Getting Named by AI: The New Metric Dividing Brands and Publishers

Leave a Comment / AI Playbook / By

Are AI Citations Worth Chasing? What the Research Actually Shows

Brands are being told that getting cited by AI is the new first page of Google. Publishers are watching their traffic collapse because of it.

Both things are true. And that contradiction is at the heart of one of the most consequential shifts in the history of digital media.

AI citations — the references that platforms like Google AI Overviews, ChatGPT, and Perplexity attach to their generated answers — are simultaneously a visibility opportunity for brands and an existential threat for the publishers whose content makes those citations possible. The research is now substantial enough to say who is benefiting, who is not, and why the split is structural rather than accidental.

How AI Search Is Changing Online Visibility

Search has operated the same way for two decades. A user types a query. An engine returns ranked links. The user clicks.

That model is changing faster than most of the web has adjusted to.

Platforms like Google AI Overviews, ChatGPT Search, Perplexity, and Microsoft Copilot now respond to queries by generating a direct answer, synthesizing information from multiple sources into a single conversational response. When they attribute that material, the attribution is called a citation. When they do not attribute it, the source website receives neither credit nor traffic.

Zero-click searches rose from 56 percent of queries in May 2024 to nearly 69 percent by May 2025. For publishers, that trajectory has been devastating. For brands trying to reach buyers, it has created a new and poorly understood visibility problem: how do you get cited by AI, and does it actually help when you do?

The answer depends almost entirely on who is asking.

What Is an AI Citation and How Does It Work?

How AI Platforms Decide Which Sources to Cite

AI search platforms do not rank pages the way traditional search engines do. They use a process called Retrieval-Augmented Generation, or RAG, retrieving relevant material from an index and then using a large language model to synthesize it into a response. When they attribute the material they used, that attribution is a citation.

A page ranked sixth in traditional Google results might receive more AI citations than the page ranked first. Being referenced matters more than being ranked. That single inversion upends two decades of SEO logic.

Citation behaviour also varies dramatically by platform, and the differences matter strategically:

  • Perplexity displays citations prominently after every claim and favours niche expert sources and review platforms
  • Google AI Overviews shows a small set of expandable source links and has historically favoured top-ranking pages, though that correlation is weakening
  • ChatGPT references sources in-text when its browsing feature is active and leans toward high-authority domains and structured listings

A Yext analysis of 6.8 million citations across ChatGPT, Gemini, and Perplexity found that 86 percent of citations come from sources brands already control, including websites, verified directory listings, and structured data. Reddit and similar forums accounted for just 2 percent of citations once location context and query intent were applied.

AI Mentions vs Citations vs Recommendations: What Is the Difference?

These three terms are used interchangeably across the industry. They describe meaningfully different outcomes.

  • Mention — a brand name appears in an AI-generated response with no source attribution
  • Citation — a platform attributes a specific claim or piece of content to a named source, with or without a visible link
  • Recommendation — an AI platform names a brand as the preferred answer to a buyer’s specific question

BrightEdge analysis of thousands of ChatGPT prompts found that ChatGPT mentions brands 3.2 times more often than it cites them. Nearly half of prompts contain zero brand mentions at all. Commercial intent language such as “deals” or “where to buy” drives four to eight times higher mention rates than informational queries.

AirOps research across 45,000 citations found that brands earning both a citation and a mention were 40 percent more likely to resurface in subsequent queries than those earning citations alone. The combination of signals matters more than either one individually.

Do AI Citations Actually Drive Traffic?

Why Citation Rates Are Rising While Click Rates Fall

The data presents a genuine puzzle that sits at the centre of the AI citation debate.

AI search usage is growing fast. ChatGPT receives over 4.5 billion monthly visits, and Perplexity processes more than 500 million searches per month. Yet the actual clicks that result from those interactions remain stubbornly small.

Only 1 percent of searches resulted in users clicking a link within a Google AI Overview, according to Pew Research Center’s analysis of 68,879 unique Google searches. Users who encountered an AI summary clicked on traditional search results just 8 percent of the time, compared to 15 percent without one. They were also more likely to end their browsing session entirely.

Organic CTR for queries featuring AI Overviews dropped 61 percent between June 2024 and September 2025, based on Seer Interactive’s analysis of 25.1 million impressions across 42 organizations. Ahrefs research from December 2025found AI Overviews now correlate with a 58 percent reduction in click-through rates for position-one content.

AI search is reaching more people. It is sending fewer of them anywhere.

When AI Citations Do Drive Measurable Results

The picture is not entirely grim. The key distinction is relative, not absolute.

Being cited does not generate substantial traffic on its own. But it generates significantly more than not being cited on the same query. Brands cited in AI Overviews earn 35 percent more organic clicks and 91 percent more paid clicks than non-cited competitors on identical queries, according to Seer Interactive. The gap has widened as overall click-through rates have declined.

The traffic that does arrive tends to be more qualified. Buyers reaching a website via AI search have typically already completed their research inside the AI interface. They arrive with intent formed.

This reframing is important. It explains why brands are investing in AI visibility even when the click numbers do not yet justify it on a surface reading. And it explains why publishers, whose business model depends entirely on those clicks, are responding with legal action.

“We’re definitely moving into the era of lower clicks and lower referral traffic for publishers.” — Stuart Forrest, global director of SEO digital publishing, Bauer Media, speaking to the BBC

What Does the Research Say About AI Citations?

What the GEO Research Found About AI Visibility

The academic foundation for much of the current AI citation debate comes from a peer-reviewed study published at the KDD 2024 conference. Researchers from Princeton University, Georgia Tech, the Allen Institute for AI, and IIT Delhi tested nine content optimization strategies across 10,000 queries to measure what made content more likely to appear in generative engine responses.

The results were clear. Visibility in AI-generated responses could be improved by up to 40 percent using specific techniques:

  • Adding statistics from credible sources increased AI visibility by 22 percent
  • Including direct quotations from subject matter experts boosted it by 37 percent
  • Incorporating citations to authoritative references further improved citation likelihood

What did not work: keyword stuffing performed 10 percent worse than the baseline. The signals that predict AI visibility differ meaningfully from those that predict traditional search rankings.

The study also found an unexpected equalizing effect: the citation-focused method produced a 115 percent visibility increase for websites ranked fifth in search results, while top-ranked sites actually lost ground on average. Lower-ranked sites have the most to gain from optimizing for AI citation.

How Citation Signals Differ Across ChatGPT, Perplexity, and Google

A strategy that works on one AI platform may have no effect on another. Citation signals differ fundamentally by platform:

That last point demands attention. Ranking well on Google no longer guarantees AI visibility, even within Google’s own AI product.

AI citation behaviour is also highly inconsistent. There is less than a one-in-100 chance that ChatGPT or Google’s AI, asked the same question 100 times, will return the same list of brands in any two responses, according to SparkToro research from January 2026. When AI Overviews generate a new answer, 45.5 percent of citations are replaced with new ones.

“If you’re not in the search engine index, you can’t be in the AI summary citation.” — Jess Sholtz, former CMO at Ringier Media, speaking to Digiday

The floor is still traditional SEO. But the ceiling is something else entirely, and each platform is defining its own rules for who reaches it.

Why Publishers Are Losing Traffic to AI Search

How AI Overviews Are Affecting Publisher Traffic

For brands, AI citations represent an emerging opportunity. For publishers, they are the operating mechanism of an ongoing financial crisis.

Publishers whose revenues depend on advertising tied to pageviews have watched AI platforms summarize their content without directing meaningful traffic back to their sites. Some have reported CTR drops of up to 89 percent since Google rolled out AI Overviews to U.S. users in May 2024. The financial damage has been concrete:

  • Business Insider: 55 percent organic traffic decline, followed by 21 percent staff cuts
  • HuffPost: approximately 50 percent decline in search referrals
  • Digital Trends: from 8.5 million monthly clicks to under 265,000, a 97 percent collapse

Educational platforms have been hit particularly hard. Chegg reported a 49 percent decline in non-subscriber trafficbetween January 2024 and January 2025, coinciding directly with AI Overviews answering the study questions that previously drove students to the site. Chegg filed an antitrust lawsuit against Google in February 2025. Penske Media Corporation filed a separate antitrust suit in September 2025, arguing Google was “cannibalizing” publisher traffic. Over 50 AI-related copyright lawsuits have now been filed against major technology companies.

“Google’s AI results don’t just excerpt; they replace. Pew’s research shows many users stop at the AI answer. That’s direct substitution, without fair bargaining or licensing.” — Jason Kint, CEO of Digital Content Next, writing in DCN’s member analysis

Which Domains Benefit Most From AI Citations?

Beyond individual traffic losses, there are structural concerns about who the citation system rewards.

The top 20 domains capture 66 percent of all AI Overview citations, with Wikipedia, YouTube, Reddit, and Amazon accounting for a disproportionate share. The system rewards scale and incumbency. Smaller publishers trying to break through face a structurally disadvantaged starting position.

AI Overview content changes 70 percent of the time for the same query, and when a new answer is generated, nearly half of citations are replaced. Visibility is volatile by design.

Google has proposed licensing arrangements to compensate publishers for content access. Most have rejected the terms as inadequate. The UK Competition and Markets Authority has launched a formal investigation into whether Google is abusing its search monopoly in the process. Google has indicated it is exploring options that would allow publishers to opt out of AI Overviews without losing traditional search visibility, but such controls are not yet widely available.

The gap between how brands and publishers experience AI citations reflects a deeper structural problem. One side gains a new channel for influence. The other loses the traffic model that made content creation economically viable in the first place.

What Happens to AI Citations Next?

AI Search Regulation and Legal Challenges

The legal environment is moving fast and on multiple fronts simultaneously.

Infringement cases filed against AI companies more than doubled in 2025, rising from around 30 to over 70 active cases. Anthropic reached a $1.5 billion settlement in a class action brought by authors. The New York Times, Condé Nast, Vox, The Atlantic, and others have filed suits against OpenAI and Google. The UK’s competition regulator is actively investigating Google’s search practices. The EU AI Act, which came into effect for general-purpose AI models in August 2025, now requires AI providers to publish summaries of training data and respect publisher opt-out rights.

Publishers are exploring licensing deals, revenue-sharing citation models, and paywalled content tiers that AI scrapers cannot access. Whether any of these frameworks will meaningfully rebalance the relationship between platforms and publishers remains unresolved.

How Brands Are Measuring AI Search Performance

For brands, the strategic direction is clearer, though still evolving.

The marketers adapting most effectively have largely stopped measuring success through rankings and traffic volume alone. They are tracking citation frequency, share of voice across AI platforms, and how their brand is described inside AI responses, not just whether it appears. It is a shift from measuring distribution to measuring perception.

Only 16 percent of brands currently track AI search performance systematically, according to industry research. Conductor’s research found that 32 percent of digital marketing leaders now name generative engine optimization their top priority for 2026, with an average of 12 percent of 2025 digital budgets already allocated to GEO initiatives.

Gartner projects that by 2028, up to 25 percent of all searches will move to generative engines.

That forecast may prove conservative. What the research makes clear is that the transition is already underway, and it is not moving at the same speed for everyone.

For brands with the infrastructure to optimize across platforms, track citation frequency, and produce the structured, authoritative content that AI systems favour, the shift represents a genuine opportunity. For publishers whose business models were built on the reciprocal relationship between content creation and search traffic, it represents something closer to a forced renegotiation, one being contested in courtrooms and regulatory offices simultaneously.

AI citations are not a neutral metric. They reflect choices made by platform designers about whose content gets surfaced, whose authority gets recognized, and whose traffic gets kept. Those choices are still being challenged.

The question was never really whether AI citations matter. The question is who decides what they are worth, and to whom.

Related Posts