AI answers about spend management software and comes first in 84%. Answers cite its own site in 64%, but Google AI Overviews names Ramp in 96% while citing ramp.com in 2%. The counts show what the answers hold, not why.

    84%of AI answers name Ramp first

  • Analysis · 4 min read

    AI names the market leader first in two of six categories

    Outside sources name the market leader in six categories. AI answers name that brand first in two. In the other four it ranks second to sixth, 14 to 41 points behind the most-named brand, so market size and AI standing need separate scoreboards.

    2 of 6Categories where AI names the market leader first

  • Case study · 4 min read

    HubSpot is named ahead of the biggest CRM vendor

    The largest CRM vendor by market share is not the brand the answers name most. HubSpot scores 99 and Salesforce 83. Assistants name HubSpot first in 49% of answers and Salesforce first in 26%. The gap sits in four of eight questions, and the lead stays in CRM.

    99 vs 83Share of CRM answers naming HubSpot vs Salesforce

Not on the list yet? Check your own brand.

See which brands appear in AI recommendations for your category, and where your brand stands among them.

Check my brand, free