Treating data as something delivered.
WHAT A DATA PRODUCT IS
A dataset or service with an owner, defined quality, documentation and consumers.
WHAT DISTINGUISHES IT FROM A TABLE
Somebody is accountable for it, and consumers can rely on stated guarantees.
WHAT IT SHOULD DECLARE
What it contains Its grain Its freshness Its quality guarantees How changes are communicated Who owns it
WHY THAT FRAMING HELPS
It moves data from a by-product of pipelines to something delivered deliberately.
WHAT DATA MESH PROPOSES
Domain teams owning their own data products, with a shared platform.
WHAT IT SUITS
Large organisations where a central team becomes a bottleneck.
WHAT IT REQUIRES
Genuine capability in each domain team A platform making it practical Federated governance
WHAT IT DOES NOT SUIT
Small organisations, where it produces inconsistency rather than autonomy.
WHAT TO ADOPT FROM IT REGARDLESS
Ownership, documented interfaces, and stated guarantees.
WHAT TO MEASURE FOR ANY DATA PRODUCT
Whether consumers use it, and whether they trust it.
WHAT TO DEPRECATE PROPERLY
Anything being replaced, with notice and a migration path.
WHY
Silently removing a table breaks someone's report without warning.
