Name the Owner and Source of Each Metric

A metric with no owner is a metric nobody can change, which sounds harmless until the first time it needs changing.

The owner is whoever is accountable for the definition

Not who builds the report. Not who reads it most. The person whose judgment the definition encodes, and who has the authority to change it. That constraint is the same one the decision guide applied and it fails the same way: an owner who cannot change the definition has been assigned blame.

One source of truth per metric

The rule is short: a metric computed in two places will disagree in one of them, and you will discover this in a meeting.

One contract owns the period definitions, the category derivation, and the amount semantics. Everything else reads from it. The reporting interface must not become a second calculation engine, which it does gradually and without anybody deciding to, one convenient local tweak at a time.

Definitions travel with the number

A category is not a label. It is a definition plus the rule that assigns records to it, and both belong where the number is read rather than in a document somebody would have to go and find.

This sounds like documentation hygiene and it is actually about trust: a reader who cannot see the definition has to either accept the number or reject it, and most people accept it, which is the worse of the two failure modes.

Separate configuration from consumption

The people who define periods, categories, and visibility are not the people reading the output. Give the definitions their own ownership, their own change path, and validation before a change publishes. A definition change that silently alters every historical number is one of the more alarming things that can happen to a reporting system, and it happens routinely.