What is Entity?
Nati Elimelech·Updated 29.07.2026 An entity is a brand, person or concept search engines and AI models treat as one thing, with a name, attributes and links to other entities.
When a search engine or an AI model meets a name, it does not see a text string. Google calls this things, not strings: its Knowledge Graph is what Google describes as a database of billions of facts about people, places and things, and every entry in it carries a name, attributes and relations to other entities.
How entities work
The problem entities solve is ambiguity. Google’s own example, from the 2012 Knowledge Graph announcement: “Taj Mahal” can be the monument, the musician, a casino or a restaurant, and a system that understands entities knows which one the query means. Google’s Knowledge Graph is the best-documented implementation of the idea, but the idea is not Google’s: any system that answers questions needs a way to decide which “Taj Mahal” it is being asked about.
There is no single central registry of entities: each system keeps a knowledge graph of its own. Public databases like Wikipedia and Wikidata serve as shared identity anchors that machines read too, and they are a strong signal, but they are one source among several that such graphs are built from.
How entities relate to SEO and GEO
A well-identified entity is the foundation: without it, the system is not sure who you are, and everything written about you scatters across name variants and contradictions. In query fan-out, some of the sub-queries touch the field’s entities directly, and an entity the system is not sure about simply does not come up in them.
The difference between an entity and a keyword
A keyword is what people type, an entity is what the system understands. You compete for a keyword in every search anew; an entity is built over time and serves every query that touches it. Keyword SEO asks “what do I want to show up for”, entity work asks “who am I in the system’s eyes”.
Questions about Entity
How do you build a strong entity?▼
A strong entity is built on consistency: same name, same facts and same context on your site, in schema, in profiles and in external mentions. Google documents that Knowledge Graph facts come from a variety of sources, including information received directly from content owners, so every source that contradicts the others makes the identification harder. That part is in your control.
The second part less so: Google itself says a Knowledge Panel, for example, is created only when there is enough information on the open web. Real entity work starts with making those sources exist. The markup comes after them.
Does schema markup help entity recognition?▼
Schema markup helps recognition, but in a more modest role than it usually gets credit for: per schema.org, the sameAs field points at a page that unambiguously identifies the entity, like a Wikipedia or Wikidata entry. Google documents Organization markup as helping it disambiguate your organization from others in its search results, and defines sameAs more narrowly, as links to profile pages.
As for AI: Google writes that no special schema is needed to appear in AI Overviews and AI Mode. Google does not speak for other engines, and Microsoft and OpenAI have published no equivalent statement about schema and AI answers.
Do you need a Wikipedia article to be an entity?▼
An entity does not require a Wikipedia article: it is a record in a knowledge graph, and it exists without any encyclopedia entry. A Wikipedia article or a Wikidata item is a strong identity signal, and schema.org names exactly those two as sameAs examples, but they are reinforcement, not an entry condition.
Is an entity the same as a Knowledge Panel?▼
An entity is not a Knowledge Panel. The panel is one display of an entity Google is confident enough about, and Google writes that it is created automatically, and that when and how it appears is not something Google can or would influence. The entity exists without a panel: it is the record in the knowledge graph, not the box in the results.