Replatform
also called Lift and Reshape
Migrating an application largely as-is while making targeted changes to exploit the target platform — usually the best return per unit of effort.
Sits between rehosting, which changes nothing and captures nothing, and refactoring, which changes everything at high cost and high risk.
The changes that typically pay for themselves: moving a self-managed database to the managed equivalent (removing patching, backup and failover work), containerising so deployment becomes repeatable, externalising configuration and secrets, replacing a local file dependency with object storage, and putting real logging and metrics in.
None of these require understanding or restructuring the business logic, which is the expensive and risky part. That is exactly why the ratio is good — the operational benefits arrive without the application-comprehension cost.
The judgement call is knowing when to stop. Each additional change increases risk and testing scope; the discipline is to take the operational wins now and leave restructuring to a separate, separately-justified decision.