Google cites itself #2 in AI Mode, data shows
Google's own properties hold the top two citation spots in AI Mode. For brands, the competition was never just with each other.
Key takeaways
- Google and YouTube, both Google-owned, hold the top two citation positions in AI Mode.
- Business Profile cards and Product Knowledge Panels drive Google's self-citation, not editorial content.
- Citation share is zero-sum: every Google-hosted card displaces a brand's owned-domain content.
- Brands with incomplete or inconsistent Google entity footprints lose AI Mode visibility before a user reaches their site.
- Treating Google Business Profiles as primary content real estate is now a prerequisite for AI search visibility.
Google's own properties account for the second-highest share of citations in AI Mode, beaten only by YouTube. Search Engine Journal reports this finding from a domain-level citation analysis, with Google's Business Profile cards and Product Knowledge Panels driving the bulk of those self-referential links.
Read that again slowly. Google built an AI answer engine. Then it cited itself more than any third-party publisher on the open web.
The mechanism matters more than the ranking
This is not an accident of algorithm. Google's Knowledge Graph has always privileged structured, entity-rich data that Google itself hosts. Business Profiles, Shopping panels, local cards: these are Google products, governed by Google's data standards, updated through Google's own interfaces. When AI Mode assembles an answer, it pulls from the same corpus that powers traditional search, and that corpus is heavily weighted toward Google-verified entities. The result is a citation table where Google competes against the open web and wins silver.
YouTube taking gold is equally instructive. Google owns YouTube. So the top two cited domains in Google's AI Mode are both Google properties. Third-party publishers, brand websites, and independent sources are fighting for position behind a company that controls the playing field and has put its own assets on it.
For brands in financial services, industrial manufacturing, or the UN system, this creates a structural problem. Their authoritative content sits on independent domains. ISO standards live on iso.org. World Bank research lives on worldbank.org. A Holcim sustainability report lives on holcim.com. None of these benefit from the preferential structured-data pipeline that feeds Google's own properties. They must earn citations through content signals that Google's first-party assets bypass entirely.
What Google's self-citation actually displaces
Citation share is a zero-sum game inside any single AI response. Every slot Google's Business Profile card occupies is a slot a brand's owned content does not. For queries with commercial or local intent, where Business Profiles are most prominent, this displacement is direct and measurable. A user asking AI Mode about a construction materials supplier in Germany gets a Google-hosted card before they see anything from the supplier's own site.
The deeper implication is that the AI Mode citation pool may be narrowing precisely where brands assumed they had the most to gain. Early thinking about generative search held that long-form, expert content would finally beat thin pages. That remains partially true for informational queries. For navigational and commercial queries, the data suggests the answer is increasingly a Google property, not the brand itself.
This matters acutely for B2B brands whose purchase consideration cycle runs through multiple touchpoints. A procurement officer at a multilateral institution using AI Mode to shortlist suppliers will surface Google-hosted entity cards before she reaches the supplier's own domain. If that card is incomplete, outdated, or simply less prominent than a competitor's, the brand has lost the first moment of AI-mediated consideration without ever knowing it happened.
Where brands can actually compete
Google Business Profiles and Knowledge Panels are not static. They are editable, verifiable, and responsive to structured data quality. Brands that have treated these as secondary to their own site architecture have, in effect, ceded the most-cited territory in AI Mode to Google's defaults.
The practical consequence: any brand serious about AI visibility must now treat its Google-hosted entity footprint as primary content real estate, not a directory listing maintained by an intern. Product Knowledge Panels require consistent structured data signals across the web, accurate entity disambiguation, and active verification. Business Profiles need current information, category precision, and attribute completeness.
For large industrials and financial institutions with complex product portfolios, this is non-trivial. A single corporate entity may have dozens of Business Profiles across regions, each feeding separately into AI Mode's citation pool. Inconsistency across those profiles creates fragmentation: AI Mode cites a partial or conflicting version of the brand rather than a coherent one.
The brands that emerge from this structural shift with intact AI visibility will be those that recognised Google's self-citation for what it is: not a quirk of early-stage AI search, but a preview of a permanently tiered citation ecosystem where Google-hosted content gets first call.