Cloud migrations rarely fail because of the cloud. They fail because of what nobody knew about the existing systems: a forgotten dependency, a nightly job that writes to a shared folder, a licence tied to a physical server, or a bill nobody planned for.
This checklist walks through the phases that prevent those surprises.
1. Know why you are migrating
Write down the goals, because they decide the trade-offs later. Common ones: closing a data centre, scaling with demand, improving reliability and recovery, releasing faster, or reducing hardware and maintenance burden. Agree how you will measure success, such as cost per month, availability, recovery time or release frequency.
2. Assess what you have
Build an inventory of every application, database, server and integration, with:
- owner and business importance;
- dependencies (which systems talk to which, including file shares and scheduled jobs);
- data volumes and growth;
- performance, availability and recovery requirements;
- licensing constraints;
- security, compliance and data-residency requirements for your sector and country.
Dependency mapping is the step most often rushed and the one that causes most cutover failures.
3. Choose a strategy per application
| Strategy | What it means | When it fits |
|---|---|---|
| Rehost | Move as-is to cloud servers | Deadlines, low-risk first wave |
| Replatform | Small changes, e.g. a managed database | Quick wins in reliability and admin effort |
| Refactor | Redesign for cloud-native services | Core systems that need to scale or change often |
| Repurchase | Replace with a SaaS product | Commodity functions such as email or HR |
| Retire | Switch it off | Duplicate or unused systems |
| Retain | Leave it for now | Not worth moving yet, or blocked by constraints |
Most estates use several strategies. Retiring unused systems is the cheapest migration of all.
4. Build the landing zone first
Before moving anything, set up the foundation:
- account or subscription structure for production, staging and development;
- networking, with private connectivity to remaining on-premises systems where needed;
- identity and access management with single sign-on, least-privilege roles and multi-factor authentication;
- a security baseline: encryption, centralised logging, threat detection and guardrails that block risky configurations;
- backup and disaster recovery policies;
- tagging standards, so every resource has an owner, environment and cost centre.
5. Plan the data migration
- Classify data and confirm where it is allowed to be stored and processed.
- Choose the method per dataset: one-off transfer for static data, continuous replication for live databases.
- Clean up before moving. Archiving obsolete data saves storage and time.
- Validate after moving: row counts, checksums and application-level checks that the data is complete and correct.
6. Migrate in waves
Start with a low-risk application to prove the process, then group the rest into waves by dependency and business priority. For each wave:
- Replicate and test in the new environment.
- Rehearse the cutover end to end.
- Agree the go/no-go criteria and the rollback plan.
- Cut over in a low-traffic window, with the owners on hand.
- Watch closely for several days before decommissioning the old system.
7. Control cost from day one
- Right-size resources based on real usage, not the old server sizes.
- Turn off or scale down non-production environments outside working hours.
- Use managed services where they reduce operational effort.
- Use commitment discounts for steady workloads once usage is understood.
- Set budgets and alerts, and review spending by owner every month.
8. Run and improve
Migration is the start, not the end. Put monitoring and alerting in place for availability, performance, errors and cost; automate deployments with CI/CD and infrastructure as code; test backups and recovery regularly; and keep improving the systems you rehosted, since that is where much of the long-term value sits.
The short version
Know why you are moving, map every dependency, choose a strategy per application, build a secure landing zone, migrate in rehearsed waves, and give every cloud resource an owner and a budget. Do those well and the migration itself becomes routine.