A replacement programme is often presented as a choice between an expensive old platform and a more capable new one. That comparison can be useful when both alternatives are available. During a real transition, however, the enterprise may have neither the option to stop the old service immediately nor the ability to use the proposed successor. It must fund and govern the interval. If the business case leaves that interval vague, the apparent economy of replacement can depend on work that has simply been omitted.
The roadmap is not an available service
In insurance, an older product engine may continue to support obligations after new sales have moved elsewhere. The technology can be unattractive to maintain and still perform work that the organisation needs. Calling it legacy does not decide when that work can cease. Equally, describing a successor as strategic does not establish that it can yet receive the existing responsibility. The engineering portfolio needs an account of capability and acceptance, not only a preference for the future architecture.
I faced that problem in a June 2025 review. Our budget assumed that an older engine would soon stop needing specialist attention, while its successor still had acceptance work ahead. The replacement sponsor wanted the remaining specialists moved onto new development. Maintenance colleagues, worried about losing the ability to support the existing service, preferred to preserve their arrangement without a defined end. Each position reflected a real concern. Neither gave the enterprise a complete transition choice.
A bridge carries an opportunity cost
I recommended an eighteen-month bridge costing JPY240 million. It retained the specialist capacity and environment needed to support the existing engine, with named ownership, a defined remit and reviews tied to the successor's acceptance evidence. To provide the engineering capacity, we deferred two discretionary analytics features. That deferral belonged in the decision. Presenting the bridge as absorbed work would have hidden the fact that the same people could not deliver every competing commitment at once.
The proposal did not promise that the replacement would finish within eighteen months. It established a funded period and the evidence needed for subsequent decisions. At the March 2026 review, the older service remained supported and two of three planned acceptance stages for the successor were complete. The third was still open. That was the position I reported. I did not claim retirement savings from an engine that had not retired, or treat approval of a bridge as proof of the successor's eventual readiness.
There is an understandable objection to explicit bridge funding: it can make delay easier to tolerate. A programme might continue extending the old service while the replacement loses urgency. That is a reason to govern the bridge, not a reason to pretend the continuing service costs nothing. The remit should distinguish maintenance needed for responsible operation from discretionary enhancement. Review points should examine both the current service obligation and the successor's progress. Further funding should require a current decision, rather than follow automatically from the existence of the first bridge.
Exit evidence matters more than an announced date
The opposite failure is to use an announced retirement date as a substitute for acceptance evidence. A date can organise work and expose a deteriorating schedule. It cannot make an incomplete successor capable. If removing support on that date would leave a required service without a responsible owner, the portfolio has to confront the choice. It may accelerate acceptance work, revise scope, extend the bridge or select another route. Hiding the choice inside a maintenance reduction makes it harder for the accountable leaders to examine.
I find it helpful to separate three propositions in the investment account. First, what makes the proposed successor worth pursuing? Second, what existing work must continue during transition? Third, what evidence would justify transferring or ending that work? These propositions are connected but not interchangeable. A strong argument for the successor does not answer the second and third questions. Nor does the continuing cost of the predecessor establish that every replacement proposal is sound.
People create another constraint that a purely financial comparison can miss. The specialists maintaining the older engine may also be the people needed to explain its behaviour to the replacement team. Moving all of them into development can weaken current support. Leaving them entirely in maintenance can deprive the new programme of essential knowledge. The allocation should identify those competing needs and the consequences of each choice. A label such as shared resource does not create additional hours or resolve the priority when both services need attention.
The bridge should also preserve an intelligible exit record. When the successor takes responsibility, the receiving team needs to know what has moved, what remains elsewhere and which acceptance evidence supports the transfer. A portfolio report that celebrates closure without identifying the residual obligations can leave small but consequential tasks ownerless. I do not suggest retaining every old arrangement indefinitely. I want the final decision to be supported by a clear account of the work that has actually changed hands or ceased to be required.
Finance colleagues can improve the discussion by keeping approval, expenditure and realised benefit separate. Approval of JPY240 million was not a claim that the entire amount had already been spent at the first review. Deferring two features was an allocation choice, not necessarily a permanent saving. Completing an acceptance stage was progress toward readiness, not the retirement benefit itself. These distinctions make the portfolio less visually simple, but they prevent unlike measures from being added together as though they described one result.
For directors, the useful challenge is whether management has priced a credible transition rather than merely an attractive destination. What happens to the existing obligation if the successor takes longer? Which resources are needed by both? What is being deferred to maintain them? Who can approve another period, and on what evidence? The board does not need to choose the engineering architecture to ask whether the proposed timing depends on an unsupported assumption about service continuity.
I would include the bridge alternative in the original replacement case, before the schedule hardens into a public internal commitment. That does not mean budgeting for every conceivable delay. It means examining a plausible transition route and its costs while leaders still have choices about scope and sequence. A replacement programme is more credible when it explains the years in between. The enterprise is responsible for those years too, even when the most persuasive slide in the proposal describes what comes after them.