You need funding for eighteen months of platform work with no customer-visible features. How do you build the case?

Do not lead with the technology

"We need to modernise the platform" is a request for trust. Lead with a number the business already recognises as a problem.

Find the cost that already exists

Three usually available and rarely added up:

Delivery drag. Lead time now versus what it was, and the cost of the difference in delayed revenue or in engineering time spent on coordination rather than features.

Run cost. Engineering time spent operating and patching, expressed as a proportion of the team. "Forty percent of four teams" is a headcount number a CFO can act on.

Risk realised. Incidents in the last year attributable to the current design, with their business cost.

Then state the outcome, not the output

Not "migrate to Kubernetes". "Reduce environment provisioning from six weeks to one day, and cut the operating share from forty percent to fifteen." Those are testable claims with a date.

Stage it so it can be stopped

Eighteen months unbroken is a large ask. Three phases with a value checkpoint each is a materially easier decision, because the sponsor can stop after phase one. That also forces the plan to deliver something in phase one, which is a discipline worth having anyway.

Include the decommissioning

A programme that only adds does not land. The saving comes from what is retired, and naming it — with dates and licence figures — is what makes the business case close.