Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Google AI Overviews Go Global — and Structured Data Is the New Authority Signal
Google AI Overviews serve over one billion users monthly across 200+ countries. The data shows structured markup and entity clarity now outweigh backlinks for AI citation selection.
Google's AI Overviews are no longer an experiment. They are live in more than 200 countries and territories, available in over 40 languages, and serving more than one billion users monthly. What was once Search Generative Experience — a limited beta that drew skepticism — is now the default way Google surfaces answers for a growing share of queries.
The expansion itself is not surprising. What is notable is the signal it sends about what Google now values in the content it cites. The emphasis has shifted, measurably, from backlink-based authority toward structured data and entity clarity. For anyone producing content that needs to be found, this is not a minor update. It is a structural change in how visibility works.
The Scale of Deployment
Google's most recent expansion, announced in May 2025, brought AI Overviews to over 200 countries with support for Arabic, Chinese, Malay, Urdu, and dozens of other languages. In the US and India, the feature is driving a 10% increase in search usage for queries that trigger AI Overviews. More recently, the rollout has extended into European markets including Germany, Belgium, Ireland, Italy, Austria, Poland, Portugal, Spain, and Switzerland.
A custom version of Gemini 2.5 now powers the more complex queries. Each AI Overview contains roughly 169 words and 7.2 links on average, according to analysis by Mike Khorev. These are not traditional blue links. They are curated citations embedded within generated text — and the criteria for selection look different from what most publishers are used to.
Structured Data Is the New Authority Signal
The most significant pattern in the data is the declining correlation between traditional domain authority and AI citation selection. Domain Authority's correlation with rankings has dropped to r=0.18, down from r=0.26. That is a meaningful decline in a metric that has anchored SEO strategy for over a decade.
This is not a marginal advantage. It is a structural one. And it aligns with what both Google and Microsoft have publicly stated: that they use schema markup directly in their generative AI features because it is efficient, precise, and easy for machines to process.
Why Entity Clarity Matters More Than Backlinks
AI Overviews do not evaluate pages the way traditional ranking algorithms do. Google's AI systems break content into chunks, not pages. They evaluate individual passages for semantic completeness, entity density, and topical coverage. Research indicates that self-contained passages of 134–167 words with 15 or more recognized entities per 1,000 words perform best.
This means a well-structured page from a lower-authority domain can outperform a high-authority page that lacks clear entity resolution. In fact, 47% of AI Overview citations come from pages ranking below position five in traditional organic results. The playing field has shifted — not toward obscure sites, but toward content that machines can parse unambiguously.
The emerging discipline of Generative Engine Optimization (GEO) reflects this shift. Where SEO optimized for crawlers and link graphs, GEO optimizes for language models and knowledge graphs. The toolkit is different: JSON-LD schema, entity-rich prose, and explicit topic relationships replace keyword density and anchor text strategies.
What Google's Documentation Tells Us
Google's own search documentation and updates reinforce this direction. Schema markup is no longer positioned as a tool for earning rich snippets. It is described as the way AI systems interpret meaning and relationships. The implication is clear: machine-readable entity graphs are becoming essential infrastructure, not optional enhancement.
The SchemaApp analysis of 2025 trends noted that Google regularly retires old or redundant markup types while doubling down on the ones that feed AI understanding. The direction is toward fewer, more meaningful structured data implementations — not more markup for its own sake.
The Citation Advantage
Pages cited in AI Overviews see measurable traffic benefits. According to available analysis, cited pages gain 35% more organic clicks and 91% more paid clicks compared to uncited competitors on the same SERP. Content with authoritative citations embedded in structured data shows a 132% visibility increase.
The cosine similarity threshold matters too. Pages whose content achieves a similarity score above 0.88 with the AI-generated summary are selected 7.3 times more often. This is a proxy for topical alignment — content that directly and clearly addresses the query in machine-parseable terms wins.
What This Means for Publishers
The strategic implications are straightforward:
Implement comprehensive JSON-LD schema — Article, Organization, Person, FAQPage, and HowTo types are the baseline. Link entities to Wikidata where possible.
Write entity-dense, self-contained passages — 134–167 words per section, with clear topic sentences that an AI system can extract without surrounding context.
Prioritize entity resolution over keyword optimization — Make it unambiguous what your content is about and who is saying it.
Audit your structured data implementation — Schema that is technically valid but semantically thin will not help. The markup needs to express genuine relationships, not just metadata.
AI Overviews are not replacing organic search. They are reshaping which content gets selected for the most visible positions in it. The publishers who adapt their technical infrastructure to match how AI systems actually read content will capture the citation advantage. The ones who keep optimizing exclusively for backlinks will watch their visibility erode — not because backlinks are worthless, but because they are no longer sufficient.
James Calder is the editor of The Search Signal, covering AI-powered search, generative engine optimization, and the future of brand discovery.
Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Perplexity's Personal Computer turns a Mac mini into a persistent, always-on AI agent that works across your local files and apps around the clock — a new category of ambient computing.
