Letting others answer their own questions.
WHAT IT REQUIRES
Data modelled so it is understandable without explanation Documentation people can find Consistent definitions Training A route to ask when stuck
WHY IT FREQUENTLY FAILS
Raw or poorly modelled data is exposed, and users produce wrong numbers confidently.
WHAT TO EXPOSE
Modelled tables with clear grain and naming.
WHAT NOT TO EXPOSE
Raw layers Intermediate models Tables whose meaning requires context
WHAT TO PROVIDE
A catalogue with descriptions and owners Examples of common questions and how to answer them Certified datasets, marked as trusted
WHAT CERTIFICATION MEANS
A dataset reviewed, tested and owned, distinguished from exploratory ones.
WHY THAT DISTINCTION MATTERS
It tells users which numbers they may rely on.
WHAT TO MONITOR
Which tables are actually queried Which queries fail What people ask for repeatedly
WHY THAT LAST POINT
Repeated requests indicate a model that should exist.
WHAT TO ACCEPT
That self-service reduces requests but does not eliminate them.
WHAT TO PREVENT
Users building their own parallel definitions.
