The AI Platform Citation Source Index 2026: The 50 Websites AI Engines Cite Most

Why "what does ChatGPT cite first" replaced "what does Google rank first"

For a generation, Google PageRank was the most consequential algorithm in information distribution. It decided which pages surfaced first, and in doing so decided which institutions, brands, and individuals were visible. Every communications strategy, every SEO budget, every earned-media placement, every Wikipedia entry was built around a single foundational question: what does PageRank reward?

That question is now a legacy question. Its successor is more consequential, more measurable, and far less well understood: what do ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews actually cite when they assemble an answer? Between August 2024 and April 2026, six of the largest citation-tracking studies ever conducted produced overlapping datasets covering more than 680 million individual citations. Everything-PR Research has synthesized that data into the first consolidated ranking of the fifty sources that compose the modern AI answer.

The findings contradict most prevailing assumptions about AI behavior. Platforms do not treat the open web as a level playing field. They concentrate citation share in a remarkably small set of domains. They privilege user-generated community content over most institutional journalism. They treat a single encyclopedia as near-foundational training material. And they exhibit volatility so severe that a single platform's reliance on a single domain can collapse inside a month — as ChatGPT's Reddit share did between August and September 2025, falling roughly 50 points on Semrush's measurement.

For communicators, publishers, educators, institutions, and brand owners, this is the working map of the new discovery terrain. The organizations that understand it will compound visibility over the next 24 months. The organizations that do not will discover, too late, that they have been running campaigns against a set of rules whose mechanics were already documented.

How the AI Platform Citation Source Index was built: methodology and source studies

This index synthesizes verified citation data from six of the largest published studies of generative AI citation behavior conducted between August 2024 and April 2026, covering more than 680 million individual citations across the five dominant AI answer engines.

Source studies and scope

• Profound's longitudinal citation study (August 2024 – October 2025), covering 680 million+ citations across ChatGPT, Google AI Overviews, and Perplexity, with updates through Reddit's own Q2 2025 shareholder disclosure.
• Semrush analyses of 150,000 citations (June 2025) and 230,000 prompts over 13 weeks (August – November 2025), documenting citation volatility after the September 2025 Google num=100 parameter change.
• Ahrefs Brand Radar analysis of the 100 most-cited domains in ChatGPT (September 2025), broken out by US and global coverage.
• Search Engine Land consolidated cross-platform analysis (October 2025), covering Reddit, YouTube, LinkedIn, Wikipedia, and Forbes as the leading multi-platform citation sources.
• Muck Rack's journalism citation study (July 2025), analyzing more than one million citations to measure how often AI platforms rely on journalistic content and which publishers are cited most.
• Visual Capitalist's synthesis of LLM citations across more than 150,000 queries (August 2025), documenting aggregate ranked frequency data.

Everything-PR Research consolidated these studies into a cross-referenced citation-share signal — defined as the average rank position of each domain across the source studies, weighted by each study's citation volume — and constructed a ranked index of the 50 domains that collectively define the modern AI answer. We grouped the index into six functional categories — Community & Conversation, Encyclopedic & Reference, Professional & Identity, Video & Audio, Editorial & News, and Commerce & Review — to reflect the distinct strategic approaches each category requires. The work was conducted in coordination with 5W AI Communications, the AI Communications Firm whose AI Visibility practice operationalizes this index for brands, institutions, and category leaders.

Four structural findings that will shape AI visibility strategy through 2027

September 2025 → September 2026: Twelve months of volatility, in numbers

One year on from the platform shocks that reset how communicators think about AI citation, the pattern is measurable — and it does not resemble the search era it replaced.

The 50-point swing

In early August 2025, Semrush measured ChatGPT citing Reddit at roughly 60% of AI Overviews-aligned queries. By mid-September, that share had collapsed to approximately 10% — a 50-point swing inside six weeks. Reddit's multi-engine aggregate share held near 40% over the same period. The volatility was platform-specific, not market-wide, and it followed a single upstream Google search parameter change (the num=100 deprecation). One product decision at one platform moved half of a channel's citation share out of ChatGPT's answer.

The Forbes doubling

Post-September 2025, Forbes' ChatGPT citation share roughly doubled. LinkedIn climbed to #5. PR Newswire became a direct input to the ChatGPT retrieval layer. The engines redistributed the share Reddit lost across editorial and wire — not evenly, not to the biggest brands, but to the sources with the highest structural extractability.

The concentration held

Through twelve months of platform-specific volatility, the top 15 domains held approximately 68% of consolidated AI citation share. The identities inside the top 15 rotated. The concentration itself did not move. That is the structural finding: AI citation is aggressively concentrated, and the concentration is more stable than any individual domain's position inside it.

The journalism split

Journalism's share of AI citations held at 27% of all queries — rising to 49% on time-sensitive queries. Evergreen and how-to answers moved further toward user-generated communities. The category the engines lean on most for recency is the same category losing share on the questions buyers ask every day.

What it means for a 2026–2027 program

Diversification across the top 15 is not optional; it is risk management. A brand whose Citation Share depends on a single source is one platform-policy change away from disappearing from the answer. The organizations compounding visibility over the next 24 months are the ones whose presence is distributed across community, encyclopedic, video, professional, editorial, and commerce citation surfaces at once.

The seven-step operating checklist: how to earn Citation Share in 2026

1. Audit brand presence across the top fifteen sources before the long tail. An organization without verified presence across Reddit, Wikipedia, YouTube, LinkedIn, Forbes, Reuters, and four to five category-specific domains is invisible to 68% of the AI citation pipeline. Audit first; strategize second.
2. Build Wikipedia as infrastructure, and resource it accordingly. ChatGPT cites Wikipedia at 26–48% of its top-10 share. A stub entry, an outdated infobox, or a missing citation-density score is direct Citation Share leakage. Wikipedia is a resourced function, not a one-off.
3. Treat Reddit as a strategic channel, not a tactical one. Reddit is the single most-cited domain across AI platforms. A brand without a defensible Reddit presence — organic thread density, AMA cadence, subreddit-specific credibility — is being written out of nearly half the queries buyers now ask. See the 50 subreddits AI engines cite most.
4. Map journalism targets to platform-specific citation patterns. ChatGPT favors Forbes, Business Insider, Reuters. Perplexity favors NIH, Microsoft, Britannica. Claude favors NYT, Atlantic, New Yorker, Economist. Gemini favors Google properties and YouTube. Google AI Overviews favors Reddit and Wikipedia. Same story, different targets.
5. Treat LinkedIn as a citation channel, not a social-media channel. LinkedIn is a top-five multi-platform citation source and dominant in B2B and executive queries. Company Pages, employee thought leadership, and long-form articles are all directly retrievable content — treated by the engines as authority rather than social ephemera.
6. YouTube is the only serious video-citation asset. Any video-content strategy that does not prioritize YouTube is not a serious AI visibility strategy. Transcript density, chapter markers, structured descriptions, and pinned comments all lift extractability.
7. Plan for volatility, not stability. Diversification across the top fifteen sources is no longer optional; it is risk management. A brand whose Citation Share depends on a single source is one platform-policy change away from disappearing from the answer.

Operating Model and Engine Guides

The Index maps where engines pull from. The operating model for earning a place in those sources is the GEO Operating Stack. Engine-by-engine playbooks:

• How to Rank on ChatGPT · ChatGPT Citation Source Index 2026
• How to Rank on Claude · Claude Citation Source Index 2026
• How to Rank on Gemini · Gemini Citation Source Index 2026
• How to Rank on Perplexity · Perplexity Citation Source Index 2026
• How to Rank on Google AI Overviews and AI Mode · Google AI Overviews Citation Source Index 2026

Practitioner companion: How PR Teams Use Claude — the working guide for communications professionals using Claude inside daily workflows: drafting, editing, document analysis, message development, copy-paste prompts, and Claude Projects for agency operations.

Related EPR Coverage

• EPR Research Index — every research series in one place
• AI Communications Master Hub — the keystone cluster
• The Citation Share Index — the brand layer companion
• Who Controls AI Answers — the vertical-by-vertical source map
• The GEO Operating Stack — the fourteen-layer operating model
• The AI Communications 100 — the figures shaping the AI answer layer
• AI Communications & GEO: The Practitioner's Guide
• The 50 Subreddits AI Engines Cite Most — the Reddit citation map