A glossary entry usually holds the term, a plain-language definition, the team that owns it, how it is calculated where that applies, and the data fields that carry it. Active customer, for example, might mean a customer with a paid order in the last 90 days, owned by finance and stored in two specific tables. Vocabulary standards such as W3C SKOS describe how to publish such terms so other software can read them.
Most of the value comes from the agreement behind each entry. Sales may count a customer at signature and finance at first payment, and the glossary records which of the two a given metric uses.
What a business glossary entry usually holds:
- one definition written for the people who use the term
- an owner who approves changes
- calculation rules and known exceptions
- links to the tables and fields that carry the term
How is a business glossary different from a data dictionary?
- A data dictionary describes technical fields: name, type, table and allowed values.
- A business glossary describes business meaning, and one glossary term can map to several fields.
- An ontology goes further and defines how the terms relate to one another.
Where does a business glossary break in AI work?
A definition that lives only in the glossary does not reach the data. An AI assistant asked for active customers may read a table where closed accounts were never filtered out and report a number that matches no agreed definition. The glossary helps an AI task only when someone checks its definitions against the records that task uses, including the exceptions, with business rule validation, and checks again when a definition or the data changes.