Knowledge graph
In short
A knowledge graph is a data structure that represents real-world entities (companies, people, products, locations) as nodes and the relationships between them (employs, owns, competes with, located in) as edges, each with a defined meaning.
Because relationships are explicit, you can ask connected questions, such as "which subsidiaries of this parent use a data warehouse?", and get precise answers. AI systems use knowledge graphs to ground responses in structured facts.
Core building blocks
- Entities: The things you track, each with a unique ID.
- Relationships: Typed links between entities, such as "subsidiary of".
- Attributes: Facts on a node, such as employee count or industry.
- Ontology: The schema that defines allowed entity and relationship types.
Where GTM teams use it
- Account hierarchies: Model parents, subsidiaries, and brands for ABM and territories.
- Buying committees: Link contacts to roles, reporting lines, and past interactions.
- Competitive and partner maps: See who uses, resells, or integrates with whom.
- AI grounding: Give assistants structured facts to retrieve instead of guessing.
Example
Say a prospect is a subsidiary of a group that is already your customer. A knowledge graph exposes that link, so the rep asks the existing champion for an introduction instead of cold emailing. A flat list would have shown two unrelated accounts.
How to build one
- Choose the questions: Decide which connected questions it must answer.
- Define the schema: List entity and relationship types before loading data.
- Resolve entities: Match records so each real company appears once. See entity resolution.
- Load and source-tag: Record where each fact came from and when.
- Maintain: Review and expire outdated relationships.
Common mistakes
- Modeling everything: Start with a narrow schema tied to real questions.
- Duplicate nodes: Unresolved duplicates break traversal.
- No provenance: Facts without sources are hard to trust or correct.
Knowledge graph FAQs
What is a knowledge graph?
A knowledge graph is a structured map of entities, such as companies and people, and the typed relationships between them. It lets systems answer connected questions and gives AI tools a reliable set of facts to retrieve.
How is a knowledge graph used in B2B sales?
Sales teams use it to model account hierarchies, buying committees, partner and competitor relationships, and shared customers. Seeing those links helps with territory design, warm introductions, and prioritizing accounts connected to existing customers.
What is the difference between a knowledge graph and a database?
A relational database stores data in tables, and relationships are implied through keys. A knowledge graph makes relationships first-class, typed, and directly queryable. You can store a knowledge graph in several kinds of databases.
Do I need an ontology to build a knowledge graph?
A basic ontology helps. Defining entity and relationship types up front keeps data consistent and queries meaningful. It can start small, with a few types tied to your first use case, and grow over time.
