Skip to content
Chalk Labs / AI SEO
Business impact

Why PR Matters for AI Search Visibility

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

PR matters for AI search because assistants build answers from many web pages, and what other sites say about your brand is part of what they find. In Ahrefs' study of 75,000 brands, web mentions of a brand tracked AI visibility more closely than links did. That is correlation, not proof, and studies disagree on how much AI cites earned media. 1

  • Brand mentions across the web correlated with AI visibility more strongly than backlinks in Ahrefs' 75,000-brand study, though that shows correlation, not cause.
  • Four vendor studies give four different answers on how much AI cites earned media, because they define the term differently.
  • An unknown brand can gain from its own pages; an established brand drew most new mentions from third-party content.
  • Paid placements carry legal duties: the FTC expects clear disclosure of paid connections, and you answer for what agencies do for you.
  • Measure PR with a fixed list of prompts, run monthly, recording whether you are named and which sources are cited.

Why does PR matter for AI search visibility?

PR matters because AI assistants answer from many web pages, and independent pages about your brand are part of that evidence. Earned media means coverage you did not pay for, such as news stories, analyst notes and reviews. A branded web mention is your brand name appearing on another site, linked or not. 2

In Ahrefs' December 2025 study of 75,000 brands, branded web mentions correlated with AI visibility at 0.664 for ChatGPT, 0.709 for AI Mode and 0.656 for AI Overviews. Domain Rating, a link-strength score, came in at 0.266, 0.285 and 0.326. 1 A correlation of 1.0 would be a perfect match, so these are moderate. Ahrefs states that correlation isn't causation. 1

An earlier Ahrefs run on Google AI Overviews, published in May 2025, found 0.664 for mentions and 0.218 for backlinks. 2 Both studies used brands with a Domain Rating above 40, so they say little about small brands.

What a mention gap could cost you each month

Enter how often buyers ask AI about your category and how often a rival is named instead of you. The result is the size of the gap that PR and better pages are aiming at. It is not a forecast of what any campaign will win back.

Roughly how many times a month buyers ask about what you sell.
Of the AI answers, how often a competitor is named and you are not.

What this adds up to

About 72 AI answers a month name a rival and not you. At these rates that puts about 2.2 enquiries, 0.4 customers and $1,728 in revenue at stake each month, or $20,736 a year.

About 72 AI answers a month name a rival and not you. At these rates that puts about 2.2 enquiries, 0.4 customers and $1,728 in revenue at stake each month, or $20,736 a year.
Buyers asking AI a month
120
Times you are absent
72
Answers that name a rival, not you.
Enquiries at stake a month
2.2
Deals at stake a month
0.4
Revenue at stake a month
$1,728
Revenue at stake a year
$20,736

How it is worked out

  • Buyers asking AI = buyer questions × share who ask an AI assistant.
  • Times you are absent = buyers asking AI × share of answers that name a rival and not you.
  • Enquiries at stake = times you are absent × enquiry rate.
  • Deals at stake = enquiries at stake × close rate.
  • Revenue at stake = deals at stake × average customer value. The yearly figure is the monthly figure × 12.
Size this for my company (free proposal)

What do the studies say about earned media, and why do they disagree?

They disagree widely. Muck Rack, a PR software vendor, says 84% of citations in AI answers come from earned media, based on more than 25 million prompts since July 2025. 4 Its page does not define earned media, and lists ChatGPT, Gemini and Claude only as integrations, not as the systems behind the 84% figure.

Profound analysed 11.84 billion citations across eight AI models between April and July 2026 and found brand-owned sites were about 57% of citations. 3 Yext, which sells listings software, analysed 6.8 million citations in four industries and says 86% came from brand-managed sources. 5 Yext's own categories add to 94%, not 86%, and it counts reviews and social posts as brand-managed. 5

Two of the four come from firms whose products fit their headline: Muck Rack sells PR software and Yext sells listings software. That does not make the numbers false. It does mean you should ask three questions before relying on one: who ran it, what does “earned” mean, and which prompts were used?

Four studies on where AI citations come from. Our reading of each page on 20 October 2026.
StudyRun byHeadlineWhy to be careful
Ahrefs, Dec 2025Ahrefs (SEO software)Branded web mentions correlate 0.656 to 0.709 with AI visibilityCorrelation, not cause; brands with Domain Rating above 40 only
Profound, 2026Profound (AI visibility software)About 57% of citations from brand-owned sites; ChatGPT 47% brand and 30% earnedObservational; earned includes institutional sources and PR wire
Muck Rack, 2025 onMuck Rack (PR software)84% of citations from earned mediaNo definition of earned media; no method stated
Yext, 2025Yext (listings software)86% from brand-managed sourcesFour industries and three models; its category shares add to 94%

When does your own site do the work, and when does PR?

Your own pages do more when AI knows little about you, and third-party coverage does more when it already knows you. Ahrefs tested this with 34 self-promotional pages and 9,886 AI answers between February and May 2026. For a little-known conference brand, 82% of new mentions came from answers that cited the authors' own pages. 6

For the established product, only 6% of new mentions came from answers citing the authors' pages, and 94% came from third-party content. 6 The authors read this as self-promotion working best where AI has a gap in its understanding. It is one small test by a firm that sells the tools involved, so treat it as a lead, not a rule.

Industry matters too. In Profound's data, the median SaaS company got only 11.4% of its citations from earned media, while the median pharma and biotech company got 59%. 3 ChatGPT leaned on earned media most, at 30% of its citations, and Gemini leaned on brand sites most, at 69%. 3

Decision rules. These are our suggestions, built on the studies above.
Your situationStart withWhy
AI never names youClear own pages: what you do, for whom, with factsOwn pages filled the gap for the little-known brand 6
AI names rivals and not you, and your press is thinIndependent coverage and honest review presenceThe established brand drew 94% of new mentions from third-party content 6; one small vendor test, so treat it as a starting hypothesis
AI names you but gets facts wrongCorrect your own pages and profiles firstThese are the sources you control directly
You sell in a field where earned media dominatesPress, analysts and institutionsPharma and biotech median earned share was 59% 3

Which outlets and sites do AI engines cite most?

A short list of sites dominates, and it is not only news. In Ahrefs' September 2026 snapshot of US queries, Reddit led ChatGPT's most-cited domains at 16.8% mention share, then Wikipedia at 7.0%, Consumer Reports at 3.7% and Forbes at 3.1%. 7 Mention share here is each site's share among the top 50 sources only, and the article does not say how many queries it covers. 7

Being retrieved is not the same as being cited. In an Ahrefs study of 1.4 million prompts from February 2025, ChatGPT cited 12.01% of the news pages it retrieved, against 88.46% of pages from general search results. 8 Many retrieved pages may never have been opened, so the gap is not a pure quality test, and results may differ for current versions. 8 A press hit gives ChatGPT something to read, but it does not promise a citation.

So pick outlets for reasons beyond the logo. Run your own prompts, note which domains each answer cites, and pitch the ones that appear for your topic.

What counts as earned coverage, and where is the legal line?

Independent means someone else chose to cover you without being paid. Wikipedia's rules for companies set a useful standard: they presume notability only after significant coverage in multiple reliable sources independent of the subject. They say press releases and paid or sponsored articles do not count. 9 AI engines do not follow Wikipedia's rules, but the test shows what editors treat as real independence.

Paid placements raise legal duties. The FTC says a connection between an endorser and a brand that consumers would not expect should be disclosed clearly, and that your company is ultimately responsible for what others do on its behalf, including an outside firm. 10 The Endorsement Guides do not have the force of law themselves, but practices that conflict with them can lead to enforcement. 10 This is not legal advice, and we have no data on how often AI cites sponsored posts.

What does a PR plan for AI visibility look like?

Start from the prompts, not the press list. The sheet below uses a made-up 40-person company that sells invoice software to mid-size firms. Every number in it is invented to show the method.

Illustrative monthly tracking sheet for an invented company
  • Prompt set: 20 buyer questions, for example “best invoice automation software for mid-size firms”.
  • Run monthly in ChatGPT, Gemini, Perplexity and Google AI Mode. Record: named? position? sources cited.
  • Month 1 result: named in 4 of 20 answers. Cited domains: 2 trade sites, 1 review site, 1 analyst blog.
  • Month 1 action: pitch both trade sites a data story. Ask five customers for honest review-site feedback.
  • Month 3 check: count again. Did named answers rise above 4 of 20? Did a pitched outlet appear among the sources?

How do you measure PR's effect on AI answers?

Use a fixed prompt list and compare before and after. It will not prove that PR caused a change, because other things move at the same time, but it is the best evidence most teams can get. Track these five items:

  • Named rate: the share of prompts where an answer names your brand.
  • Source mix: which domains each answer cites, and whether your pitched outlets appear.
  • Accuracy: whether the answer describes you correctly.
  • Branded search and direct visits, which are noisy but real signals of awareness.
  • A holdout: prompts about a product you did not promote, to see background drift.

How long does coverage keep helping?

Expect it to fade, so treat a feature as a start. Profound tracked 883,000 pages across seven answer engines and found the median page lost half its citation share within 11 days of its peak. 11 About 22% of page and engine pairs were sustained, with median half-lives of 27 to 29 days. 11

News, announcement and research-report pages had the shortest half-life of Profound's four page families, at a median of 27 days, though only about a day shorter than the others. 11 That fits PR's rhythm: a story spikes and settles. Keep your own answer pages current so the facts stay available between stories. Our guide to earning brand mentions with digital PR covers the pitching side, and AI SEO vs digital PR compares the two.

What are the limits of this evidence?

Most of it is correlation or observation, run by firms that sell related tools. 1345 We found no study that isolates a PR campaign's effect on AI answers by comparing treated and untreated brands.

Chalk Labs offers AI SEO and PR services, (our PR work is led by co-founder Shilika Jain), so we have a stake in this topic. Our published search cases measure Google results, not AI citations won through PR. We do not guarantee that coverage will change what any AI assistant says about you. The wider debate is in brand mentions vs backlinks.

Frequently asked questions

Does PR help you show up in ChatGPT?

Probably, but indirectly and without proof. Ahrefs found branded web mentions correlated with AI visibility at 0.664 for ChatGPT, and it warns that correlation isn't causation. 1 ChatGPT also cited only 12.01% of the news pages it retrieved in a 1.4 million prompt study. 8 Coverage gives ChatGPT more to read, not a guaranteed citation.

Is PR better than SEO for AI visibility?

Neither wins alone in the evidence. Profound found brand-owned sites supplied about 57% of citations overall, while earned media mattered more in some fields and in ChatGPT. 3 Plan for both: clear own pages and independent coverage. Our page on AI SEO vs digital PR compares them in more detail.

Do paid or sponsored articles count?

AI can read them, but we have no data on how often it cites them. They do not count as independent coverage under Wikipedia's notability rules, and the FTC expects clear disclosure of paid connections. 910 Label sponsored content as sponsored, and keep records of what you paid for.

How do I find which outlets AI cites for my topic?

Run 20 real buyer questions through four assistants and record the domains in each answer's sources. Count which appear most. Ahrefs' monthly lists show broad leaders such as Reddit and Wikipedia, but your topic may differ. 7 Repeat monthly so you can see change.

Are press releases useful for AI visibility?

We found no study that isolates press releases. Wikipedia treats them as non-independent material, so they do not build notability there. 9 Use a release to publish clear facts on your own site and wire, and treat independent stories as the separate goal.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied) · AhrefsOpened October 20, 2026. Supports: Published 12 December 2025, 75,000 brands (Domain Rating above 40): branded web mentions correlate 0.664 (ChatGPT), 0.709 (AI Mode), 0.656 (AI Overviews); Domain Rating 0.266, 0.285, 0.326; authors note correlation isn't causation.
  2. An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied) · AhrefsOpened October 20, 2026. Supports: Published 26 May 2025: branded web mentions 0.664 and backlinks 0.218 correlation with AI Overview mentions; branded web mentions defined as the brand name anywhere on the web, linked or unlinked; correlation not causation.
  3. Where do AI citations come from? · ProfoundOpened October 20, 2026. Supports: 11.84 billion citations, 29 industries, 8 AI models, 16 April to 16 July 2026: brand sites about 57% of citations; ChatGPT 47% brand and 30% earned; Gemini 69% brand; median earned share 11.4% in SaaS and 59% in pharma and biotech; earned defined as third-party editorial coverage, institutional sources and PR wire.
  4. Generative Pulse · Muck RackOpened October 20, 2026. Supports: Vendor page stating 84% of citations in AI answers come from earned media, from more than 25 million prompts since July 2025; the page does not define earned media or state a method; ChatGPT, Gemini and Claude appear only as integrations, not as the systems behind the figure.
  5. Yext Research: 86% of AI Citations Come from Brand-Managed Sources · YextOpened October 20, 2026. Supports: Release of 9 October 2025: 6.8 million citations from ChatGPT, Gemini and Perplexity in four industries (1 July to 31 August 2025); 86% brand-managed, though listed shares (44%, 42%, 8%) sum to 94%; reviews and social counted as brand-managed; vendor study.
  6. Self-Promotional Content Works, Until It Backfires (AI SEO Experiment) · AhrefsOpened October 20, 2026. Supports: Published 6 July 2026: 34 self-promotional pages, 9,886 AI answers, 7 February to 31 May 2026; 82% of new mentions of a little-known brand came from answers citing the authors' pages; for the established brand 6% did and 94% came from third-party content; small test by a vendor.
  7. The 50 Most-Cited Websites in ChatGPT (September 2026) · AhrefsOpened October 20, 2026. Supports: Updated 2 September 2026, US queries: Reddit 16.8%, Wikipedia 7.0%, Consumer Reports 3.7%, Forbes 3.1% mention share; share measured among the top 50 sources only; query count not disclosed.
  8. Why ChatGPT Cites One Page Over Another (Study of 1.4M Prompts) · AhrefsOpened October 20, 2026. Supports: Published 15 April 2026: retrieved URLs from general search were cited 88.46% of the time, news 12.01%, Reddit 1.93%; many non-cited URLs may never have been opened.
  9. Wikipedia:Notability (organizations and companies) · WikipediaOpened October 20, 2026. Supports: Presumed notability requires significant coverage in multiple reliable secondary sources independent of the subject; press releases and paid or sponsored articles do not count.
  10. FTC's Endorsement Guides: What People Are Asking · U.S. Federal Trade CommissionOpened October 20, 2026. Supports: Material connections between endorsers and brands should be disclosed clearly when consumers would not expect them; the company is ultimately responsible for what others do on its behalf; the Guides (16 CFR Part 255) lack the force of law but conflicting practices can lead to Section 5 enforcement.
  11. The half-life of an AI citation is 11 days · ProfoundOpened October 20, 2026. Supports: 883,000 pages across seven answer engines (8 September 2025 to 8 September 2026): median page loses half its citation share within 11 days of peaking; 22% of pairs sustained with median half-lives of 27 to 29 days; news, announcement and report pages faded fastest.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.