Modernization programs are usually measured by how much has moved: applications migrated, workloads re-platformed, percentage in the cloud. Those numbers are easy to count. They say almost nothing about whether the investment was worth it.
Modernization should be an investment decision, not a migration scorecard.
Decide what to modernize, and what to leave alone
Not every system deserves modernizing. Rank candidates on the same criteria, from business value and risk to what they enable for AI. Some will earn the cost. Some should be contained. A few should be retired outright.
Fund it in stages that each earn the next
Stage-gated funding changes the conversation with finance. Each phase has a defined outcome, agreed before it starts. The next tranche is released when the last one has shown what it changed. Programs that can't show a result early are stopped early, which is cheaper than late.
Measure the capability, not the migration
The questions that matter are about what the organization can now do. How quickly can a priority use case reach value? What does it cost to run a business transaction? How much faster are decisions made? And has the legacy estate actually been switched off, or is it still running beside the new one? The scoreboard that cannot be faked is short: systems retired, consumers repointed, extracts switched off. Budget for dual-run too: until legacy is off you pay for both, so name who owns that cost.
How this goes wrong
- Modernize everything. A program sized to the whole estate, with no ranking and no end date.
- Lift and shift, then stop. The same design on new infrastructure, at higher run cost.
- Keep the old estate just in case. Two platforms, double the cost, and decisions still made from the old one.
- Report migration percentage. A number that rises steadily while business value stays flat.
- Forget the operating model. New platform, old ownership, same friction.