Cloud programmes tend to start with a destination rather than an assessment. The decision is made at a strategic level, and the engineering team is left to discover which workloads will not fit — usually during migration, when the options have narrowed.
A readiness assessment inverts that. It looks at what each workload actually needs before deciding where it should live.
Placement is a per-workload decision
Latency between an application and its database, licensing that is priced per physical core, data that must remain in a jurisdiction, and dependencies on hardware that has no cloud equivalent — each of these can determine placement on its own, and none of them appears on a strategy slide.
Assessed properly, most estates split rather than move wholesale. That is a legitimate outcome, not a failure of ambition.
Hybrid is only cheap if it is designed
A split estate costs more to run than either extreme if identity, networking, monitoring and change control are duplicated across both sides. It costs less than either if those layers are designed once and applied consistently.
The difference between the two outcomes is almost entirely architectural, and it is decided early.
Operating model, not just landing zone
Migration finishes; operation does not. Cost governance, capacity review, patching and access management all need an owner and a rhythm. Programmes that treat these as a phase-two concern generally discover the bill before they discover the process.
In short
Assess per workload, design the shared layers once, and agree who operates the result before you migrate.