Capital decisions are normally governed by a case: here is the spend, here is the return, here is what we expect to see and by when. The current wave of technology commitment has largely escaped that discipline, because the cost of being seen to do nothing has felt higher than the cost of spending badly.
That asymmetry is the interesting part, and it is temporary. It ends the first time a board is asked to explain a run rate rather than a budget.
Two different decisions wearing one name There is a capability decision and there is a capacity decision. Boards routinely approve the second while discussing the first.
A capability decision buys the ability to do something the business could not previously do. It is small, reversible, and its real output is information. A capacity decision commits the business to running something at scale — infrastructure, licences, data work, headcount, a vendor relationship with switching costs. It is large, slow to unwind, and its value depends on adoption that has not happened yet.
The first should be approved quickly and often. The second deserves the same underwriting as any other durable asset, and rarely receives it.
The question is not whether the technology works. It is what would have to be true for this to have been the right amount to spend.
The cost tail Technology capital has an unusual shape. The build is visible and bounded. The run is neither. Integration maintenance, data preparation, refresh cycles, vendor escalators and the people who hold it all together are recurring, and they scale with usage rather than with revenue.
Where these budgets go wrong, it is seldom the initial number. It is that the initial number was treated as the number.
A reversibility test Before committing, four questions are worth answering in writing:
If adoption is half what we assumed, what does it cost us annually to keep running this? Which of these commitments can we exit inside twelve months, and at what price? Which operating metric should move, by when, and who reports it? What would we have to observe to conclude this was wrong — and who is permitted to say so out loud? The fourth is the one most often missing. It is also the one that decides whether a programme can be stopped or only quietly continued.
What governance looks like here Not a steering committee. A small number of named owners, a short list of operating metrics reviewed on the same calendar as everything else, and a stated review point at which continuation is an active decision rather than a default.
The businesses that will look sensible in three years are not the ones that spent least. They are the ones that can explain, commitment by commitment, what each was for and what came back.
This note concerns capital discipline and governance. It is not investment, legal or tax advice, and not a recommendation on any vendor, technology or security.