The Discovery Layer Has Moved
The mechanics of how people find companies changed faster than most marketing organizations could react. Search sessions on AI platforms average six minutes, compared to seconds on Google, and the queries themselves have transformed — AI search queries average 23 words, while Google queries average four. People aren't typing fragments and scanning ten blue links anymore. They're describing their situation in full and expecting a synthesized answer, one they're inclined to trust as an endpoint rather than a starting point.
The consequence for brands is structural, not cosmetic. Zero-click search — where a user gets their answer without visiting any website — reached 58.5% of U.S. searches and 59.7% of EU searches in 2025, and inside AI Mode sessions specifically, that figure climbs past 90%. When the large majority of research ends without a single site visit, "getting found" stops being the point. What matters is whether you were part of the answer the AI actually gave.
A Completely Different Signal Set
The AI's answer draws on a completely different signal set than Google's ranking algorithm did. Research from GEO firm Brandlight found that the overlap between top Google links and the sources AI systems actually cite has fallen from roughly 70% to below 20%. Princeton and Georgia Tech researchers, publishing the foundational GEO study at KDD 2024, showed that content optimized with citations, statistics, and quotations could lift AI visibility by 30 to 40% — a lever almost entirely disconnected from traditional keyword ranking.
A more recent large-scale analysis of over 100,000 prompt responses across 100+ brands found that corporate websites account for only a portion of what AI engines cite; among non-corporate sources, YouTube, Reddit, editorial coverage, and Wikipedia dominate, and the single most-cited content format is the ranked "best-of" listicle.
Correlation Strength With AI Citation (Spearman Coefficient)
Perhaps the most counter-intuitive finding comes from a 2026 synthesis of six independent studies covering 680 million AI citations: brand mentions across the web correlate with AI citation at roughly three times the strength of backlinks. Links, the currency SEO was built on, still matter. But being talked about, consistently and by independent sources, matters far more to how AI represents you than being linked to.
Presence, Identity, Visibility — and Why Most Companies Only Have One
It's useful to separate what's actually happening into three layers. First, can AI identify your company at all — does it recognize your name, your products, your founders, and distinguish you from companies with similar names? Second, once it knows who you are, does it understand you correctly — your customers, your differentiation, your business model, your place in the market? Third, when a buyer describes their problem, does AI actually recommend you, or does a competitor with a thinner product but a richer public footprint get the nod instead?
Most companies have some presence. Far fewer have a coherent identity. Almost none have engineered their visibility on purpose. And this isn't a gap that closes itself over time — AirOps' 2026 State of AI Search report found that brands earning both mentions and citations in AI answers are 40% more likely to resurface across consecutive queries than brands with citations alone. Visibility compounds for the companies that have it and stays flat for the ones that don't.
This is also where the "invisible knowledge" problem becomes expensive. AI doesn't invent an understanding of your company from nothing — it assembles one from whatever fragments of your business have become public: a blog post here, a press release there, a founder interview, an outdated product page, a stray LinkedIn post. Meanwhile, the material that actually captures how your company thinks and what it has learned — customer conversations, sales calls, support interactions, product decisions, internal Slack threads, executive insight — sits inside the organization, doing no work for your AI identity whatsoever.
Where AI Brand References Actually Come From
Source: Cintra's 2026 analysis of over 21,000 brand mentions.
What "Fixing It" Actually Looks Like
The temptation is to treat this as a technical SEO problem — fix your schema markup, speed up your Core Web Vitals, add FAQ blocks. Those things help; one documented case study showed a B2B site's AI citation rate jump from 18% to 52% after cutting page load time from 4.8 seconds to 1.9 seconds, with Perplexity citations going from zero to 38 per 200 queries. But technical fixes only clear the runway. They don't give AI anything new to say about you.
The deeper fix is closer to what the Princeton researchers were pointing at: AI rewards specificity, authority, and evidence — statistics, quotations, structured explanations of how a problem actually gets solved. That's not something a technical audit produces. It's something that only comes from the substance of the business itself: the case study that explains exactly how a customer's problem was solved, the founder's actual reasoning about a market shift, the product decision explained in enough detail that it reads as expertise rather than marketing copy.
This is precisely the translation problem most marketing teams are not resourced to solve. They know the knowledge exists somewhere in the organization. They don't have a workflow for continuously turning a support ticket, a sales call transcript, or an executive's offhand insight into a structured, citable, AI-legible piece of public knowledge — at the volume and cadence AI systems seem to reward. eMarketer found that only 23% of marketers currently invest in AI-visibility measurement at all, even though 54% say they plan to build a GEO program within six months. The intention is there. The infrastructure is not.
Why YesPress Newsroom Is Built for This Exact Moment
YesPress Newsroom exists because closing that gap by hand — one blog post, one press release, one LinkedIn update at a time — cannot keep pace with how AI actually builds understanding of a company. It doesn't treat AI identity as a single campaign to run once and move on from. It treats it as infrastructure: a continuous system that measures how AI currently sees your company, finds where that picture is incomplete or wrong, and turns the knowledge your company is already generating into the structured, authoritative public record AI needs in order to represent you accurately.
That starts with measurement, because — as the law firm case study makes brutally clear — you cannot assume your search-engine performance tells you anything about your AI visibility. YesPress continuously tracks whether AI can reliably identify your company, how it currently describes you, whether that description is consistent across platforms, which sources it's actually pulling from, and how you stack up against competitors in the same category.
But YesPress's real differentiation is what happens after the diagnosis. Most GEO tooling stops at measurement and dashboards, leaving the actual work of changing the underlying content to marketing teams who are already stretched thin. YesPress Newsroom closes that loop. It takes the raw material every company already produces — a customer success call, a product roadmap conversation, a founder's answer in a media interview, a Slack thread where the team worked through a hard tradeoff — and turns it into structured stories built to be understood by both people and AI systems.
Every one of those published stories becomes a fresh, authoritative signal in the pool of material AI draws from when it constructs its understanding of your company — precisely the kind of earned, third-party-style visibility that the research shows AI weights three times more heavily than backlinks. Because YesPress runs this as a continuous loop — measure, identify gaps, transform knowledge into stories, publish, monitor how AI's understanding shifts, repeat — the AI identity it builds compounds, rather than decaying like a one-off content push.
Being Known Is the New Being Found
The uncomfortable truth in nearly every 2026 report on this shift is the same: the brands winning AI visibility right now are not necessarily the biggest or the most established. They are the ones whose public knowledge is rich, current, and consistent enough for AI to construct an accurate story from it. Amperity's 2026 Consumer Priorities Report found that more than half of consumers say generative AI has already led them to select a brand they otherwise wouldn't have considered — and found the experience satisfying enough to act on. That is the upside of this shift: AI is actively widening the consideration set for companies that show up well, not just protecting incumbents.
The companies that treat this moment as another SEO checklist item will spend the next few years optimizing metadata while their competitors quietly become the default answer to the questions their buyers are actually asking. The companies that understand AI identity as something to be deliberately built — continuously, from the real knowledge already inside their walls — will be the ones AI trusts enough to recommend.