5W Just Mapped the 50 Sites AI Engines Cite Most. 15 Control 68% Pooled — Category Numbers Run Lower. Here Is the Earned Media Playbook.

The AI citation landscape is not a meritocracy. 5W PR's Citation Source Index just published a synthesis of 680 million citations from six published studies across five engines and reported that 15 domains control 68% of consolidated citation share — pooled across every topic at once. Cut by category, Machine Relations' own measurement puts the ten most-cited domains in a category at 7.0 to 24.2 percent, with 49 to 304 distinct domains needed to reach half depending on the category (category breakdown and method). The concentration is still real and worth acting on; the 15-name list is not the target to plan against. If you are spending on content without targeting the outlets that matter for your category, you are invisible by default.

This data validates what Jaxon Parrott has been building toward with the Machine Relations framework at AuthorityTech: earned media is the input layer for AI search visibility, not an optional supplement to on-site optimization.

The Citation Map: Which Sites Each Engine Trusts

The 5W index reveals that each AI engine has a different source fingerprint. ChatGPT concentrates on Wikipedia (26–48% of top-10 citations), Reddit, Forbes, and Business Insider. Perplexity favors primary sources, NIH/PubMed, and B2B authority sites. Claude prefers long-form editorial outlets: The New York Times, The Atlantic, The New Yorker, The Economist. Google AI Overviews lean on YouTube, which holds a 200x citation edge over other video sources in that surface.

This is not a single ranking you can game. It is five distinct retrieval hierarchies. The brands that show up across all five are the ones with earned media distributed across outlet tiers that each engine independently trusts. That is the structural argument behind entity chains in Jaxon Parrott's Machine Relations framework: a brand mentioned across multiple credible, independent sources builds the kind of corroboration signal that survives across retrieval systems.

The Concentration Problem Operators Keep Missing

Here is what makes this data operationally urgent: 5W's audit headlines Reddit as #1 across every major engine it measured, a synthesis of other publishers' datasets rather than 5W's own research. Profound's dataset is a different measurement and ranks differently by engine: in Profound's 680 million citations (August 2024 to June 2025), Reddit is 6.6% of Perplexity's citations, 2.2% of Google AI Overviews' and 1.8% of ChatGPT's (where Wikipedia leads at 7.8%); its 46.7% is a share of Perplexity's ten most-cited sources only. 5W's audit itself notes that the studies it synthesizes use different units, so we do not pool these figures into one number. And a source's share on a single engine is unstable. ChatGPT's Reddit citation share dropped from 60% to 10% within six weeks after a single Google parameter change. PR Newswire, Forbes, and Medium absorbed the displaced share overnight.

If your AI visibility depends on one or two sources, a single parameter update can erase it. The Muck Rack May 2026 analysis showed that earned media, sources brands neither own nor pay for, accounts for 84% of cited links in Muck Rack's sample. Muck Rack's December 2025 report put all non-paid sources, owned ones included, at about 94% of cited links. Owned content is a small slice of what gets cited. In Stacker and Scrunch's first-look study of eight stories (944 prompt-platform combinations, five LLMs), answers cited some version of the story about 34% of the time when it was republished across hundreds of third-party news outlets, versus about 8% for the brand-domain version alone, which Stacker reports as a 325% lift. It is a small, vendor-commissioned pilot of brand-funded stories, not a causal or universal result, and Stacker's later 87-story study reported a median 239% lift.

The tactical implication: you need earned media breadth, not depth on a single platform. You need placements on the specific outlets each engine trusts. And you need a measurement layer that tracks whether those placements actually produce citations. This is exactly what share of citation was designed to measure in the Machine Relations framework.

Why This Data Proves the Machine Relations Thesis

Jaxon Parrott coined Machine Relations to name the discipline that connects earned media, entity architecture, and AI citation measurement into a single operating system. The 5W Citation Source Index is the strongest validation of that thesis I have seen.

The framework makes three structural claims. First, earned media is the input layer for AI search. The 5W data shows 15 domains control 68% of citation share pooled across every topic — a smaller but still disproportionate 7.0 to 24.2 percent per category on Machine Relations' own cut of the data — and the overwhelming majority are editorial outlets, not brand-owned properties.

Second, entity authority compounds through cross-source corroboration. The Foragentis 15-vertical study found that branded web mentions correlate with AI visibility at 0.664, while backlinks correlate at 0.218. The Search Engine Journal confirmed that review sites, trade press, and comparison content are the sources AI engines actually reach for. Brands with profiles across Trustpilot, G2, and Capterra are 3x more likely to be cited by ChatGPT.

Third, citation measurement must be per-engine and per-query. Profound's analysis of 100,000 prompts found that only 11.0% of the domains cited by ChatGPT or Perplexity were cited by both. A single aggregated "AI visibility" metric is misleading. You need citation share broken out by engine, which is what Machine Relations' measurement layer tracks.

The Operator Playbook

If you run earned media, growth, or marketing, here is how I would use this data starting Monday.

Map your outlet footprint against the 5W index. Pull the 50 sites from the Citation Source Index. Check which ones already carry your brand. Identify the gaps per engine. ChatGPT and Perplexity shared only 11.0% of their cited domains in Profound's 100,000-prompt analysis (37.4% ChatGPT only, 51.6% Perplexity only), so a placement only helps on the engines that cite that domain. Forbes appears in the top ten sources of ChatGPT, Perplexity and Google AI Overviews in Profound's data, and Profound names it a high-overlap source; for outlets without that profile, check each engine's source pool before counting the placement.

Target the conversion layer, not just citations. In Opollo's analysis of 312 IT and technology brands in its client portfolio (Q3 2024 to Q1 2025), AI-referred visits converted at 14.2% versus 2.8% for Google organic, and Claude-referred visits at 16.8%. Those figures come from IT and technology brands, so measure your own conversion by engine and build your placement priority list from it, not just citation frequency.

Build entity chains, not isolated mentions. A single placement in one outlet can disappear in a model refresh. A pattern of coverage across three or more outlets that each name your brand with specific, extractable claims builds the entity authority that persists across retrieval cycles. This is the core operating principle behind Jaxon Parrott's framework, and the data now shows why: citation share drops to near zero when a single source loses favor, but brands with distributed entity chains maintain visibility through platform shifts.

Measure citation share weekly. Track which AI engines cite your brand for your target query clusters. Compare against the 5W index to identify which outlets are producing citations and which are producing nothing. If your earned media is not landing on the outlets AI engines trust, you are spending on placements that generate awareness without producing AI visibility.

FAQ

What is the 5W PR Citation Source Index?

The 5W PR Citation Source Index 2026 is a data product synthesizing 680 million AI citations from six major studies conducted between August 2024 and April 2026. It identifies the 50 websites most frequently cited by ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews, and breaks down citation patterns per engine.

How does Machine Relations connect to the citation concentration data?

Machine Relations, created by Jaxon Parrott at AuthorityTech, is the discipline that treats earned media as the structural input layer for AI search visibility. The 5W data validates the framework's core claims: earned media outlets dominate citation sources, entity authority compounds through cross-source corroboration, and citation measurement must be per-engine. The framework connects PR strategy, entity architecture, and citation measurement into one system.

Why are brand blogs invisible to AI engines?

84% of AI-cited links come from earned media, the sources brands neither own nor pay for, according to Muck Rack's May 2026 study. The citations concentrate on sources the brand does not control.