The three decisions you make before moving to Azure
Most cloud projects do not fail on technology. They fail because building started before three things were decided.
Moving to Azure is easy. Anyone with a credit card can create a virtual machine in ten minutes. That is why organisations often arrive and then, two years later, find themselves holding a bill nobody can explain and an architecture nobody can change.
The problem is never Azure. It is that building started before three things were decided.
1. Identity, before anything else
Identity is not a configuration detail. It is the foundation. Everything that comes afterwards (permissions, conditional access, automation, cost per team) sits on top of how identities were organised in the first month.
Postponing that decision does not make it go away. It gets made by default, by whoever creates the next resource. And undoing a badly built identity structure two years later, with production systems sitting on top of it, is the most thankless task there is.
2. How costs will be read
"What is this costing us?" is the question finance will ask, and they will ask it at the end of the first quarter.
With no tags, no meaningful resource groups and no naming convention, the answer is a long spreadsheet in which nobody can tell production apart from the test somebody left running in November.
Decide this before the first resource is created, because going back and tagging hundreds of resources afterwards is a project in its own right, and it is a project nobody approves.
3. What is not going to the cloud
This is the decision that is rarely made explicitly, and it is the one that saves the most money.
Not everything should migrate. Some systems are at end of life and should be switched off, not moved. Some workloads have a usage profile that makes the cloud more expensive than the server that has already been paid for. Some data stays where it is, for regulatory reasons.
A migration that moves everything indiscriminately is not a migration. It is a house move where you pack the rubbish and take it with you.
The pattern that repeats
These three decisions have one thing in common: they are all cheaper to make than to correct. They cost weeks at the start and they cost years at the end.
We learned that in regulated environments, in banking and in the public sector, where a deferred decision surfaces quickly. But this is not about the sector: a deferred decision always surfaces in the end.
Method at the start. Not guesswork at the end.
