Digital PR and Brand Mentions for AI Visibility

The single strongest predictor of whether an AI engine cites your brand is not your backlink profile. It is how often independent, third party sources mention your brand by name, with or without a link. Research from Seer Interactive found that brands with active third party trust signals are cited in 75 percent of AI answers, compared to just 1 percent for brands without them, a 75 times gap that is among the largest signal to output multipliers documented anywhere in GEO research so far.

This guide covers why unlinked brand mentions and earned media now outweigh traditional link building for AI visibility, what the data shows about where AI citations actually come from, and how to run a digital PR program built for this shift. For the broader mechanics behind how models choose sources at all, see our guide on how LLM models decide which brands to mention.

Facts, figures and platform behavior in this article were checked in September 2026. Citation research in this area is still young and moves quickly, so treat specific percentages as directional rather than fixed.

For over a decade, the core currency of digital PR was the backlink. That is no longer the strongest signal AI systems actually respond to.

Source: Authority Tech, GEO statistics analysis, 2025

A correlation roughly three times stronger is a large gap in any research context. Ahrefs' same study found the top 25 percent of brands by mention volume averaged 169 AI Overview mentions, compared to just 14 for the next quartile down, a difference that compounds rather than scales linearly. Separate research from LumenGEO, drawn from more than 1,000 brand audits, arrived at the same 0.664 correlation figure independently, which is a meaningful sign this pattern is not an artifact of one dataset.

This is the part that surprises most PR and marketing teams, since an unlinked mention has historically carried no measurable SEO value. The mechanism is retrieval augmented generation, the underlying technique most AI search tools use to ground their answers in real, current web content. When a model retrieves information to answer a query, it does not need a hyperlink to find your brand, it needs your brand name to appear, in context, across enough independent sources that the model treats you as a known, credible entity worth referencing.

In practice, this means AI engines cross reference brand claims against independent sources before including them in an answer. Earned media provides exactly that independent verification layer. A brand that only ever talks about itself, on its own website, has no outside confirmation for a model to check its claims against, no matter how well written that owned content is.

The data on this is now large enough to be unambiguous. Muck Rack's May 2026 study analyzed more than 25 million links cited by ChatGPT, Claude and Gemini across 17 industries and found that earned media accounts for 84 percent of all citations, with journalism specifically contributing 27 percent of that total. Paid and advertorial content, by contrast, accounted for just 0.3 percent.

Sources: Muck Rack, "What Is AI Reading?", May 2026, and BuzzStream, January 2026

That last row is worth sitting with. A tactic many PR teams still lean on heavily, distributing a press release through a wire service, produced essentially no measurable AI citation value in BuzzStream's analysis, at 0.04 percent of citations. Meanwhile genuine editorial coverage, the kind a journalist chooses to write because the story or the data is actually interesting, accounts for the large majority of what AI engines actually cite.

Academic research backs this up independently. A University of Toronto study by Chen, Wang and colleagues found that AI engines show a systematic and overwhelming bias toward earned media over brand owned and social content, which is a stronger claim than a simple correlation, since it points to the pattern being a structural feature of how these systems weigh sources, not a coincidence of what happens to get published most often.

Part of why this works is that AI systems are, in effect, inheriting a trust hierarchy that already exists among human audiences. Nielsen data found that 92 percent of consumers trust earned media, meaning editorial coverage and independent recommendations, over all other forms of advertising, compared to just 41 percent who trust paid advertising. AI models trained on the same body of text that humans read and cite naturally absorb a similar hierarchy, which is one reasonable explanation for why the 84 percent earned media figure and the 92 percent consumer trust figure land so close together.

Worldcom Group's research reached a similarly striking conclusion from a different angle, finding that up to 90 percent of the citations driving brand visibility in large language models come from earned media specifically, positioning public relations, not content marketing or paid promotion, at the center of AI visibility strategy.

Earning a mention is only half the equation. The content around that mention still needs to be structured in a way a model can extract with confidence. Research from Princeton, Georgia Tech and IIT Delhi, presented at KDD 2024, found that adding statistics improved source visibility in generative engines by 41 percent, adding quotations improved it by 28 percent, and citing external sources improved visibility for lower ranked content by as much as 115 percent. The same research found that keyword stuffing actively hurt visibility, the opposite of what a decade of traditional SEO practice would predict.

This is why a purely technical fix, such as adding schema markup, is not a substitute for genuine earned coverage. An Ahrefs study tracking pages that added JSON-LD schema found no statistically significant change in AI citations on its own. Structured data still has real value for machine readability, but the larger, better documented levers for AI visibility remain earned media, brand mentions and genuinely extractable, data backed content, a pattern we cover in more detail in our guide on why your brand doesn't appear in AI search.

Not every AI model treats brand mentions the same way. One 2026 analysis found Claude mentions brands in 97.3 percent of its responses, the highest rate of any major AI model, and draws heavily from LinkedIn and professional network data specifically when doing so. That is a useful, actionable detail for B2B brands in particular, since it means a well maintained LinkedIn company page with regular thought leadership content is not just a marketing nicety, it is functioning as a real entity signal for at least one major model.

Publish original data on a regular cadence. A small, genuinely original data study, a customer survey, an industry analysis, or a proprietary benchmark, gives journalists a number worth quoting, and numbers are exactly what earn citations, per the Princeton and Georgia Tech research above. One modest study per quarter is a realistic starting cadence for most teams.

Pitch for relevance, not reach. Journalists reject pitches that do not fit their beat at a high rate, and prefer pitches backed by original data or research. A single, well targeted pitch to the right reporter with a genuine data point outperforms a broad blast to a long list.

Treat unlinked mentions as a real success metric, not a consolation prize. Given the mention to visibility correlation is roughly three times stronger than the correlation for backlinks, a mention without a link earned through genuine editorial coverage is not a missed opportunity, it is often the primary win.

Keep your entity presence current everywhere, not just on your own site. Consistent brand descriptions, complete LinkedIn and Google Business Profile data, and accurate directory listings all feed the same verification layer AI systems use to decide whether to trust a claim about your brand.

Refresh cornerstone content regularly. Research compiled in 2026 found pages that go more than three months without a meaningful update are over three times more likely to lose AI visibility.

Measure brand mention velocity, not just backlinks earned. Since mentions correlate far more strongly with AI visibility, a campaign report built entirely around domain rating and referring domains is measuring the wrong primary outcome for GEO purposes, even though those metrics still matter for traditional SEO.

Relying on wire distributed press releases. At 0.04 percent of citations, this remains one of the least effective tactics available for AI visibility specifically, even though it can still serve other communications purposes.

Treating GEO as a separate program from existing PR and earned media work. Brands that build a standalone AI visibility initiative disconnected from their PR strategy tend to create redundant effort without improving either outcome, since the underlying lever, earned third party mentions, is the same one PR teams have always pursued.

Assuming schema markup alone will move the needle. Structured data supports machine readability but has not shown a measurable, independent effect on AI citation rates in controlled testing. It belongs in a program, not in place of one.

Chasing volume of coverage over relevance and data quality. Keyword stuffed, low substance content actively hurt visibility in the KDD 2024 research, which suggests quantity without genuine informational value works against a brand rather than for it.

Our guide on tracking your brand across ChatGPT, Perplexity and Gemini covers setting up this kind of ongoing measurement across platforms.

Do unlinked brand mentions really matter if they carry no direct SEO value?

Yes, and increasingly more than backlinks do for AI visibility specifically. The correlation data is consistent across multiple independent studies, and the underlying mechanism, retrieval augmented generation checking your brand against independent sources, does not require a hyperlink to work.

Should we stop pursuing backlinks entirely and focus only on mentions?

No. Backlinks still matter for traditional SEO and referral traffic. The point is that a PR or content strategy built purely around earning links, while ignoring unlinked editorial mentions, is now optimizing for a weaker signal when AI visibility is the goal.

Is a press release ever worth doing anymore?

Wire distributed releases specifically show almost no measurable AI citation value. A press release that leads to genuine journalist pickup and independent editorial coverage is a different outcome entirely, and that coverage is where the real value sits.

How quickly does a digital PR campaign show up in AI citations?

Most digital PR campaigns produce measurable results within 3 to 6 months. Given that AI systems also weight freshness and re-crawl content on their own schedules, a single placement is best treated as one data point in an ongoing program rather than a one time event.

If you want to see how much of your current AI visibility is coming from earned mentions versus your own website, and where the biggest gaps are, you can check it directly.