Published on

Organisation first, automation second

Automating a disorganised organisation does not organise it: it multiplies its disorder at machine speed. Before governance tools you need three much simpler things.

Service line Multicloud Governance

The tool does not decide who the owner is

Many cloud governance initiatives start by buying or configuring tools: automatic policies, compliance dashboards, alerts. Before long the same unanswered questions appear: who approves this exception?, who owns this account?, is this rule mandatory or is it a recommendation?

The tool works perfectly. What does not exist is the organisation that should be telling it what to do.

Who first, the tool secondAutomating without organising leaves alerts, rules and exceptions with no owner; organising first turns them into responsibilities.GOVERNANCE · ORGANISATIONWho first, the tool secondAutomating without organisingToolRulesAlertsAlerts that go to nobodyRules nobody knows are mandatoryExceptions with no controlOrganising and then automatingOwnersStandardDecisionsToolEvery alert has an ownerEvery rule has a level and a checkEvery exception has a dateN100F · n100f.com

The three things that come first

Owners. Every domain, every account and every deliverable has a responsible person, by name. Without that, every alert belongs to everybody, which is to say nobody.

A written standard. What is permitted, what is mandatory and what is merely advisable. Without a standard, every review is an opinion.

Decision levels. What is decided at each level, who decides it and how anything that does not fit gets escalated. Without that, everything goes up to the board or nothing leaves the teams.

With those three things settled, automating is almost mechanical: the tools enforce what the organisation has already decided.

Governance is not control

The word 'governance' sounds like a restriction. Done well, it is the opposite: teams know what they can do without asking permission, who to ask when something is unclear and how to request an exception when they genuinely need one. Fewer meetings, fewer blockages, fewer surprises.

The route

This series starts with the organisation (domains, roles and decision levels), moves on to the standard (how to write it, how to make it evolve and how to manage its exceptions), goes down to governing several clouds at once (account structure, identity and inherited policies) and ends with how governance itself is measured and versioned.


How we apply it: Multicloud Governance.

If your cloud cost keeps growing and nobody knows why, or your architecture depends on whoever built it, let’s talk.

Let’s talk