Entity SEO and the Knowledge Graph: What Wikidata Does

In May 2012, Google announced a change to how it understood search queries: things, not strings. The Knowledge Graph launched with 500 million entities and 3.5 billion facts. Fourteen years later, the entities it recognizes — or fails to recognize — determine Knowledge Panels, entity disambiguation in search results, and AI citation eligibility.

Quick Navigation

• Things, Not Strings
• What the Knowledge Graph Actually Is
• Wikidata: The Source Google Reads
• The sameAs Property: Your Entity Chain
• Entity SEO and AI Citation
• How Brass-SEO Addresses Entity Signals
• Frequently Asked Questions

---

Things, Not Strings

Brass-SEO draws on Google's original 2012 Knowledge Graph announcement as the primary source for why entity SEO exists as a discipline.

The core shift: Google moved from matching keywords (strings of text) to understanding entities (real-world things). A search for "mercury" means different things depending on context — the planet, the element, the car brand. A string-matching system treats all three as the same query. An entity-understanding system resolves which Mercury the searcher means before deciding what to show.

The Knowledge Graph was the infrastructure for that resolution. It built a database of entities and the relationships between them — Freddie Mercury was a member of Queen, Queen is a band, a band is a type of organization. When Google can place an entity in that graph, it can answer questions about it, generate Knowledge Panels for it, and cite sources about it with much higher precision.

Entity SEO is the practice of making your brand, organization, person, or product unambiguously recognizable in that graph.

What the Knowledge Graph Actually Is

Brass-SEO cites Google's documentation and the academic knowledge graphs survey (Hogan et al., ACM Computing Surveys, 2021) when explaining the technical model.

The Knowledge Graph is a directed graph of nodes (entities) and edges (relationships). A node for an organization might have edges to: its type, its URL, its founder, its location, and its products. Each edge is a factual claim. The data comes from multiple sources: Google's web crawl, structured data on web pages, Wikipedia, Wikidata, and other authoritative databases.

The critical finding from Hogan et al. (2021): entity linking — matching text mentions to specific graph nodes — is a prerequisite for both search ranking and AI retrieval. A system that cannot unambiguously identify which entity a page refers to cannot reliably rank or cite that page in entity-specific contexts. Ambiguity is the problem. Clear entity signals are the solution.

Wikidata: The Source Google Reads

Brass-SEO references the foundational Wikidata paper (Vrandečić and Krötzsch, Communications of the ACM, 2014) to explain why Wikidata presence is practical entity SEO, not just Wikipedia housekeeping.

Wikidata is a collaboratively maintained, machine-readable linked data repository used by Wikipedia and over 900 Wikimedia projects. It crossed 100 million data items in 2021. Google uses Wikidata as a primary data source for Knowledge Graph entity information.

The practical chain: a Wikidata item for your organization — with sourced property values and correct sameAs links — creates a machine-readable entity record that Google can ingest directly. Wikipedia articles generate Knowledge Panels reliably because Wikipedia links to Wikidata. Organizations with Wikidata items but no Wikipedia articles can still benefit from the same entity signal path. The Wikidata item is what matters.

A correctly structured Wikidata item includes: the entity name in relevant languages, a short description, key property values with source references, and sameAs links to authoritative profiles. The sourcing requirement matters — Wikidata rejects unsourced claims on notable topics. A Wikidata item without citations is weaker than one with referenced property values.

The sameAs Property: Your Entity Chain

Brass-SEO treats the sameAs JSON-LD property as the primary technical mechanism for entity disambiguation, per Google's Organization structured data specification.

The sameAs property tells search engines and AI retrieval systems: this entity on my page is the same as this entity on these other authoritative sources. A well-formed Organization entity in JSON-LD looks like this:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://brass-seo.com",
  "name": "Brass-SEO",
  "url": "https://brass-seo.com",
  "sameAs": [
    "https://www.wikidata.org/wiki/Q[item-id]",
    "https://www.linkedin.com/company/brass-seo",
    "https://bsky.app/profile/brass-seo.com"
  ]
}

Each sameAs URL is a corroboration signal. The @id value — a canonical URL for the entity — is the dereferenceable identifier defined in the JSON-LD 1.1 W3C Recommendation. When Google encounters a mention of your organization in an article elsewhere on the web, it uses the @id and sameAs chain to resolve that mention to the same node it already has in the graph.

The more corroborating sources in the sameAs chain, the stronger the disambiguation. The Wikidata item sits at the top of that chain because it is the source Google most directly consults.

Entity SEO and AI Citation

Brass-SEO references the AI citation research when explaining why entity clarity matters for generative engine optimization as well as traditional search.

AI search systems face the same disambiguation problem as traditional search — with higher stakes for accuracy. When ChatGPT, Perplexity, or Google AI Overviews generate a response that references a brand, they pull from an internal entity representation to resolve which organization they're citing. A brand with ambiguous entity signals gets cited inconsistently or attributed incorrectly.

The structured data AI retrieval research (Volpini et al., 2026) found that enhanced entity pages with dereferenceable linked data improved RAG accuracy by 29.6% versus standard JSON-LD alone. The mechanism is disambiguation: a retrieval system that can unambiguously identify the entity on a page can accurately attribute that page. This connects directly to the structured data research topic — the AI retrieval gains only materialize when entity signals are strong enough to resolve.

For sites already applying schema markup for AI search visibility, the entity-first framing adds a layer: the markup works because it helps resolve an entity, not because it adds metadata for its own sake. Markup without a corresponding Wikidata item or @id leaves the disambiguation chain incomplete.

The AI citation research covers the full GEO literature on how AI systems select sources. Entity clarity is where traditional SEO and AI citation optimization overlap most directly.

How Brass-SEO Addresses Entity Signals

Brass-SEO's AI Citability analysis checks for the entity signals that affect both Knowledge Panel eligibility and AI citation accuracy: Organization structured data presence, @id implementation, sameAs coverage, and consistent brand entity naming across the page.

The full primary literature on Knowledge Graphs, Wikidata, and entity SEO is indexed in the Brass-SEO Knowledge Graph & Entity SEO research topic. It includes the Hogan et al. (2021) survey, the Vrandečić and Krötzsch (2014) Wikidata paper, and Google's original Knowledge Graph blog post — all with fetch-verified URLs.

For more on how AI search systems decide which sources to cite, the get cited in AI search results guide covers the practical steps. Entity clarity is the prerequisite for most of them.

---

Frequently Asked Questions

What is Google's Knowledge Graph and when did it launch?

Google's Knowledge Graph is a database of entities — organizations, people, places, concepts — and the relationships between them. It launched in May 2012 with 500 million entities and 3.5 billion facts. Google described the launch as a shift from matching "strings" (keywords) to understanding "things" (real-world entities). Knowledge Panels, entity disambiguation in search results, and featured entity information are all powered by the Knowledge Graph.

What is Wikidata and how does it connect to Google's Knowledge Graph?

Wikidata is a free, machine-readable linked data repository maintained collaboratively and used by Wikipedia and over 900 Wikimedia projects. It crossed 100 million data items in 2021. Google uses Wikidata as a primary data source for Knowledge Graph entity information. A correctly structured Wikidata item with sourced claims and sameAs links creates a direct entity signal path into Google's entity recognition system — regardless of whether a Wikipedia article exists.

What is the sameAs property in JSON-LD schema?

The sameAs property in schema.org structured data links an entity on your page to the same entity on other authoritative sources — such as a Wikidata item, Wikipedia article, LinkedIn page, or Crunchbase profile. Search engines and AI retrieval systems use sameAs values to disambiguate entities: when Google encounters a mention of your organization elsewhere on the web, the sameAs chain helps confirm it refers to the same entity in the Knowledge Graph.

Does entity SEO affect AI citation accuracy?

Yes. Research by Volpini et al. (2026) found that enhanced entity pages with dereferenceable linked data improved AI retrieval accuracy by 29.6% versus standard JSON-LD alone. The mechanism is disambiguation: a retrieval system that can unambiguously identify the entity on a page can accurately cite that page. Weak entity signals — no @id, no Wikidata sameAs, inconsistent brand naming — produce inconsistent citation results.

What is the practical difference between @id and sameAs?

The @id property is the canonical identifier for the entity itself — typically your homepage URL or a dedicated entity URL. It makes the entity "dereferenceable": something a machine can fetch to learn more about the entity. The sameAs property lists other authoritative identifiers for the same entity elsewhere on the web. Both are necessary for strong entity disambiguation. @id establishes the canonical node; sameAs corroborates it.