Top 10 domains AI search engines cite most in 2026

Reddit, YouTube, LinkedIn and Wikipedia supply a large share of what ChatGPT, Perplexity and Google's AI answers cite. The ranking changes by engine, and it can change in a week.

Rows of server racks with blue status lights in a data centre server room
Photo: Carl Lender from Sunrise, USA, CC BY 2.0, via Wikimedia Commons

The most cited domains in AI search in 2026 are Reddit, YouTube, LinkedIn, Wikipedia, Forbes, G2, Yelp, Facebook, Medium and TechRadar, in that order, according to Peec AI's March 2026 analysis of 30 million sources cited by ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews in the US. Reddit and YouTube are the only two domains in every engine's top five; ChatGPT leans on Wikipedia, Google's engines on YouTube and social sites, and Perplexity on Reddit.

Key takeaways
  • Reddit and YouTube are the only domains in every engine's top five (Peec, March 2026).
  • Wikipedia leads ChatGPT: 13.15% of its US citations in Similarweb's early-2026 count.
  • YouTube is first on both Google AI surfaces, AI Mode and AI Overviews.
  • ~60% → ~10%: ChatGPT answers citing Reddit, August to mid-September 2025. Rankings move fast.
  • 74.5%: share of citations that actually supported their sentence in a 2023 audit.

The ranking criterion is simple: how often a domain is directly cited in AI answers, counted across large samples of prompts. I use Peec's cross-engine list for the top 10 because it is the newest and widest, and add per-engine numbers from three more studies:

StudySamplePeriod
Peec AI30 million cited sources, five enginesMarch 2026
Profound680 million citationsAug 2024 to Jun 2025
Semrush100 million citationsJul to Oct 2025
SimilarwebChatGPT web browsing, USJan to Feb 2026
The story

Around May 2025, the Wikimedia Foundation saw unusually high amounts of apparently human traffic to Wikipedia, mostly from Brazil. When it updated its bot detection, much of the surge turned out to be bots built to evade detection. After reclassifying the March to August data, the Foundation's Marshall Miller reported on 17 October 2025 that human pageviews were down roughly 8% on the same months of 2024.

His explanation: search engines increasingly use generative AI to answer people directly, and younger readers look things up on social video. Yet almost all large language models train on Wikipedia, and search and social platforms keep surfacing its content. The site sits near the top of AI citation lists while fewer people click through, which is why the Foundation is asking AI platforms to attribute it clearly and send readers back.

Wikimedia Foundation, Diff: New user trends on Wikipedia (17 October 2025)
30 millioncited sources in Peec's March 2026 study
7.8%of all ChatGPT citations went to Wikipedia (Profound, 2024-25)
6.6%of all Perplexity citations went to Reddit (Profound)
~60% to ~10%ChatGPT answers citing Reddit, Aug to Sep 2025 (Semrush)

Which domains do AI search engines cite most?

Across five engines, Peec ranks Reddit first and YouTube second, followed by LinkedIn and Wikipedia. The engine-by-engine tops differ more than the overall list suggests.

EngineTop 5 cited domains (Peec, March 2026)Leaning
ChatGPTWikipedia, Reddit, Forbes, TechRadar, LinkedInEncyclopaedic and editorial
PerplexityReddit, YouTube, LinkedIn, Wikipedia, G2Community and reviews
GeminiReddit, YouTube, Wikipedia, Medium, PCMag / ForbesMixed
Google AI ModeYouTube, Reddit, Facebook, LinkedIn, YelpVideo and social
Google AI OverviewsYouTube, Reddit, Facebook, LinkedIn, MediumVideo and social
Share of all citations by domain and engine, Aug 2024 - Jun 2025
See the data as a table
ChatGPT (%)Google AI Overviews (%)Perplexity (%)
Wikipedia7.80.6
Reddit1.82.26.6
YouTube1.92.0
Quora1.5
LinkedIn1.30.8
Forbes1.10.60.7
G21.10.6
Gartner0.71.0
NerdWallet0.80.60.6
Yelp0.8

Source: Profound, AI platform citation patterns, 680 million citations · Gaps mean the domain was outside that engine's top 10 in the study.

Rows of Google bicycles parked on Charleston Road at the Google campus in Mountain View
Google's campus in Mountain View. Its two AI surfaces, AI Mode and AI Overviews, put YouTube, a Google property, at the top of their citations.Photo: Dietmar Rabich, CC BY-SA 4.0, via Wikimedia Commons

1. Reddit

Reddit is the most cited domain across AI search in Peec's 2026 data, first or second on every engine it tested. Profound found it was the single largest source for Perplexity (6.6% of all citations) and Google AI Overviews (2.2%), and second for ChatGPT.

It is also the most volatile. Semrush tracked 230,000 prompts weekly and saw ChatGPT cite Reddit in close to 60% of responses in early August 2025, collapsing to around 10% by mid-September; Reddit still remained one of ChatGPT's two most cited domains afterward.

Answer engines value Reddit for first-hand experience in people's own words, which is exactly what brand pages lack. The only durable way in is genuine participation in the subreddits your buyers read.

  • Peec rank, all engines: 1
  • Perplexity share (Profound): 6.6% of all citations
  • ChatGPT, Similarweb Jan-Feb 2026: 11.97% of citations

2. YouTube

YouTube is second overall and first on both of Google's AI surfaces, AI Mode and AI Overviews, in Peec's data; Profound measured it at 1.9% of AI Overview citations and 2.0% of Perplexity's.

Peec found AI answers pull from both video descriptions and auto-generated transcripts, including Shorts.

  • Peec rank: 2; first on AI Mode and AI Overviews
  • AI Overviews share (Profound): 1.9%

3. LinkedIn

LinkedIn is third overall and appears in four of the five engines' top fives. Semrush found Google's AI Mode cited LinkedIn in nearly 15% of responses through mid-2025, and that LinkedIn and Forbes rose on ChatGPT after its September 2025 change.

For B2B brands this is the most controllable entry on the list: posts and articles by named people at your company are citable sources.

  • Peec rank: 3
  • AI Mode responses citing it (Semrush, 2025): nearly 15%

4. Wikipedia

Wikipedia is ChatGPT's top source: 7.8% of all ChatGPT citations in Profound's data, 47.9% of the citations going to its top 10 domains, and 13.15% in Similarweb's January to February 2026 count. It is absent from the top five of Google's AI Mode and AI Overviews.

Peec notes Wikipedia works twice: models cite it live and also learned from it in training. A Wikipedia page is earned through notability and independent coverage, not written by the brand.

  • ChatGPT share of all citations: 7.8% (Profound), 13.15% (Similarweb, 2026)
  • Semrush: from about 55% to under 20% of ChatGPT responses, Aug-Sep 2025

5. Forbes

Forbes is the top editorial publisher on the list and gained after ChatGPT's September 2025 change; Semrush found it doubled its number of cited ChatGPT responses after September. Profound measured it at 1.1% of ChatGPT citations.

Contributed and staff articles both get cited, which is why "featured in Forbes" placements now have a second life in AI answers.

6. G2

G2, the software review site, is sixth overall and appears in Perplexity's top five only; Profound measured 1.1% of ChatGPT citations and 0.6% of Perplexity's. For SaaS companies, verified reviews on G2 are among the most direct routes into "best tool for X" answers.

7. Yelp

Yelp ranks seventh, driven by local and service queries on Google AI Mode and Perplexity (0.8% of Perplexity citations in Profound's data).

In India, where Yelp is little used, the equivalent signal is likely to be Google Business Profile reviews and local directories, though none of these US-based studies measured that.

8. Facebook

Facebook is eighth, in the top five for both Google AI surfaces. Peec notes Facebook blocks Perplexity's crawlers in its robots.txt, so it does nothing for that engine, but public group posts and page posts are cited by Google's.

9. Medium

Medium is ninth, in the top five for AI Overviews and Gemini, and among ChatGPT's biggest gainers after September 2025 in Semrush's data. Its mix of long, specific how-to posts suits retrieval.

10. TechRadar

TechRadar closes the top 10, in ChatGPT's top five for product and comparison queries; Profound measured 0.9% of ChatGPT citations. Like Forbes it lost some ChatGPT share in September 2025, which shows that even favoured publishers are reweighted without notice.

The top 10 at a glance

RankDomainWhere it is strongestWhat a brand can do
1RedditPerplexity, all enginesDisclosed, useful participation
2YouTubeGoogle AI Mode and AI OverviewsVideo with clean transcripts
3LinkedInFour of five enginesPosts by named people
4WikipediaChatGPTEarn notability; never self-edit
5ForbesChatGPTEarned editorial coverage
6G2PerplexityVerified software reviews
7YelpAI Mode, Perplexity (local)Local reviews
8FacebookGoogle AI surfacesPublic page and group posts
9MediumAI Overviews, GeminiSpecific how-to posts
10TechRadarChatGPT (product queries)Product reviews and comparisons

Why do ChatGPT, Perplexity and Google cite different sites?

Because each retrieves from a different index with different rules:

  • ChatGPT search leans on encyclopaedic and editorial sources.
  • Perplexity leans on community discussion.
  • Google's AI features lean on Google's own properties and social platforms; Semrush also found AI Mode shows a clear preference for Google-owned domains such as YouTube.

The clearest academic evidence on how concentrated those choices are comes from a 2025 study of news citations, and an earlier audit of whether citations actually support the answer.

News Source Citing Patterns in AI Search SystemsKai-Cheng Yang · arXiv 2507.05301 · 2025

Using the AI Search Arena dataset of 24,069 conversations from March to May 2025, the study extracted 366,087 citations from 12 AI search models by OpenAI, Perplexity and Google. Only 9% of citations went to news outlets. Citation was highly concentrated: the top 20 sources took 67.3% of OpenAI models' citations, against 31.9% for Google and 28.5% for Perplexity. Among news citations, OpenAI's models leaned heavily on Reuters (22.8%) and AP (12.2%); Google's top news source was indiatimes.com (4.3%), and Perplexity's was BBC (3.2%). The limit: a three-month snapshot of black-box systems, covering news only.

Read the paper (free)
Evaluating Verifiability in Generative Search EnginesNelson F. Liu, Tianyi Zhang, Percy Liang · Findings of EMNLP 2023 (arXiv 2304.09848)

The authors audited answers from four generative search engines of the time (Bing Chat, NeevaAI, Perplexity and YouChat) against the sources they cited. Only 51.5% of generated sentences were fully supported by their citations, and only 74.5% of citations actually supported the sentence they were attached to. The limit: the engines tested have since changed, and two no longer exist, but the finding explains why being cited is not the same as being quoted accurately.

Read the paper (free)
UserquestionQuery rewriteinto searchesSearch indexReddit, YouTubeWikipedia, pressAnswer withcitations
Answer engines rewrite a question into several searches, retrieve candidate pages, and cite the ones the model drew on. Which index and which sites they favour differs by engine.

How often do AI citation rankings change?

Often enough that a ranking older than a quarter should be rechecked. The sharpest recorded shift came in mid-September 2025, when ChatGPT's citations of Reddit fell from about 60% of responses to about 10%, and of Wikipedia from about 55% to under 20%, within weeks.

ChatGPT responses citing a domain, 2025 ~60%~10%~55%<20% Early AugMid-SepEarly AugMid-Sep RedditWikipedia
The September 2025 reset, drawn from Semrush's weekly tracking of 230,000 prompts. Values are approximate, as Semrush reported them.

What caused the drop?

Some SEOs linked the drop to Google removing its num=100 results parameter around 11 September 2025, which limited deeper result pages to outside tools.

Semrush's own analysis doubted that was the whole cause, noting only about 34% of Reddit's Google rankings sit in positions 21 to 100, and suggested OpenAI deliberately reduced over-reliance on a few sites. The winners on ChatGPT were PR Newswire, Forbes and Medium.

Do not build an AI-visibility plan on one domain. Even favoured sites are reweighted without notice.

How engines choose sources is explained in how AI Overviews and ChatGPT choose sources, and how the old SEO playbook led here is in the history of SEO.

ChatGPT citation share by domain, US, Jan-Feb 2026
See the data as a table
Share of citations (%)
wikipedia.org13.15
reddit.com11.97
openai.com6.21
walmart.com2.9
youtube.com2.67
linkedin.com2.42
reuters.com2.27
nih.gov2.22
google.com2.17
forbes.com1.38

Source: Similarweb AI Citation Analysis, ChatGPT web browsing, US

One detail in the Similarweb data deserves attention: openai.com was ChatGPT's third most cited domain, ahead of Reuters and LinkedIn. An answer engine citing its own publisher is a reminder that these are commercial products, not neutral libraries.

A Wikipedia advertisement on a digital billboard in Times Square, New York
A Wikipedia billboard in Times Square. Wikipedia is ChatGPT's single most cited domain, and it reaches models twice: through live retrieval and through training data.Photo: Kevani for Wikimedia Foundation, CC BY-SA 4.0, via Wikimedia Commons

What should a brand do with this list?

Be present, honestly, where the engines already look, and make your own pages easy to quote. The two Reddit threads in this post's forum list ask exactly this, and the consistent answer from practitioners is that third-party mentions move AI visibility more than on-site tweaks.

  1. Map your queries to engines. If your buyers use ChatGPT, Wikipedia and editorial coverage matter; if they use Google, YouTube and social matter more.
  2. Answer in the communities. Useful, disclosed participation on Reddit and LinkedIn is citable; astroturfing gets removed, and Reddit's weight on ChatGPT has already been cut once.
  3. Publish video with clean transcripts. YouTube is first on Google's AI surfaces, and transcripts are read.
  4. Earn reviews where your category is judged. G2 for software, Yelp or local equivalents for services.
  5. Write quotable pages. Specific numbers, named sources and direct answers, the edits the GEO study found lift visibility by up to 40%. The full method is in how to get cited by ChatGPT and Perplexity, and evidence on what works in does AEO/GEO work.

For more on how AI is changing marketing work, see the applied AI page on this site.

Questions people ask

What sources does ChatGPT cite most?

Wikipedia and Reddit, by a wide margin. Similarweb's January to February 2026 count put Wikipedia at 13.15% and Reddit at 11.97% of ChatGPT's citations in the US, followed by openai.com at 6.21%.

Why does Perplexity cite Reddit so much?

Perplexity weights community discussion heavily: Reddit was 6.6% of all its citations and 46.7% of the citations going to its top 10 domains in Profound's 2024 to 2025 data. Reddit offers first-hand answers to exactly the specific questions people ask answer engines.

Does Google AI Overviews cite Wikipedia?

Less than other engines do. Profound measured Wikipedia at 0.6% of AI Overview citations, and Peec's 2026 data has it outside the top five for both AI Overviews and AI Mode, which favour YouTube, Reddit and social platforms.

Can a brand pay to be cited by AI search engines?

Not directly in organic answers. Brands influence citations by earning mentions on sites the engines trust and publishing pages that are easy to quote. AI answers also cite sources inaccurately: one audit found only 74.5% of citations supported the sentence they were attached to.

Sources

  1. Peec AI: Top domains cited by AI search, analysis of 30M sources (2026)
  2. Peec AI: Most-cited domains by engine (2026)
  3. Profound: AI platform citation patterns (2025)
  4. Semrush: The most-cited domains in AI, a 3-month study (2025)
  5. Similarweb: The most cited domains by LLMs (2026)
  6. Yang: News Source Citing Patterns in AI Search Systems, arXiv (2025)
  7. Liu, Zhang and Liang: Evaluating Verifiability in Generative Search Engines, arXiv (2023)
  8. Aggarwal et al.: GEO, Generative Engine Optimization (2024)
  9. Pew Research Center: AI summaries and clicks (2025)
  10. exchange4media: The ad that never loaded (2026)
  11. Wikimedia Foundation, Diff: New user trends on Wikipedia (October 2025, source for the story)
AI citationsChatGPTPerplexityGoogle AI OverviewsAI ModeRedditWikipediaGEO
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Marketblog.

Keep reading in Marketblog

From the other blogs