A company must modernise a portfolio of applications. How should each be assigned a modernisation approach, and what is the most common misassignment?

The approaches and when each fits

• Retire — nobody depends on it, or the dependency can be removed. The highest-return option and the one most often skipped, because switching something off requires a conversation.
• Retain — leave it alone. Appropriate when the system works, is cheap to run, and changes rarely. Saying so explicitly protects capacity.
• Rehost — move it without changing it. Fast, low risk, delivers infrastructure benefits only. Correct when the driver is a datacentre exit or a hardware end-of-life and the application itself is fine.
• Replatform — move it with targeted changes, such as swapping a self-managed database for a managed one. A good default when the driver is operational cost.
• Refactor / rearchitect — restructure while preserving behaviour. Expensive, and justified when the business needs capabilities the current structure cannot support.
• Repurchase — replace with a commercial product. Correct when the capability is commodity and the organisation's version has no advantage.

The most common misassignment

Choosing refactor when replatform or retire would do, because refactoring is the most architecturally interesting option and because "we will fix it properly" is easier to propose than "we will leave it alone" or "we will switch it off".

The second most common: rehosting everything on the theory that it can be improved later. It usually is not, and the estate has moved without improving while consuming the modernisation budget.

What should drive the assignment

The business driver, not the technical condition. A datacentre exit with a deadline argues for rehosting almost everything. A cost problem argues for replatform and retire. A need for capabilities the current architecture cannot provide argues for refactoring the specific components that block them — and only those.

The discipline that keeps it honest

Assign the cheapest option that satisfies the driver, and require a written justification for anything beyond it. Most portfolios then resolve into a large number of retire and retain decisions, a moderate number of replatforms, and a small number of genuine rearchitectures — which is a far better allocation than treating every system as a modernisation project.