A study published July 9, 2026 by Notified (the newswire infrastructure behind GlobeNewswire) found that 99.3% of press releases distributed through GlobeNewswire were cited by ChatGPT or Claude — with an average time-to-citation of just eight hours. A separate finding in the same study: press releases with structured formatting generated 2.6x more citations than unstructured equivalents covering the same announcements.
Press releases were supposed to be dying. Instead, they’ve become one of the fastest mechanisms available for getting content into AI search results. This post explains why the channel works, what structure drives the citation multiplier, and how to implement it without a PR agency budget.
Why AI Systems Are Especially Receptive to Press Releases
The underlying reason is citation infrastructure. Major newswires — GlobeNewswire, PRNewswire, BusinessWire, AccessNewswire — are among the most consistently crawled domains across every major AI platform. Reddit holds the #1 citation position across LLMs (cited at roughly 40% frequency), and Wikipedia dominates ChatGPT’s top-10 citation share at 26–48%. Newswires sit just below that tier, but they offer something the top-tier platforms don’t: you can publish to them directly.
AI systems trust newswires for a structural reason: they’re timestamped, attributed, and formatted consistently. A press release carries a dateline, a contact name, a company, and a specific claim — exactly the metadata that allows an AI model to treat it as a citable source rather than ambient content. When an AI model is asked “what has Company X announced recently?” the newswire entry is the ground truth it reaches for.
The citation dynamic is also different from organic content. A blog post typically needs inbound links and domain authority before it appears in AI results. A GlobeNewswire release starts showing up in ChatGPT and Claude responses within hours of distribution, regardless of your own domain’s authority.
The 8-Hour Citation Window
Notified’s study tracked citation timing across 25 million press releases distributed through GlobeNewswire in 2025 and the first half of 2026. The average time-to-citation was eight hours. That means a press release published at 9 AM EST is typically appearing in AI-generated answers by 5 PM the same day.
The speed matters because AI citation recency is real. Research from Ahrefs (May 2026) found that content published or significantly updated within the past 13 weeks is disproportionately cited in AI responses, with a sharp drop in citation frequency for older material. Press releases have an inherent freshness advantage: they’re time-stamped to a specific date, and their structure signals that they contain current information.
This creates a compounding effect for brands that publish consistently. Each release enters the citation pool within hours. If a user asks an AI “what’s the latest from [company]” three weeks after a release, the release often still satisfies that query because it carries an explicit date and a current announcement.
The 2.6x Structure Gap: What It Actually Means
Notified’s AI Press Release Optimizer tool produced the 2.6x citation figure by comparing optimized (structured) versus unoptimized (unstructured) releases covering identical announcements. The structured releases got 2.6 times as many citations across AI platforms in the 30-day window after publication.
Structure here means four specific things:
A lead paragraph with exact figures in the first 50 words. AI systems extract the opening of a document first. A release that leads with “Company X raised $12M in Series A funding to expand its AI search optimization platform to European markets” gives the model a citable claim immediately. A release that leads with “Company X is thrilled to announce a milestone in its growth journey” forces the model to read further to find extractable content — and often doesn’t wait.
Subheadings that match query patterns. Press releases with H2 or bold-formatted subheadings aligned to search intent (“Who This Affects,” “What Changes,” “Key Data Points”) get cited more often because the model can locate the relevant section for a given query without parsing the entire document. A company announcing a product update should have a subheading like “How the Update Changes [Specific Workflow],” not just “Product Details.”
Quotations with attribution. Notified found that releases containing at least one quotation from a named executive were cited 34% more frequently. The likely mechanism: a named quote gives the AI model a citable human source within the document, not just a claim. “According to Jane Doe, CEO of [Company]” is more extractable than an anonymous assertion.
Specific numbers, not ranges or vague claims. “The company serves more than 200 customers” is vague. “The company serves 214 enterprise customers across 18 countries” is citable. AI models cite specifics because they need to answer specific questions. Every claim in the body of your press release should carry a number, a date, or a proper noun.
The Mentions vs. Citations Distinction You Need to Track
The Notified study tracks citations — cases where the AI response references the press release as a source. A separate metric that often gets conflated is brand mentions: cases where a company’s name appears in an AI answer without that answer citing the company’s own content.
These are fundamentally different outcomes. A mention means your brand is referenced. A citation means your content is the source. For most AI search queries, being the cited source is more valuable: it drives the user to click through to your domain, it positions your content as authoritative, and it influences how future AI responses characterize your company.
The press release channel is unusual in that it can produce both simultaneously. A well-structured announcement generates citations on the newswire URL and, if the press release links to your website or product page, often drives secondary citations of your domain as the AI learns to associate the announcement with your brand’s web presence.
What This Costs and Who It’s Actually For
A single GlobeNewswire or PRNewswire distribution costs between $500 and $2,500 depending on word count and regional targeting. BusinessWire is comparable. AccessNewswire starts lower, around $300 for basic distribution.
That price point makes press releases a viable tactic for companies with genuine news to distribute — product launches, funding rounds, partnership announcements, study publications, leadership changes — but an expensive one for manufactured content. AI systems are getting better at distinguishing announcements with substance from releases that are purely promotional. A release announcing “Company X Continues to Be a Leader in Its Category” produces fewer citations than one announcing “Company X Releases Study Showing 34% Improvement in [Specific Metric].”
The ROI calculation has changed significantly in 2026. Before AI citation patterns became measurable, press releases were justified (or not) by earned media pickups in trade publications. Now there’s an additional value layer: a $500 release that generates citations in ChatGPT and Claude for 90 days is providing persistent brand visibility in the discovery channel where more than 1 billion monthly users are researching products and services. For categories where AI search drives purchase decisions, that’s a different math than traditional PR reach.
Four Release Types That Generate Outsized AI Citations
Based on the Notified data and the broader pattern of what AI systems cite, four press release formats consistently outperform generic announcements:
Data release announcements. “Company X publishes study on [specific topic with specific finding]” is the highest-citation format. The AI model needs data to answer quantitative questions. If your release is the source of specific data, you become the citation for every query touching that topic. A SaaS company publishing a “State of [Industry Category] Report” with even 50 respondents will generate more citations than ten product launch releases.
Product milestone announcements with exact metrics. “Company X reaches 10,000 paying customers, 90-day retention rate at 84%” is more citable than “Company X announces strong growth.” The specificity is what makes it useful to an AI generating an answer about the company’s traction.
Partnership announcements naming both parties. These generate citations on both sides — the release shows up when users ask about either company. A partnership announcement between a mid-market SaaS vendor and a known enterprise platform creates citation surface for queries about both organizations.
Leadership/expert commentary releases. A release where a named executive takes a position on an industry trend — “CMO of Company X: Brands That Don’t Optimize for AI Citations Will Lose 30% of Discovery Traffic by 2027” — functions as thought leadership content. The AI model cites it when answering questions about trends in the category, not just questions about the company.
How to Track Whether Your Press Releases Are Getting Cited
Standard PR metrics (pickup count, potential reach, AVE) don’t tell you whether your release appeared in ChatGPT or Perplexity responses. To measure that, you need to run the specific queries a user would ask if they were looking for information your release covers, then check whether the AI response references your newswire distribution.
The queries to test are the ones a buyer or journalist would ask: “What has [Company] announced recently?” “Who are the leaders in [category]?” “What data exists on [topic your study covers]?” For each query, check ChatGPT, Perplexity, Claude, and Google AI Mode. Your release should appear in at least two of the four if it was well-structured and distributed through a major newswire.
You can run a free AI visibility audit at ai-visibility.llmagnet.com to see how your brand currently appears across AI platforms — and identify gaps in which queries your content is failing to surface for.
Conclusion
The press release stopped working for traditional SEO years ago. It works for AI citation optimization in 2026 for a specific reason: newswires are in the trust tier that AI systems draw from, and a well-structured release enters the citation pool within hours of distribution.
The 2.6x citation multiplier from structured formatting — lead paragraphs with specific figures, subheadings that match query patterns, named executive quotes, and exact numbers — is the tactical difference between a release that gets cited and one that gets ignored. The 99.3% citation rate for GlobeNewswire distributions means the channel is reliable; the structure is what determines how often and how prominently.
If your company has announcements worth making — data publications, product milestones, partnerships, leadership commentary — press releases distributed through major newswires are now one of the fastest paths into AI-generated answers. The investment is modest relative to the visibility window it creates.
→ Check whether your brand currently appears in AI answers: ai-visibility.llmagnet.com (free audit)