The Claude Citation Source Index 2026 is Everything-PR Research's ranking of the 50 domains Claude cites most when assembling answers. Synthesized from six independent citation-tracking studies covering more than 680 million citations across the major AI engines, plus a Profound dataset of 3.25 billion AI citations and an arXiv analysis of 10,000+ Claude health citations. The Claude citation map looks materially different from ChatGPT's. Claude is the most selective of the major AI engines — its social citation rate is 3.99% versus 15.3% for Google AI Overviews and 11.3% for Perplexity. Only 36% of Claude's journalism citations come from the past 12 months versus 56% for ChatGPT — a structural preference for stable, authoritative, peer-reviewed content over fresh news. The cited sources skew long-form editorial (NYT, The Atlantic, The New Yorker, The Economist), peer-reviewed academic (Nature, arXiv, NIH/PubMed), and institutional (Mayo Clinic, CDC, SEC, Federal Reserve). Median Domain Authority of cited Claude health sources is 92. A separate retrieval reality matters: Claude uses Brave Search as its web-search backend, and Profound found an 86.7% overlap between Claude-cited URLs and Brave's top organic results. Brave visibility is the leading indicator for Claude visibility — Google ranking is not.
3.99% Claude's social citation rate (lowest of major engines) | 97.8% Share of Claude health citations from institutional sources | 92 Median Domain Authority of cited Claude sources | 86.7% Citation overlap between Claude and Brave Search top results |
1. Claude is the authority engine
If ChatGPT is a popularity engine and Perplexity is a community engine, Claude is the authority engine. The Profound, arXiv, and Lantern data converge on the same finding from three different methodologies: Claude cites fewer sources, weights them more heavily, and prefers institutional and peer-reviewed material over social and user-generated content. The social citation rate of 3.99% is not a rounding error — it is roughly a quarter of Google AI Overviews' rate and a third of Perplexity's.
The mechanism is partly architectural. Anthropic emphasizes high-quality editorial training data, and Claude's web retrieval routes through Brave Search rather than Google. Brave's index over-weights documentation-style pages, government sites, and academic publishers — and Claude inherits that bias. The mechanism is partly editorial: Anthropic's published model behavior favors source-backed reasoning over crowd consensus. The downstream effect is that Claude is the AI engine where prestige editorial, institutional research, and peer-reviewed academic publication earn citations at a rate not seen anywhere else.
This matters because Claude is the enterprise leader. Anthropic holds 40% of the enterprise LLM API market and an estimated 54% of the enterprise coding market. For any brand whose buyers are enterprise B2B, regulated industries, healthcare, financial services, government contracting, or technical procurement, Claude visibility is not optional — and Claude visibility runs on a different playbook than ChatGPT visibility.
"Claude does not cite a brand's own description of itself if it cannot find third-party validation elsewhere. It does not cite a summary of a study if the source is accessible. For brands accustomed to running PR through owned media, Claude is the audit."
— Everything-PR Research, June 2026
2. Methodology
This Index is a synthesis of independent third-party citation data, weighted toward Claude-specific datasets. Source studies:
- Profound longitudinal citation study (August 2024 – October 2025): 680 million citations across major AI engines; later expanded to 3.25 billion AI citations with engine-specific social citation rates.
- Profound Brave Search overlap analysis (via Tagliaferro, 2025): 86.7% overlap between Claude-cited URLs and Brave's top organic results for the same queries.
- arXiv Claude health citation analysis: 10,000+ Claude health citations; 97.8% from established institutions; median Domain Authority 92.
- Semrush 325,000-prompt cross-engine study (Q1 2026): Claude's share of LinkedIn citation (5.3%) versus ChatGPT (14.3%) and Google AI Overviews (13.5%).
- Oltre / eSEOspace Claude URL pattern analysis (May 2026): 2,170-URL Claude citation dataset; 56% under /blog/ paths, 47% listicle-style paths, 24% with year tokens in URL.
- Lantern AI Citation Content Visibility Report (February 2026): 200 million+ citations collected directly from ChatGPT, Perplexity, Gemini, and Claude.
Consolidation logic: where studies disagree on rank inside the top 50, this Index favors larger sample size and engine-specific data over multi-engine aggregates. Where studies disagree on attribution (e.g., whether to credit YouTube as 'social' or 'video'), this Index uses the functional category most useful to operator decision-making.
Volatility caveat: Claude is the most stable of the major engines on a quarterly basis. The top 20 is unlikely to shift more than two positions per quarter. The long tail (positions 30–50) reflects active churn as Anthropic's Brave Search backend evolves and as Claude usage shifts toward enterprise queries with different citation patterns. This Index will be reissued quarterly.
This synthesis was produced in coordination with 5W AI Communications, the AI Communications Firm whose AI Visibility practice operationalizes this index for brands, institutions, and category leaders.
3. The 50 domains Claude cites most
| Rank | Source | Category | Lean / Cluster | Share Signal | Note |
|---|
| 1 | Wikipedia | Encyclopedic | Foundational reference | Top-cited across every Claude study | Anthropic training corpus + Brave Search retrieval both surface Wikipedia first. |
| 2 | The New York Times | Editorial — News/Long-form | Stable authority | Top editorial source in 5W consolidated data | Claude leans long-form, source-backed reporting; NYT is the canonical anchor. |
| 3 | The Atlantic | Editorial — Long-form | Over-indexes on Claude | Materially higher Claude share than ChatGPT | Claude's structural preference for long-form analysis is most visible here. |
| 4 | The New Yorker | Editorial — Long-form | Over-indexes on Claude | Recurring top-10 in Claude-specific tracking | Citation depth on long-form essays and reported features. |
| 5 | The Economist | Editorial — Business/Policy | Over-indexes on Claude | Recurring top-10 | Policy and global-business queries route here disproportionately. |
| 6 | Reuters | Editorial — Wire | Stable | Top wire source on time-sensitive queries | 36% of Claude journalism citations are <12 months old; Reuters is the freshness anchor. |
| 7 | Bloomberg | Editorial — Business | Stable | Recurring top-15 | Markets, executives, public-company queries. |
| 8 | The Washington Post | Editorial — News | Stable | Recurring top-15 | Politics, policy, national security. |
| 9 | Financial Times | Editorial — Business | Stable | Higher share than on ChatGPT | Paywalled but extractable lede/headline; global business depth. |
| 10 | The Wall Street Journal | Editorial — Business | Stable (paywalled discounted) | Recurring top-20 | Business/markets default. |
| 11 | Nature | Academic — STM | Over-indexes on Claude | Claude's selective bias toward peer-reviewed sources | Median Domain Authority of cited Claude health sources: 92 (arXiv). |
| 12 | NIH / PubMed | Institutional — Health | Foundational health authority | 97.8% of Claude health citations from institutional sources | Government-grade biomedical research. |
| 13 | arXiv | Academic — STM Preprints | Over-indexes on Claude | Standard Claude reasoning anchor on technical queries | Brave Search surfaces arXiv preprints; Claude cites them directly. |
| 14 | Harvard.edu | Academic | Stable authority | Recurring top-20 | Domain authority signal; institutional credibility. |
| 15 | MIT.edu / MIT Tech Review | Academic + Editorial | Stable | Recurring top-20 | Engineering, AI, science queries. |
| 16 | Stanford.edu | Academic | Stable | Recurring top-20 | Research authority signal. |
| 17 | Harvard Business Review | Editorial — Management | Stable | Recurring top-25 | Management frameworks; Claude favors structured reasoning. |
| 18 | Brookings | Research — Policy | Over-indexes on Claude | Recurring top-25 | Policy, economics, governance queries. |
| 19 | Pew Research | Research — Polling | Stable | Recurring top-25 | Polling, demographics, opinion data. |
| 20 | OECD | Research — Policy | Over-indexes on Claude | Recurring top-30 | Global policy and economic data. |
| 21 | World Bank | Research — Economic | Over-indexes on Claude | Recurring top-30 | Global development and economic indicators. |
| 22 | IMF | Research — Economic | Over-indexes on Claude | Recurring top-30 | Macro and country data. |
| 23 | Statista | Reference — Data | Stable | Recurring top-30 | Numeric anchor; chart and stat extractability. |
| 24 | Britannica | Encyclopedic | Stable secondary reference | Higher share on Claude than on ChatGPT | Where Wikipedia is challenged, Claude back-fills with Britannica. |
| 25 | Reddit | Community / UGC | Sharply under-indexed | Claude social citation rate: 3.99% (lowest of major engines) | Claude cites Reddit far less than ChatGPT, Perplexity, or Google AIO. |
| 26 | BBC | Editorial — News | Stable | Recurring top-35 | Global news authority; archive depth. |
| 27 | The Guardian | Editorial — News | Stable | Recurring top-35 | UK editorial; long-form. |
| 28 | NPR | Editorial — News | Stable | Recurring top-35 | Public-media authority signal. |
| 29 | ProPublica | Editorial — Investigative | Over-indexes on Claude | Recurring top-35 | Investigative depth; original document publication. |
| 30 | Mayo Clinic | Institutional — Health | Foundational consumer-health | Recurring top-35 | Symptom, condition, treatment queries. |
| 31 | CDC.gov | Institutional — Health | Foundational | Recurring top-35 | Public-health authority. |
| 32 | WHO.int | Institutional — Health | Foundational | Recurring top-35 | Global health authority. |
| 33 | SEC.gov | Institutional — Finance | Foundational | Recurring top-40 | Public-company filings; 10-K/10-Q retrieval. |
| 34 | Federal Reserve | Institutional — Finance | Foundational | Recurring top-40 | Monetary policy and economic data. |
| 35 | Gov.uk | Institutional — Government | Stable | Recurring top-40 | UK government information depth. |
| 36 | LinkedIn | Professional | Stable | 5.3% of Claude responses (Semrush) | Cited but at materially lower rate than ChatGPT (14.3%) or AIO (13.5%). |
| 37 | YouTube | Video | Under-indexed vs other engines | Lower Claude share than ChatGPT/Gemini | Claude does not lean on transcript content the way Gemini does. |
| 38 | GitHub | Technical / Reference | Stable | Top developer-docs citation | Repository docs, READMEs, issues. |
| 39 | Stack Overflow | Community — Technical | Stable | Top developer Q&A citation | Engineering queries. |
| 40 | Mozilla Developer Network | Technical — Docs | Over-indexes on Claude | Recurring top-45 | Web standards reference; documentation depth. |
| 41 | Microsoft Learn | Technical — Docs | Stable | Recurring top-45 | Microsoft ecosystem documentation. |
| 42 | AWS Documentation | Technical — Docs | Stable | Recurring top-45 | Cloud-platform reference. |
| 43 | Forbes | Editorial — Business | Sharply under-indexed vs ChatGPT | One sample dataset contained zero Forbes URLs on Claude | Claude's selective bias filters out contributor-network content. |
| 44 | CNBC | Editorial — Business | Stable | Recurring top-50 | Business news and earnings. |
| 45 | McKinsey | Research — Consulting | Stable | Recurring top-50 | Industry research; sector synthesis. |
| 46 | Deloitte Insights | Research — Consulting | Stable | Recurring top-50 | B2B research. |
| 47 | RAND | Research — Policy | Over-indexes on Claude | Recurring top-50 | Defense, policy, public-sector research. |
| 48 | Council on Foreign Relations | Research — Policy | Over-indexes on Claude | Recurring top-50 | International policy queries. |
| 49 | IEEE | Academic — Engineering | Stable | Recurring top-50 | Standards and engineering literature. |
| 50 | Brave Search index | Retrieval Infrastructure | Foundational mechanism | 86.7% citation overlap with Brave top results | Not a content source per se; the retrieval backend that determines Claude's eligible set. |
Rankings reflect synthesized citation share inside Claude specifically, June 2026 equilibrium. The selective citation rate means rank movement of 3–5 positions can reflect modest sample-size noise; the cluster patterns are stable.
4. Category leaders inside Claude
4.1 Encyclopedic & Reference — Wikipedia
Foundational and stable. Anthropic's training corpus weighting plus Brave Search's reference-heavy index make Wikipedia the unambiguous #1. Britannica is the strongest secondary, with a higher relative share on Claude than on ChatGPT.
4.2 Editorial & News — The New York Times (and the long-form cluster)
Claude's editorial leader is NYT, but the structural finding is the over-indexing of long-form: The Atlantic, The New Yorker, The Economist, and Financial Times all earn higher relative Claude share than they do on any other engine. Reuters and Bloomberg anchor the time-sensitive cluster.
4.3 Academic & Peer-Reviewed — Nature, NIH/PubMed, arXiv
This is where Claude is structurally different from every other engine. Nature, NIH/PubMed, and arXiv are not edge cases on Claude — they are top-15 citation sources. For research-grounded queries, Claude routes to the primary source even when the consumer-facing summary outranks it on Google.
4.4 Institutional & Government — Mayo Clinic, CDC, SEC, Federal Reserve
Government-grade authority. Claude's median Domain Authority of 92 on health citations is driven by this cluster. For any brand in healthcare, financial services, or regulated industries, the institutional cluster is the citation environment to compete inside.
4.5 Research & Policy — Brookings, Pew, OECD, World Bank, RAND
Claude over-indexes on think tanks and supranational research bodies. Policy, economics, and governance queries route disproportionately here. For B2B and public-affairs communications, this cluster is a Claude-specific lever that does not appear in ChatGPT or Perplexity ranking analyses.
4.6 Technical & Documentation — GitHub, MDN, Microsoft Learn
Documentation-style content is what Brave Search retrieves well, and Claude over-indexes on it. For technical product communications, the citation surface is product documentation, repository READMEs, and developer-platform reference material — not press releases or contributor articles.
5. Four structural findings
Finding 1. Claude is the only major engine where prestige long-form editorial outpaces tabloid and wire. The Atlantic, The New Yorker, The Economist, and Financial Times earn materially higher relative share on Claude than on any other engine. For brands placing thought-leadership op-eds, the prestige editorial cluster is more valuable on Claude than the contributor-network cluster (Forbes, Business Insider) that drives ChatGPT visibility.
Finding 2. Claude does not reward contributor-network publishing. One published Claude citation dataset contained zero URLs from Forbes or Bloomberg, and eSEOspace reported only 7% of Claude citations in their sample came from mainstream news domains. The Forbes/Business Insider/Inc.com playbook that delivers on ChatGPT under-delivers on Claude. Editorial placements must come from owned editorial desks, not contributor programs.
Finding 3. Brave Search ranking is the leading indicator. 86.7% citation overlap between Claude responses and Brave top organic results means the practical question is not "is my brand visible on Claude" — it is "is my brand visible on Brave Search." For 90% of brands, Brave ranking has never been measured. It is now the most important search ranking to audit.
Finding 4. Owned-channel content is structurally undervalued. Claude does not cite a brand's own description of itself if it cannot find third-party validation. The mechanism is cross-verification. Brands accustomed to running PR through owned blog, press kit, and About page content are running into a Claude-specific wall: the engine requires third-party signal. This is not opinion — it is documented retrieval behavior.
6. What to do
- Audit your top-15 Claude source presence first. Wikipedia, NYT, The Atlantic, The New Yorker, The Economist, Reuters, Bloomberg, Washington Post, Financial Times, WSJ, Nature, NIH, arXiv, Harvard.edu, MIT.edu. If you cannot account for verified presence on at least 6 of these, Claude visibility is structurally out of reach.
- Earn Brave Search visibility, not just Google ranking. Run a parallel Brave audit for your top 50 buyer-intent queries. Where you rank in Brave top 10, you are eligible to be cited by Claude. Where you do not, you are not.
- Invest in long-form, source-backed editorial placement. Op-eds in The Atlantic, The Economist, HBR, Financial Times, Brookings, and Foreign Affairs are Claude-visibility assets. Forbes and Business Insider contributor placements do not deliver here.
- Map your category to its institutional authority cluster. Healthcare brands compete in Mayo Clinic / CDC / NIH / WHO. Financial-services brands compete in SEC / Federal Reserve / IMF / Brookings. Build third-party citation inside that cluster, not around it.
- Cite the primary source in your own content. Claude rewards source-backed reasoning. Pages that link out to peer-reviewed studies, government data, and institutional research earn Claude citations at materially higher rates than pages that cite only secondary sources.
- Reduce social-channel weighting in your AI visibility scorecard. Claude social citation rate is 3.99%. Spending equal AI-visibility effort on Reddit for Claude and ChatGPT is mis-allocation.
- Plan a 6-month minimum horizon. Claude rewards stable authority, not freshness. Brands starting from zero on Claude should expect 6 months of authority-building before consistent citations appear. The compound is real, but the lag is real too.
- AI Platform Citation Source Index 2026 — the master index this satellite belongs to.
- The ChatGPT Citation Source Index 2026: The 50 Domains ChatGPT Cites Most.
- The Gemini Citation Source Index 2026: The 50 Domains Gemini Cites Most.
- The Perplexity Citation Source Index 2026: The 50 Domains Perplexity Cites Most.
- The Google AI Overviews Citation Source Index 2026.
- The 50 Subreddits AI Engines Cite Most: Inside Reddit's Citation Map 2026.
- GEO Operating Stack.
- AI Communications Master Hub.
- EPR Research Index.
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Thirty-plus publications. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.
