How to build a business glossary alongside a data model
Record agreed business definitions with the model they describe, and keep the technical dictionary alongside them.
When this is useful
- Two teams report different numbers for the same metric and the difference turns out to be definitional.
- A data model is being built and the column names are outrunning the agreement about what they mean.
- A governance programme needs definitions that are actually connected to data.
Before you start
- A model to attach definitions to, and the people who can actually settle a definition.
Step 1: Record the business term
The Business Glossary holds the agreed meaning in business language. Write the definition someone outside the data team would accept.
Data ▸ Business Glossary
Step 2: Record the technical definition
The Data Dictionary carries the technical detail for the model itself, and the Metadata Repository is where the wider metadata sits.
Data ▸ Data DictionaryData ▸ Metadata Repository
Step 3: Connect the two
A glossary term with no data behind it is a wiki page. Attaching it to the entities and attributes it describes is what makes the definition enforceable and findable later.
Data ▸ Subject Areas
What happens next
The definitions become part of every lineage trace and impact assessment that crosses them, which is where a definitional mismatch usually reveals itself.
Example
An "active customer" definition agreed once and attached to the four entities that implemented it differently — the reconciliation work that followed had a clear target rather than a debate.
Tips
- Define the disputed terms first. The uncontroversial ones can wait forever without cost.
- Record who agreed a definition. A definition with no owner does not survive its first challenge.
Limitations
- The glossary and dictionary are separately enabled.