What happens when definitions live only in reports?

If metric logic is hidden in a dashboard or personal SQL, other teams cannot easily discover, reuse, or audit it. A metrics store catalogs what metrics exist, how they are calculated, who maintains them, and where they can be used.

What should a metrics store record?

Entries may contain a business definition, formula, source data, allowed dimensions, time window, refresh cadence, owner, certification status, and change history. Some systems compile definitions into queries; others serve as a metadata catalog while another service executes them. Some cache results. “Store” does not necessarily mean every metric value is stored.

For net revenue, a definition should specify refunds, taxes, event time, and allowed breakdowns. Analysts and agents can use the same version. When finance changes the policy, the store should record effective dates and prompt affected reports to be checked again.

How does it differ from a metrics layer or warehouse?

A metrics layer is the architectural capability for consistent metrics; a metrics store is commonly the component that holds their definitions. A warehouse stores underlying data, while a glossary explains business terms. Vendor boundaries vary. Lineage and quality information still matter for trustworthy results. Defining business semantics in the data engine illustrates how metrics and dimensions can be organized.

FAQ

Is a metrics store a table of KPI values?

Not necessarily; its central purpose is managing definitions.

Can it include uncertified metrics?

Yes, if status and ownership are clear.

How are duplicate metrics prevented?

Use naming, stewardship, search, and review of overlapping definitions.