Muck Rack’s May 2026 analysis of 25 million AI citations found earned media accounts for 84%, while paid media accounts for just 0.3%. This guide covers why AI engines structurally prefer earned media over owned content, what the brand mention data shows about AI citation probability, which content formats and publication types earn the most citations, and how to build the earned media programme that moves AI citation rates in 2026.
Earned media accounts for 84% of all AI citations. Muck Rack’s May 2026 Generative Pulse study analysed more than 25 million links across ChatGPT, Claude, and Gemini in 17 industries and found the same pattern across three consecutive editions: earned media at 82% to 89%, paid media at just 0.3%. Brands with genuine earned media earn AI citations. Those without are largely absent from AI-generated answers, regardless of how strong their owned content is.
Key takeaways
- Muck Rack found earned media accounts for 84% of all AI citations
- Brands in the top 25% for web mentions earn 10x more AI citations
- Brand mentions predict AI visibility three times better than backlinks
- Journalism accounts for 27% of AI citations and 49% on time-sensitive queries
We ran an AI citation audit for a B2B software brand last month. Despite solid SEO health, it appeared in AI-generated answers for just two of the fourteen category queries we tracked. Both citations pulled from a year-old TechCrunch piece and a G2 review the brand didn’t know existed. Our ContextualJourney™ platform maps exactly where those gaps sit before we recommend anything.
Earned media for AI citations: why AI engines cite what they cite
AI engines retrieve information from sources they’ve learned to trust, not through keyword matching. Generative AI tools learn during training which types of sources are reliable and which are self-serving. Third-party pages pass the credibility test because they come from parties with no direct financial interest in the subject. Brand-owned content fails the same test.
AI search engines show systematic bias toward earned media over brand-owned and social content, according to University of Toronto research. The researchers concluded the primary strategy is to dominate earned media to build AI-perceived authority. Fullintel and University of Connecticut research independently found 89% of AI-cited links were earned media and 95% were unpaid.
The pattern across three consecutive editions suggests this is structural, not a model quirk. AI engines treat brands with consistent earned media coverage as authoritative. Brands relying on owned content find those inputs don’t translate into AI citation outcomes. Greg Galant, CEO of Muck Rack, put it plainly in Muck Rack’s Generative Pulse: for communications teams, earning coverage in the right outlets has real consequences beyond traditional metrics.
How AI systems recognise and cite earned media
AI systems process text from across the web during training, learning which types of content appear in contexts associated with trust, accuracy, and editorial credibility. Earned media carries specific signals: named journalists, editorial oversight, correction policies, and no financial relationship between publisher and subject. A feature article about a brand in a trade publication reads very differently from the same brand’s own blog post.
When AI cites a brand in response to a buyer query, it’s almost always drawing from third-party sources rather than the brand’s own domain. Earned media provides third party validation that AI systems treat as a credibility signal in ways that owned content structurally cannot. Earned media distribution across multiple independent publications multiplies this effect.
Press coverage in industry publications and earned media mentions across third-party sites create the independent editorial record AI engines retrieve from for category queries. Brand visibility in generative search is built through media relations, PR strategy, and consistent editorial coverage.
AI citation sources: the platform breakdown
Each major AI engine sources its answers differently, but the preference for earned media is consistent across all of them:
| Platform | Citation behaviour | What earns citations |
|---|---|---|
| ChatGPT | Cites in 96% of responses, avg 5 citations | Wikipedia, industry publications, third-party editorial |
| Gemini | Cites in 82% of responses, avg 8 citations | Brand-owned structured content alongside earned editorial |
| Claude | Cites in 55% of responses, avg 13 citations | High-credibility academic and editorial sources |
| Perplexity | Real-time retrieval from indexed web content | Trade press, review platforms, Tier-1 earned coverage |
| Google AI Overviews | Journalism doubles for time-sensitive queries | News coverage and category-native editorial media |
Google AI Mode citations show similar concentration toward editorial and third-party sources. Google AI Overviews now trigger on approximately 48% of all tracked queries according to BrightEdge’s analysis. AI Overview citations from outside the organic top 100 are dominated by YouTube at 18.2% (Ahrefs, March 2026), confirming video has become a significant earned media citation surface.
Brand mentions and AI visibility: what the data shows
Brand mentions (linked and unlinked references to a brand name across third-party web content) are the strongest measurable predictor of AI citation rates. The correlation between brand web mentions and AI Overview visibility stands at r=0.664 according to Ahrefs and LumenGEO’s 2026 analysis. Backlinks correlate at r=0.218. Domain authority correlates at r=0.18.
Evertune.ai’s analysis of 75,000 brands found the top 25% for web mentions earn over 10x more AI citations than the next quartile. The top quartile averages 169 AI mentions versus 14 for the next tier. The gap compounds: more mentions produce more AI citations, which produce more branded searches, which signal authority to AI systems, which produce more citations.
Why brand mentions predict AI citation rates
Brand mentions work as an AI citation predictor because they’re a proxy for something AI systems genuinely value: evidence that independent sources are discussing, verifying, and referencing the brand. When multiple editorial publications, review sites, and industry forums reference a brand in similar terms, that consensus tells AI models what the brand does and that it can be trusted.
The mechanism is machine relations: the relationship between a brand and the AI systems that learn about it from the web. A brand that appears consistently across trade publications, industry forums, and editorial blogs builds a richer machine-readable identity than one that lives primarily in its own content. Research from Evertune.ai, LumenGEO, and Ahrefs puts earned media density above domain authority, backlinks, and keyword optimisation as a predictor of citation probability.
Web mentions versus backlinks for AI citations
The r=0.664 vs r=0.218 gap between mentions and backlinks changes which activities deserve strategic priority. Backlink acquisition, guest posting, and traditional SEO tools all build the metric that correlates least strongly with AI visibility. Earned media programmes that generate brand mentions across independent publications build the metric that correlates most strongly.
This doesn’t mean backlinks are irrelevant. They still correlate with traditional search rankings and domain authority signals that some AI platforms weigh. But for brands investing in AI visibility, the return on earned media coverage is materially higher than the return on equivalent link-building investment. Our digital PR and AI search guide covers how to build the earned media programme that moves AI citation rates.
The role of journalism in AI citations
Journalism accounts for 27% of AI citations, a figure steady at 25-27% across all three editions of Muck Rack’s study. For time-sensitive queries, journalism’s share rises to approximately 49% according to analysis of Muck Rack’s citation data. Tier-1 publications (the New York Times, Wall Street Journal, Business Insider) carry disproportionate citation weight because they’ve passed the editorial credibility threshold AI models use.
Why editorial media placements carry citation weight
Editorial media earns citation weight through four signals AI models trust: editorial oversight, named journalists, correction policies, and established reputations for factual accuracy. A news article in a major publication has passed an editor’s review before publication under that outlet’s editorial standards. Embargoed briefings allow journalists time to prepare richer coverage, producing more durable AI citations than a brief mention.
BuzzStream’s January 2026 study of 4 million citations from 3,600 AI prompts across 10 industries found editorial blog and content pages account for 53.46% of all AI citations. News pages account for 14.09% and social content for 8.71%. The dominant citation class is substantive editorial content that addresses a question in depth.
Industry publications and third-party editorial coverage
Beyond Tier-1 journalism, industry-specific publications carry significant citation weight for category-level AI queries. Third-party editorial coverage in trade press produces highly targeted AI citations, reaching buyers when they’re actively evaluating options in a category. A strong narrative around AI research and category expertise is vital for earning coverage in the publications AI engines retrieve from most consistently.
Alongside traditional editorial media, YouTube has become a significant citation surface. Bluefish data reported by Adweek in January 2026, drawn from 6.1 million citations across four independent research firms, found YouTube appears in 16% of LLM answers, overtaking Reddit at 10%. YouTube accounts for 18.2% of AI Overview citations sourced from outside the organic top 100, according to Ahrefs March 2026 research.
Conference talks, product demos, and expert interviews on YouTube generate AI citations independently of text-based coverage.
What the AI citations come from: the complete picture
Understanding which content types AI cites most frequently is as strategically important as understanding why earned media dominates. The BuzzStream January 2026 dataset covers 4 million citations across 10 industries.
What AI answers reveal about content format
Editorial blog and content pages account for 53.46% of all AI citations. Within that category, comparative content is the most cited at 26.92%, followed by market analysis at 23.06%, and definition and explainer content at 20.57%. Ranqo’s June 2026 study of 102 brands found best-of listicles account for 35.7% of content-level AI citations.
Citation behaviour splits clearly by format:
- Listicle and ranking formats perform strongly for evaluative queries
- News coverage earns citations for time-sensitive queries
- Press releases and owned content almost never earn direct AI citations
Measuring AI citation outcomes
Measuring earned media’s impact on AI visibility requires different tools from traditional PR measurement. AI mentions differ from citations: a brand can appear in AI answers without a linked citation, and both contribute to brand visibility in generative search. Tracking both through consistent prompt sets across ChatGPT, Perplexity, AI Overviews, and Google AI Mode maps earned media activity to citation outcomes.
Running 30 to 50 target prompts across major AI platforms weekly reveals which publications are appearing in citations for a brand’s core category queries. That data shows which media placements are producing direct AI citation outcomes and which are building brand visibility without yet appearing as citations.
Building the earned media presence that drives AI citations
Muck Rack, Evertune.ai, BuzzStream, and multiple 2026 AI citation studies all point to the same strategic priorities. Brands that earn AI citations consistently:
- Appear across multiple independent publications in their category
- Have earned coverage in high-authority outlets AI engines treat as credible references
- Generate enough organic third-party discussion that AI models have encountered them in multiple contexts
Earned media strategy for AI citation rates
An earned media strategy built for AI citation rates prioritises coverage breadth, because citation probability increases when a brand appears across multiple independent sources. The Stacker December 2025 analysis found that distributing content across a wide range of publications increases AI citations by up to 325%. A data-led campaign placed with twenty relevant publications produces more AI citation value than an exclusive placement with one major outlet.
Recency matters alongside breadth. Half of all AI citations in the Muck Rack study came from content published within the last 11 months. AI retrieval systems weight recent content, and consistent earned media output keeps a brand’s citation footprint current. Earned media distribution through consistent PR strategy is the operational mechanism that builds and sustains AI citation rates over time.
Brand mentions and digital PR
Building brand mention density across digital channels is the most direct way to move AI citation rates. Every mention in a publication, review platform, or editorial adds a data point to the reference web AI systems draw on. Digital PR is the practice of building that density systematically. The target is the specific publications and platforms AI engines retrieve from for the brand’s core category queries.
User-generated content, community discussions, and forum mentions also contribute to brand mention density. Community sentiment and engagement on forums are important for AI models mining conversational data. Reddit, specialist communities, and industry forums all appear consistently in AI citation sources, and PR teams building AI citation strategy need to include them in their target list.
If your brand isn’t earning AI citations, here’s what the data says
The brands that earn consistent AI citations share a common thread: they’ve built the kind of earned media presence that AI systems were trained to trust. They appear in editorial media, in independent reviews, in analyst reports, and in community discussions. If AI-generated answers in your category describe competitors accurately and describe your brand inaccurately or not at all, the earned media footprint that AI systems are retrieving from is your competitors’, not yours.
Talk to the FirstMotion team to map exactly where your brand sits in AI-generated answers for your core category queries and which earned media activities will close the gap most efficiently.
Most brands we audit appear in fewer AI-generated answers than they expect, and the gap is almost always an earned media gap, not a content gap. Our ContextualJourney™ platform maps which publications AI engines are retrieving from for your category queries before we recommend anything.