What everybody does and what nobody does
In IT functions that grow without structure, two symptoms appear at once. There is work done two or three times (two teams documenting the same network, three scripts for the same thing) and work nobody does (who reviews access every quarter?, who maintains the tagging standard?).
Both have the same cause: nobody has written down who is responsible for what.
A map on three levels
Domains. Large areas of responsibility that are stable over time: platform, security, networks, data, cost, architecture. They do not depend on individuals or on the org chart of the moment.
Roles. Within each domain, the functions needed: who defines the standard, who enforces it, who reviews it. One person can hold several roles; a role can be shared between several people, but always with a named holder.
Deliverables. What each role produces or maintains, concretely: the tagging standard, the quarterly access review, the account inventory, the monthly cost report. Each deliverable has one single owner.
Written, versioned and validated
The map is not an org chart on a wall. It is a living document:
- It is written in a structured format, so it can be queried and validated (are there deliverables with no owner?, any role with no holder?).
- It is versioned like code: every change has a date and a reason.
- It is validated before automating: the governance tools (alerts, approvals, reports) are configured from this map, not the other way round.
When you notice it working
When an alert arrives, you know who it goes to. When something needs approving, you know who does it. When somebody leaves, you know exactly which deliverables have to be reassigned. And when a new task appears, the first question stops being 'who takes care of this?' and becomes 'which domain does it fall in?'.
Knowing who is responsible for each thing does not yet say who can decide what. That is what decision levels are for.
How we apply it: Multicloud Governance.