Knowledge graph
A map of real-world things and how they relate, built by an engine so it can reason about facts and connections rather than matching words on a page.
When an AI engine seems to "know" things about a business, it is not reading your homepage in the moment. It is consulting a map it built earlier. That map is a knowledge graph.
What a knowledge graph is
A knowledge graph is a structured map of real-world things and how they relate. It holds entities, companies, people, products, places, and the connections between them: this person founded this company, this company sells this product, this product serves this market.
The reader is a machine, and the graph is how it reasons. Instead of matching words, it follows connections. Ask it about a topic, and it can pull in everything it has linked to that topic and answer from the relationships, not just the page in front of it.
Why engines build one
A graph lets an engine answer questions that no single page states outright. If it knows your company serves a market, and it knows what that market needs, it can connect the two without you spelling it out.
It also lets the engine resolve ambiguity. Two businesses with similar names become two separate nodes, each with its own facts. The clearer and more consistent your identity, the cleaner the node it can build for you.
The old way and the new way
The old way ranked pages: a list of documents that matched a query, ordered by relevance.
The new way reasons over a graph: a model of things and relationships that an engine draws on to compose an answer. Your pages still feed it, but the unit of understanding shifted from the document to the entity and its connections.
The damaging admission
You do not control the knowledge graph. The engine builds and curates it from many sources, and no tool can guarantee it adds you, or that the facts it holds about you are right. Anyone promising to put you in a knowledge graph is overselling.
What you can control is whether your own site states a clear, consistent identity that is easy to read. If your pages are unreadable to a machine or contradict each other, you make it harder for an engine to build an accurate node for you, and that is the part worth fixing first.
How to check where you stand
The practical starting point is whether engines can read your site cleanly enough to attach facts to you at all. That is measurable.
Run a free scan on any URL to see how clearly engines read your pages, or read how AI engines decide what to recommend for the wider context.