Moving to the cloud promises flexibility, resilience and lower operating effort. Without a plan it can also bring surprise bills, downtime and security gaps. This checklist walks through the phases that matter.
Phase 1: Assess what you have
- Inventory every application and server — what it does, who owns it, how critical it is.
- Map dependencies — databases, file shares, scheduled jobs, external APIs. Hidden dependencies are the most common cause of failed cut-overs.
- Record current performance and cost so you can compare after the move.
- Note compliance needs such as data residency or audit requirements.
Phase 2: Choose a strategy per application (the 6 Rs)
| Strategy | What it means | When to use it |
|---|---|---|
| Rehost | Lift and shift as-is | Quick moves, tight deadlines |
| Replatform | Small changes, e.g. managed database | Easy wins on maintenance |
| Refactor | Re-architect for cloud-native services | Core apps that need to scale |
| Repurchase | Replace with SaaS | Commodity tools like email or HR |
| Retain | Keep on-premise for now | Low value or blocked by constraints |
| Retire | Switch off | Unused or duplicate systems |
Phase 3: Build the landing zone first
Before moving anything, set up:
- Separate accounts or subscriptions for production and non-production
- Identity and access management with single sign-on and least privilege
- Network design, VPN or private links to your offices
- Logging, monitoring and alerting
- Backup policies and a tested restore procedure
- Budgets and cost alerts
Phase 4: Migrate in waves
Start with a low-risk application to prove the process. For each wave:
- Replicate data and test the application in the cloud.
- Agree a cut-over window with the business.
- Prepare a rollback plan with a clear decision point.
- Cut over, verify with real users, then decommission the old system only after a stable period.
Phase 5: Optimise after go-live
This is the step most teams skip — and where most savings are found.
- Right-size instances using actual usage data.
- Schedule non-production environments to switch off out of hours.
- Use reserved or savings plans for steady workloads.
- Clean up unattached disks, old snapshots and idle load balancers.
- Review costs monthly with the application owners.
Key takeaways
Assess dependencies thoroughly, choose a strategy per application, build security and monitoring before the first move, migrate in small waves with rollback plans, and keep optimising after go-live.
