Cloud Migration: A Checklist Before You Move Anything

Cloud migration projects go wrong less often because of the technology and more often because of decisions that were never made explicitly before the migration began. A checklist worked through in advance catches most of the common, expensive mistakes.

Model the real cost before committing, not just the headline compute pricing. Cloud costs include compute, storage, data transfer (which is often underestimated and can become a significant line item), and the operational cost of the expertise needed to manage cloud infrastructure well. A migration justified purely on "the cloud is cheaper" without a realistic cost model frequently produces an unpleasant surprise on the first few months of invoices.

Decide the migration strategy deliberately: lift-and-shift (moving systems largely as-is) is fastest and lowest-risk but captures the least benefit, since it does not take advantage of cloud-native efficiencies. Re-platforming (moving with moderate optimisation) balances speed and benefit reasonably well for most businesses. Full re-architecting captures the most benefit but takes the longest and carries the most risk, and is usually only justified for systems that will run at significant scale.

Security and access control need to be redesigned for the cloud model, not simply copied from on-premises assumptions. Cloud environments have different default security postures, and misconfigured cloud storage or access permissions are a leading cause of data exposure incidents — a proper migration plan includes a specific security review, not an assumption that existing security practices transfer automatically.

Sequence migration by risk and dependency, not by convenience. Non-critical, standalone systems make good candidates to migrate first, building organizational experience and confidence before tackling business-critical, highly interdependent systems where a mistake has real operational consequences.

Plan for a period of parallel running or a clear rollback plan for critical systems. Knowing exactly how to revert if the migrated system does not perform as expected, before the migration happens rather than improvised during a crisis, meaningfully reduces the real-world risk of the entire project.

Train the team on operating in the new environment before full cutover, not after. Cloud infrastructure often requires different operational skills and monitoring approaches than on-premises systems, and a technically successful migration can still create real operational problems if the team managing it day to day was not properly prepared for how the new environment behaves.