What is a knowledge graph?

Wikipedia defines it as "a knowledge base that uses a graph-structured data model or topology to represent and operate on data", often used "to store interlinked descriptions of entities" (Wikipedia, read 1 November 2026). The same article says: "There is no single commonly accepted definition of a knowledge graph."

The term has two audiences. Data engineers use it for graph databases inside a company. Marketers usually mean Google's Knowledge Graph, the store of facts behind Google's knowledge panels. This entry covers both, with the focus on search.

How does a knowledge graph work?

The basic unit is a fact made of three parts: a thing, a relation and a value. "Acme Help Desk", "founded in", "2021". Each thing is an entity with its own ID. Facts connect entities, so a system can answer a question by following links rather than matching words.

Google introduced its Knowledge Graph in May 2012 with the line "things, not strings". At launch it held "more than 500 million objects" and "more than 3.5 billion facts", built from sources including Freebase, Wikipedia and the CIA World Factbook (Google blog, 16 May 2012, read 1 November 2026).

Google's knowledge panel help adds that facts come from public sources, licensed data and content owners who suggest changes (Google Help, read 1 November 2026).

Developers can look up entities through the Knowledge Graph Search API. Google says it returns individual entities in JSON-LD using schema.org types, and warns it "is not suitable for use as a production-critical service" (last updated 2024-04-26).

Why does the knowledge graph matter for B2B SaaS?

If Google knows your company as an entity, it can connect your name to your category, founder and products. Without that, your brand is only a string of letters that may match other things.

AI answers work from the pages a system finds. Clear, consistent facts about your company across your site and third-party pages give any system less to guess. Google says Organization markup helps it "disambiguate your organization in search results" (Google Search Central, last updated 2026-09-08). None of the Google pages we read describe a way to submit a company to Google's Knowledge Graph directly.

Example (hypothetical)

A hypothetical knowledge graph holds three linked entities: Acme Help Desk (a software company), Jane Doe (a person) and help desk software (a category). Facts link them: Jane Doe founded Acme Help Desk; Acme Help Desk makes help desk software. A question like "who founded Acme Help Desk" is answered by following one link.

  • A knowledge panel is a box in Google results that shows facts from the Knowledge Graph.
  • Wikidata is an open, editable knowledge base built on the same idea of items and statements.
  • Entity SEO is the marketing work of making your company a clear entity.
  • sameAs is markup that links your site to your other profiles, a hint about which pages describe the same entity.

For how entities fit into getting named in AI answers, see the LLM SEO guide.