TrendlyAI
Menu

Digital PR for AI Citations: Get Covered Where AI Looks

By Bhuwan Aryal, founder of TrendlyAIPublished 10 min read

Digital PR helps AI visibility because AI answers draw heavily on earned media: news coverage, expert articles, research and reference sites, not ads. To get covered where AI looks, answer journalist requests in your area of expertise, publish original data reporters can use, and pitch only stories that are true and useful to their readers.

Key takeaways๐Ÿ”—

Which publishers do AI answers cite?๐Ÿ”—

AI answers cite a mix of news outlets, trade publications, reference sites and community sites, and the mix changes by engine and by question type. The strongest single dataset on earned media comes from Muck Rack, which sells PR software, so weigh that, but its numbers are consistent across editions.

In its May 2026 "What is AI Reading?" update, Muck Rack analyzed more than 25 million links cited by ChatGPT, Claude and Gemini across 17 industries. It found:

Domain-level studies of other engines point the same way. Here is what three of them show for major publishers:

StudyEnginePublisher findings
Ahrefs (September 2026, 3 million+ US queries)Google AI OverviewsForbes No. 12 (0.8% of citations), The New York Times No. 26 (0.5%), CNBC No. 44 (0.3%)
Profound (680 million citations, August 2024 to June 2025)ChatGPTForbes 1.1% of citations, Reuters 0.6%
Profound (same dataset)Google AI Overviews, PerplexityForbes 0.6% and 0.7%
Muck Rack (May 2026)ChatGPTAxios in the top three domains in 13 of 17 industries

Those shares look small, but each is a single publisher competing with millions of sites. The lesson is not "get into Forbes." It is that AI engines trust independent, edited coverage, and that includes the trade publications your buyers already read. For the full list of domains ChatGPT leans on, see our breakdown of the most-cited domains in ChatGPT.

Why does earned media matter more than your own site?๐Ÿ”—

Earned media matters because an AI engine answering "what's the best tool for X" wants sources that are not the vendor. Your own site can explain what you do, but a reporter, analyst or reviewer saying it carries more weight, the same way it does with a human buyer.

Earned coverage also travels. One quote in a trade article can get picked up by newsletters, roundups and "best of" lists, and each of those is another page that names you in the context of your category. That is the pattern we describe in our guide on how to get mentioned in ChatGPT: the more independent pages that connect your brand to a buyer's problem, the more likely an engine is to make the connection too.

This does not make your own site irrelevant. Coverage often links back to your pages, and engines still need a clear, accurate source for your pricing and features. Think of PR as the part of AI visibility you cannot write yourself.

How do journalist request platforms work?๐Ÿ”—

Journalist request platforms let reporters post questions they need expert sources for, and let you reply with a short, quotable answer. If the reporter uses it, you get quoted, usually with your name, title and company. Here are four that are active as of October 2026:

PlatformHow it worksCost for sources (as of October 2026)
HARO (Help a Reporter Out)Email digests of reporter queries. Cision shut it down as part of Connectively in December 2024, then sold it to Featured.com, which revived it in April 2025.Free for sources and journalists at relaunch, per the acquisition announcement; check the HARO site for current terms
QwotedExpert profile plus pitches to reporter requestsFree Basic plan with 2 pitches a month and a 2-hour pitch delay. Pro is $149.99/mo, or $1,188/year
FeaturedFinds journalist requests, podcasts and other media opportunities in one placeFree plan with a limited daily AI allowance. Lite $39/mo, Pro $99/mo, Pro+ $299/mo (cheaper billed yearly)
Source of SourcesRun by HARO founder Peter Shankman. Up to three emails a day of queries; you reply directly to the reporterFree

A few rules make these platforms work:

  1. Answer only where you are a real expert. Source of Sources says off-topic or spammy replies get you removed with "no exceptions, no appeals." Reporters on every platform think the same way.

  2. Lead with the quote. Put a two or three sentence answer the reporter can paste, then one line on who you are. Skip the company history.

  3. Be specific. A number from your own experience, a clear opinion, or a concrete example beats a generic tip.

  4. Write it yourself. Reporters are flooded with AI-written pitches. When Featured bought HARO, its CEO Brett Farmiloe said that "as AI floods the internet with generic content, journalists need credible, human sources more than ever." A pasted, generic answer is exactly what reporters are trying to filter out.

  5. Respond fast. Queries often have short deadlines. A free plan's delay can cost you placements, which is the real reason to consider paying.

Why do original data studies work for AI citations?๐Ÿ”—

Original data works because reporters need numbers with a source, and AI engines love passages that contain a specific number tied to a named source. A small, honest dataset you own can be quoted for years.

You do not need a huge survey. Good data studies for small companies include:

Make it easy to cite. Publish the study on your site with the main finding in the first paragraph, a short methodology, the sample size and dates, and a chart or table. State the limits plainly. Then pitch it to the reporters and trade publications who cover your buyers, and use it in your journalist request replies. Our content strategy for AI citations covers how to structure the page itself.

Never invent or inflate the numbers. A data study that falls apart under a reporter's questions does lasting damage, and once an AI answer repeats a wrong number with your name on it, it is hard to take back.

How do you pitch honestly?๐Ÿ”—

You pitch honestly by offering a real story or a real expert, saying clearly who you are, and never paying for coverage that pretends to be independent. Here is how the common tactics compare:

TacticHonest?Likely result
Replying to journalist requests where you have real expertiseYesSteady trickle of quotes in relevant articles
Pitching an original data study to reporters on your beatYesFewer placements, but each can be cited for a long time
Offering your founder for podcasts and expert interviewsYesBuilds a record of your name next to your category
Paid sponsored content clearly labeled as an adYes, if labeledFine for reach, but weak as an independent signal
Paying for "editorial" articles with no ad labelNoFTC risk, and Google treats paid links that pass ranking credit as spam
Mass-sending AI-written pitches to every queryNoIgnored, then blocked by reporters and platforms

The rules behind the "No" rows are public. The FTC's native advertising guide says an ad "shouldn't suggest or imply to consumers that it's anything other than an ad," and recommends clear labels like "Ad" or "Sponsored Advertising Content" rather than vague ones like "Promoted." Google's spam policies list "advertorials or native advertising where payment is received for articles that include links that pass ranking credit" as link spam, unless the links carry rel="sponsored" or rel="nofollow".

There is also a practical reason. Engines and readers both discount coverage that looks bought. A labeled sponsored post is a fair ad buy, but it is not the independent signal that earned media gives you.

How do you measure whether PR is improving your AI visibility?๐Ÿ”—

You measure it by tracking the AI answers themselves, not just counting press hits. Pick the buyer questions that matter, check them before a PR push, and re-check them every week afterward.

  1. List 10 to 25 buyer prompts that a reporter's coverage could influence, such as "best HR software for restaurants." Our competitor AI visibility analysis shows how to pick them.

  2. Record a baseline: which engines name you, which name competitors, and which sources they cite.

  3. Log each placement with the date, outlet and the prompts it relates to.

  4. Re-check weekly and look for new mentions, and for the outlets that covered you appearing in the cited sources.

  5. Be patient and skeptical. Coverage can take weeks to show up in answers, and answers vary from one ask to the next, so one good result is not a trend.

TrendlyAI does the re-checking for you. It scans ChatGPT, Perplexity, Claude and Google AI Overviews weekly for the prompts you choose, shows whether you or your competitors are named, and flags won citations when a gap turns into a mention. It does not track Gemini, Grok, Copilot or AI Mode. For tracking basics, see our guide to tracking brand mentions in ChatGPT, Perplexity and Google, and the GEO hub for every other way to earn citations.

Not sure where you stand today? Run the free AI visibility check. It tests 5 prompts on ChatGPT and Perplexity in about 30 seconds, no signup needed.

Frequently asked questions๐Ÿ”—

Does press coverage help you show up in ChatGPT?๐Ÿ”—

It can. Muck Rack's May 2026 study of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% of citations came from earned media and 27% from journalism. Coverage that connects your brand to a buyer's problem gives engines independent sources to draw on.

Is HARO still running?๐Ÿ”—

Yes. Cision shut HARO down as part of its Connectively platform in December 2024, then sold it to Featured.com in April 2025, which relaunched it as a free email service for sources and journalists. Check the HARO site for current terms.

Is Qwoted free?๐Ÿ”—

Qwoted has a free Basic plan for sources with 2 pitches a month and a 2-hour pitch delay. As of October 2026, its Pro plan costs $149.99 a month or $1,188 a year and allows 35 pitches a month.

Should I pay for sponsored articles to get into AI answers?๐Ÿ”—

Labeled sponsored content is legal, but it is a weak independent signal. Unlabeled paid articles break FTC guidance, and Google treats paid links that pass ranking credit as spam. Earned coverage is slower but far more durable.

What kind of data study gets picked up by reporters?๐Ÿ”—

A focused study that answers a question your buyers and their trade press care about, with a clear sample size, dates and method. It does not need to be large. It needs to be true, new and easy to quote in one sentence.

GEOAI SearchGuides

About the author

Bhuwan Aryal is the founder of TrendlyAI, a tool that tracks whether ChatGPT, Claude, Perplexity and Google AI Overviews recommend a brand or its competitors. He builds and runs the product himself and tracks TrendlyAI's own AI visibility in public.

Keep reading

New to GEO? Start with the complete GEO guide.