Press Release Distribution Won't Get You Cited in AI Search. Here's What Actually Does.

If you've been pitched on press release distribution as a path to AI search visibility, you're not alone. Several distribution platforms have spent the past year marketing their services on exactly this premise: distribute your press release widely enough, and AI models will pick it up.

A study published in March 2026 ran 4 million AI citations through a tracking tool and found something inconvenient for that pitch.

Key Takeaway: Distributed press releases account for just 0.04% of all AI citations across ChatGPT, Google AI Mode, AI Overviews, and Gemini. Original editorial content makes up 81% of news citations. For AI visibility, the channel matters as much as the content.

Written by Derek Chua, digital marketing consultant and founder of Magnified Technologies. Derek works with SMEs across Singapore to build content strategies that work in both traditional and AI-powered search.

---

What the Data Actually Shows

BuzzStream, working with Citation Labs and their XOFU monitoring tool, ran 3,600 prompts across 10 industries for one week. They tracked citations across ChatGPT, Google AI Mode, Google AI Overviews, and Gemini. Four million citations in total.

Here's what they found for press releases and syndicated content:

• Press releases distributed through syndication channels (Yahoo, MSN) accounted for 0.04% of all citations in the dataset
• Direct citations from newswires like PRNewswire: 0.21% of total citations
• Syndicated news overall (articles republished through MSN and Yahoo networks): 0.9% of total citations

For context, news publications as a whole accounted for 14% of all citations. But within that news category, original editorial content dominated at 81% of news citations. Affiliate and review content took the rest.

The split held across all three prompt types they tested: evaluative prompts ("Is Sony better than Bose?"), brand awareness prompts ("What is Chase known for?"), and informational prompts.

Why AI Models Ignore Distributed Releases

This isn't random. There's a structural reason why distributed press releases rarely get cited.

Syndicated content, by design, appears on multiple domains simultaneously. Yahoo Finance, MSN, and hundreds of aggregator sites all publish the same text. For AI models trained to prioritise authoritative, unique sources, identical content on low-authority domains signals exactly the opposite of what you want.

Google's VP of Product for Search, Robby Stein, has said that AI recommendations work similarly to how a human might research a question: they look for sites where other people are talking about you. That framing favours earned editorial coverage in publications with genuine audiences, not distributed text on aggregators.

There's also the crawl access problem. A separate study found that 71% of major news publishers now block AI retrieval bots. Even when syndicated content is technically accessible, it may be sitting on sites that AI crawlers can't or won't read.

The distributed press release gets published in all the wrong places.

The One Exception Worth Knowing

Here's where the data gets interesting for businesses that do have something to announce.

Newsroom content published on company-owned domains accounted for 18% of ChatGPT citations in the BuzzStream dataset. On Google's AI platforms, that number dropped to around 3%.

ChatGPT cited examples including Iberdrola's corporate press room and Target's corporate subdomain. Not the Yahoo Finance version of those releases. The original source, on the company's own domain.

This is a meaningful distinction. The distributed copy loses. The original source on your own website can still get picked up, at least by ChatGPT.

For most SMEs, this points toward building a proper news or resources section on your own domain rather than outsourcing your announcement content to a wire service.

What Actually Gets Cited: The Practitioner View

At Magnified, the pattern we see across client content aligns with what BuzzStream found. Content that gets cited in AI answers tends to share a few characteristics: it takes a position, it provides something specific (a number, a framework, a comparison), and it lives on a domain that AI crawlers can reach.

Press releases, by contrast, are typically written to be distributed, not to answer a question. They're about the company, not for the reader. AI models, like search engines before them, have learned to spot the difference.

The BuzzStream data broke down citations by prompt type. Evaluative prompts ("which is better") generated the most news citations at 18%. The content that dominated those evaluative citations: head-to-head comparisons, cost analyses, and reviews from outlets like Reuters, CNBC, and CNET.

That's editorial content. Written with a point of view. Not a press release.

Three Things That Actually Improve Your AI Citation Rate

The data is clear enough that we can draw practical conclusions.

1. Invest in original editorial content on your own domain

Articles, guides, comparisons, and frameworks published on your website are what AI models cite. Not because they're longer or better-optimised, but because they're original, authoritative, and accessible.

If you're a specialist in your field, the most effective thing you can do for AI visibility is publish original analysis or commentary that takes a clear position. That's what evaluative AI prompts pull from.

2. Build a proper newsroom or resources section on your website

If you do have announcements to make, publish them on your own domain first. Given ChatGPT's pattern of citing company newsrooms directly (18% citation rate), a well-maintained resources or news section on your site outperforms a wire distribution.

Structure it properly: use Article schema markup, include author attribution, date clearly, and make it crawlable.

3. Earn editorial mentions, not placements

Robby Stein's point about AI recommendations favouring sites that other people are talking about points toward genuine digital PR: getting journalists, analysts, and bloggers to mention you in original editorial pieces. Not paying for placement, not syndicating releases, but being genuinely interesting enough that others write about you.

That's harder than paying a wire service, which is exactly why it works.

What This Means for Your Marketing Budget

If you're currently spending on press release distribution and hoping it improves your AI search visibility, BuzzStream's data suggests you're not getting value for that spend on the AI visibility dimension.

That doesn't mean press releases are useless. There are still cases where they serve a purpose: investor relations, regulatory announcements, genuine news that journalists will pick up. But the AI visibility argument from distribution platforms doesn't hold up against the data.

The better allocation: put that budget into owned content on your website, improve your schema markup so AI can parse your pages accurately, and build the kind of editorial presence that earns citations rather than buys distribution.

A well-structured article on your domain, with clear author attribution and specific data, is worth more in an AI citation context than a press release republished on 200 aggregator sites.

---

Frequently Asked Questions

Do press releases still help with SEO? Press releases can still generate backlinks and earned media coverage if journalists pick them up. The BuzzStream data specifically addresses AI citations, not traditional SEO. For backlinks, a press release that earns coverage in a real publication still has value. For AI search visibility directly, the distribution channel shows almost no impact.

What types of content get cited most in AI answers? Original editorial content makes up 81% of news citations in the BuzzStream dataset. Within that, evaluative content (comparisons, cost analyses, reviews) generates the most citations overall. Content published on domains with genuine authority and editorial standards consistently outperforms syndicated or distributed material.

Does publishing a press release on my own website help? Yes, more than you might expect on ChatGPT. Company-owned newsroom content accounted for 18% of ChatGPT citations in the dataset, compared to just 3% on Google's AI platforms. Publishing your announcements on your own domain, with proper structure and schema markup, gives you a better shot at being cited directly than distributing through a wire service.

How does schema markup affect AI citations? Schema markup helps AI models correctly interpret your content: what type of content it is, who wrote it, when it was published, and what it's about. Article schema with author and date metadata signals to both search engines and AI crawlers that your content is structured, original, and attributable. It's part of what separates citable content from generic filler.

Should SMEs invest in digital PR for AI visibility? Digital PR that earns genuine editorial coverage in real publications is a better investment for AI visibility than paid press release distribution. AI models favour content from authoritative sources. If your goal is to be cited, being mentioned in an article on an established publication's domain (not its aggregator syndication) is the outcome that matters. Focus on being genuinely newsworthy, not just distributed.

---

If you want to build a content strategy that earns citations in AI search rather than renting visibility through distribution, our SEO and GEO services start with what the data actually shows.