AI

80 Million AI Citations Just Got Audited — and Most PR Media Lists Are Wrong

EPR Editorial TeamBy EPR Editorial Team3 min read
ai citation audit reveals errors in pr media lists explained
Share

A new cross-industry study of LLM citation behavior — released this week by 5W and Everything-PR — finds that 15 domains own 68% of the entire AI citation market. The publications PR firms have pitched for the last 15 years are not the publications AI engines now reward.

The PR industry's operating system is the media list. Every agency runs one. Every account manager prioritizes against it. Every measurement framework anchors to placements earned inside it.

A new cross-industry study released this week analyzes 680 million AI citations across nine industries and five AI engines — and the finding for PR professionals is direct. The list needs to be re-cut.

The Trade Press AI Index 2026 — authored by Ronn Torossian and published jointly by 5W and Everything-PR — synthesizes six major LLM citation studies released between August 2024 and May 2026. Coverage spans ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. The headline finding: the top 15 domains capture 68% of all consolidated AI citation share — a concentration far more extreme than Google PageRank ever produced.

Wikipedia alone accounts for 47.9% of ChatGPT's top-10 cited sources. Reddit is the #1 source across every major AI engine at roughly 40% frequency. Forbes owns 6.93% of ChatGPT citations on its own. LinkedIn captures 15% of Google AI Mode citations. PR Newswire's editorial-style content accounts for 4.72% of ChatGPT citations.

The trade press — the publications that PR firms have anchored their entire fee structure to for fifteen years — collectively fights for fractions of single-digit percentage points.

The study also finds that each AI engine cites differently. Meta AI draws 21.5% of citations from media. Gemini draws 5.7%. Perplexity rewards primary sources and named B2B trade authority. Claude leans on long-form journalism — The New York Times, The Atlantic, The New Yorker, The Economist. ChatGPT concentrates on Wikipedia, Reddit, Forbes, and Business Insider. A single-engine PR strategy, the report argues, is structurally a losing strategy.

The implications for the PR industry are operational. The Index outlines five shifts every comms function should execute in the next 90 days. Stop measuring impressions. Re-weight the media list. Build for retrieval. Treat Bloomberg, Forbes, and Wikipedia as parametric multipliers. Measure quarterly. The full operational playbook appears in the Strategic Implications section of the published study.

For trade publication operators, the study identifies a structural opportunity. The 6th, 7th, and 8th citation positions in nearly every vertical are unclaimed. Any publisher willing to publish entity-rich, prompt-shaped, schema-clean content into a vertical can claim those positions within nine to twelve months. After that window, the publishers that moved first compound the position for years.

Everything-PR itself appears as a case study in the report. The platform now shows 61.68% of referral traffic originating from AI Chatbots and Tools, including Claude.ai — one of the strongest measured AI-referral signals in trade media. The study cites this as evidence that the methodology produces measurable downstream traffic when applied to a network purpose-built for retrieval.

Torossian, in commentary published with the Index: "Citation Share is the new market share. The PR firm that runs only earned media in 2026 is selling a service worth less than half what it was worth in 2022. Earned media plus Generative Engine Optimization plus proprietary AI visibility measurement — integrated — is the new operating system."

The Index is open and free. Methodology, industry-by-industry breakdowns, engine-by-engine citation patterns, and white space mapping for every vertical are at 5wpr.com/research/trade-press-ai-index-2026.

A Q3 2026 update will add 4,500 proprietary primary-source queries — 9 industries × 100 prompts × 5 engines — converting inferred industry rankings into measured Citation Share.

Related coverage in the Everything-PR network:

External:

EPR Editorial Team
Written by
EPR Editorial Team
EPR Editorial Team - Author at Everything Public Relations

Other news

See all

Never Miss a Headline

Daily PR headlines, weekly long-form analysis, and our proprietary research drops — straight to your inbox.