Layer One
Structural
Wikipedia, Wikidata, knowledge graphs, industry databases. The encyclopedic, machine-readable layer.
Research · Earned Authority
The condition that can't be bought.
Earned Authority is the third condition of the Recognition Triangle: what others say about you when you are not in the room, built across three layers, Structural, Expert, and Community. It cannot be bought, faked, or tagged into existence. This page is the evidence for what actually earns it, drawn from a citation index covering more than 680 million AI citations.
Last verified 20 August 2026In one paragraph
A 2026 citation index synthesizing six major studies found that roughly fifty domains supply most of what ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews say about brands, and that the top fifteen of those alone account for 68% of the total. Reddit is the single most-cited domain of any kind. Wikipedia supplies up to nearly half of ChatGPT's top citations. And four of the world's most recognized news brands, the Wall Street Journal, the New York Times, Bloomberg, and the Financial Times, do not appear in the top twenty at all. Earned Authority is not a reflection of how well-known you are. It is a measurable, specific pattern of who else is talking about you, and where.
The Three Layers
Structural, Expert, Community. All three, measured.
Earned Authority is not one thing. It is built across three separate layers, and the citation data maps onto each of them cleanly.
Layer Two
Expert
Analyst reports, academic citations, journalism, and trade publications. The vetted, professionally-scrutinized layer.
Layer Three
Community
Forums, reviews, practitioner discussion, professional networks. The unfiltered, ongoing-conversation layer.
- Reddit is the single most-cited domain of any kind, across every major AI engine, at roughly 40% citation frequency, the largest share in the entire index - 5W, 2026
- LinkedIn jumped from the 11th most-cited source to the 5th in three months, now appearing in 14.3% of ChatGPT Search responses - 5W, 2026
On-page tactics do not build any of the three layers. A controlled study of 1,885 pages found schema markup produced no statistically meaningful citation gain on ChatGPT or Google AI Mode - Ahrefs, 2026. Earned Authority is earned off-page, by the sources above, or it is not earned at all.
The Shortlist Is Shorter Than You Think
Fifty domains decide most of what AI says.
The AI Platform Citation Source Index synthesizes six of the largest published citation studies, more than 680 million individual citations across ChatGPT, Google AI Overviews, Perplexity, Gemini, and Claude, from August 2024 to April 2026. It found that citation share concentrates on roughly fifty domains, and the concentration inside that fifty is sharper than classic Google PageRank ever showed for ordinary search results.
Source: 5W, AI Platform Citation Source Index 2026, published May 2026. A brand can be well built and still invisible to AI, simply because none of these fifty sources has ever mentioned it. That gap has a name: Authority Debt.
Authority Has To Be Monitored, Not Banked
A source's citation share can collapse in two weeks.
Even the top of the list is not stable. The most striking single finding in the index is how fast Reddit's own citation share moved, in either direction, within a single quarter.
Reddit's own ChatGPT citation share collapsed from roughly 60% to roughly 10% of prompt responses within two weeks in September 2025 - 5W, 2026. The index's own conclusion is that annual AI audits are obsolete and quarterly tracking is the necessary minimum. It is the same reason our own Recognition Baseline Study, below, is built to run on a schedule rather than once.
The Absentees
Reputation and AI citation are not the same thing.
Four of the most widely recognized news brands in the world are functionally invisible to the exact system this page is about.
“
The Wall Street Journal, The New York Times, Bloomberg, and the Financial Times do not appear in the top 20.
5W, AI Platform Citation Source Index 2026 · Read the source
This is the clearest evidence available that Earned Authority is not a proxy for brand fame. It is a specific, measurable pattern of citation that some famous, highly credible publications have not built, largely because paywalls and licensing restrictions keep AI crawlers out, while Reuters and Forbes, both structurally more open, rank far higher. Reputation among people and recognition by a machine are related, but they are not the same condition, and the gap between them is exactly what Authority Debt describes.
This page covers the citation-source evidence specifically. For how the machine decides which of those sources to trust in the first place, see the Machine Gatekeeper page. For the wider evidence base across all three conditions of the Recognition Triangle, see the main research hub.
Back to the research hubCommon Questions
Questions people ask about this page.
What is the AI Platform Citation Source Index?
A 2026 index from 5W synthesizing more than 680 million citations across ChatGPT, Google AI Overviews, Perplexity, Gemini, and Claude, drawn from six of the largest published citation studies between August 2024 and April 2026, to identify the roughly fifty domains that disproportionately supply what these systems say about brands.
Why is Reddit cited so often by AI systems?
Reddit represents the Community layer of Earned Authority: unfiltered, ongoing discussion that AI systems appear to weight heavily as a signal of what real people actually think, separate from what a brand says about itself. That is also why its citation share moves so fast; it reflects live conversation, not an indexed archive.
Does this concentration look the same in every industry?
The index reports a global pattern across five major engines, but the exact mix shifts by category. A regulated or professional-services category tends to lean harder on the Expert layer specifically, for example, while consumer categories lean harder on Community. The underlying pattern, a small set of sources doing most of the work, held across every study synthesized.
Is this page part of the same research as the main hub?
Yes, this is one of several framework-specific evidence pages under /research/, each going deeper on one condition of the Recognition Triangle. This one covers Earned Authority specifically, who gets cited and how concentrated that citation share is. The main hub covers the wider evidence base across Clarity, Credibility, and Earned Authority together.
A note on verification
All figures on this page are checked against the primary source: 5W's own published press releases for the AI Platform Citation Source Index and its Wikipedia/Reddit follow-up release, and Ahrefs' own reported schema-markup study. Where a figure could only be confirmed via a secondary write-up, that is noted in the text. If a figure here has been superseded or misquoted, tell us at info@aingworth.com and we will correct it.
See where your own Earned Authority stands today.
Run the free diagnostic to find which condition of the Triangle is leaking, or run the Recognition Check to see how often AI actually names you.