Who decides what this column means?

dbt's own 2026 State of Analytics Engineering put ambiguous data ownership among the top pains teams report, alongside trust and quality — and it is the one with the least tooling. dbt groups and model owners tell you who maintains the SQL. They do not tell you who decides what customer_status is allowed to contain, who to ask before adding a value, or who should have been consulted when someone changed it last quarter. So the answer is usually 'ask in #analytics and see who replies'. This workspace makes ownership a property of the meaning rather than of the file: every column and every vocabulary can carry an owner, a steward team and a review cadence, sitting as columns in a grid you can sort and filter. The most useful view is the one that shows what has no owner at all — that list is your actual to-do, and it is the prerequisite for every review, notification and escalation workflow you might build later.

What you do

1. Every column and vocabulary gains an owner, a steward team, a review cadence and a decision forum.
2. Filter to 'no owner'. That list is the real work.
3. Start with the vocabularies — that is where the arguments live.