The Axe and the Scalpel: How to Cut a Technology Portfolio Without Destroying Its Value
Indecision does not spare the flagship; it quietly starves it along with everything else.
Executive Summary
The instruction, when it comes, is rarely subtle: the technology budget for next year will be forty to sixty per cent lower than this year’s, and the revised plan is due in three weeks. After a period in which the harder question was how to spend fast enough, programme managers are being handed the opposite problem with none of the tools built for it. The reflexes that surface first — cut every programme by the same percentage, or protect whatever is largest and most visible — feel fair and feel safe. Both destroy value.
This paper argues for triage: the deliberate sorting of a programme portfolio into those that must be saved, those that must be stopped, and those whose fate turns on a genuine judgement — and it sets out criteria for making that sort defensibly. The recommendation is specific. Cut on the basis of the value still to come and the cost of its loss, staged against how much benefit remains unrealised — not on the basis of sunk cost, percentage fairness, or political weight.
The argument runs against two respectable objections — the discipline of finishing what one starts, and the finance instinct to cut discretionary spend first — and answers both. It also confronts the most common response of all, which is to make no decision and let the cut happen by attrition, and shows why drift is the costliest path available. Salami-slicing is exposed as the most expensive option disguised as the most even-handed. The paper closes with the governance a credible cull requires, because a triage decision that cannot be held to will quietly reverse itself inside two quarters.
The reader who takes one thing from what follows should take this: the budget cut is arithmetic handed down from above, but how it falls is a choice made below — and that choice, not the cut itself, is what determines how much value the year preserves.
The core claim: an even cut is not a neutral cut. Reducing every programme by the same proportion is a decision to damage all of them and complete none — the appearance of fairness bought at the price of value.
The moment the question changes
For most of the last few years the pressure on anyone running a technology portfolio ran in one direction. There was more to build than there were people to build it; the new-economy narrative made almost any digital investment look prudent; and the after-effects of the millennium remediation had left budgets swollen and spending habits loose. The scarce discipline was speed.
That world has turned. The capital that financed the expansion has withdrawn, the valuations that justified the spending have fallen, and boards that once waved investment cases through are now asking a colder question: what happens if we simply stop? The programme manager who spent three years learning to accelerate is being asked, more or less overnight, to become an expert in deceleration — and the profession has almost no framework for it. Budgeting methods assume growth or steadiness. Almost none tell you how to cut a portfolio by half without cutting it at random.
Why the two reflexes fail
Under time pressure, two responses dominate. Both deserve to be stated at their strongest before they are dismantled.
The even cut
The first is the across-the-board reduction: every programme loses the same percentage, so no one can claim to have been singled out. Its appeal is procedural fairness and speed — it can be issued in a memo before lunch. Its flaw is that programmes do not deliver value in proportion to their funding. Most deliver their benefit in a lump, at completion, once a threshold of investment has been crossed. Fund a programme at sixty per cent of what completion requires and you do not get sixty per cent of the benefit; frequently you get none, having spent the sixty. An even cut applied across a portfolio therefore tips a whole set of programmes below their completion threshold at once — converting a collection of viable investments into a collection of stranded ones, uniformly and expensively.
Protecting the flagship
The second reflex is to protect the largest, most visible programme — the one with executive sponsorship and a launch date already announced — and to find the savings everywhere else. Its appeal is that it defends the thing most expensive to be seen abandoning. Its flaw is that size and visibility are not the same as value still to come. The flagship may be ninety per cent delivered, its benefits nearly banked and its remaining spend low-return, while three smaller programmes with most of their value still ahead of them are sacrificed to keep it comfortable. Protecting the flagship optimises for the avoidance of embarrassment — which is not the same objective as the preservation of value, and in a hard year the difference is the whole game.
The principle: sort by value still to come
“Sunk cost is the most seductive number in a budget crisis and the least relevant. The only spending an organisation can still manage is the spending that has not yet happened.”
Triage rests on a single reorientation: judge each programme not by what it has consumed but by what it will still deliver, and by what stopping it would cost or forfeit. Two quantities matter, and neither is what a programme has spent to date.
- Value still to come — the benefit that remains unrealised and would be gained by continuing, net of the money still required to reach it. A programme that has spent heavily but stands just short of delivering has high value still to come; one that has spent little but faces a long road to an uncertain benefit has low value still to come, however modest its past cost.
- Cost of loss — what is destroyed or forfeited by stopping: not only the abandoned benefit but breakage elsewhere, contractual penalties, and the value of options the programme was quietly preserving.
Sunk cost enters the decision only as information about the future — what remains to be spent to finish — never as a debt to be honoured. The half-built system that will never justify its remaining spend should be stopped precisely because the past spending is gone whether it continues or not.
The triage method
The sort has three destinations, applied in a defined sequence.
| Category | Test | Action |
|---|---|---|
| Protect | High value still to come; near a completion threshold; stopping forfeits benefit already almost earned | Fund to completion, even at the cost of others |
| Stop | Low value still to come; far from any benefit threshold; remaining spend outweighs remaining return | Halt cleanly and bank the saving |
| Judgement | Real value ahead but reachable only with contested funding; strategic optionality in play | Decide explicitly, at portfolio level, with reasons minuted |
The method is worked in four steps.
- Freeze and inventory. Stop new commitments and list every live programme with its remaining cost to complete, its unrealised benefit, and its nearest benefit-delivery threshold. Past spend is recorded but held in a separate column — consulted for facts, ignored for the decision.
- Sort the obvious. Most portfolios contain programmes that are plainly Protect — nearly done, high return — and plainly Stop — far from value, poor remaining return. Removing these first shrinks the hard problem to the genuinely contested middle.
- Contest the middle. The Judgement cases are decided together, against the risk appetite, funding what the survivable budget can actually complete rather than spreading it thin. It is better to finish three of six than to half-fund all six.
- Stage the survivors. Re-sequence the protected programmes so that benefit lands as early as possible, converting the portfolio from parallel slow burn to serial delivery. In a cash-constrained year, the timing of benefit is very nearly as valuable as its size.
The completion threshold is the pivot. For every programme, establish the point at which further investment starts returning realised benefit. Fund past that point or not at all — the space between is where budgets are buried alive.
A worked example
Take a composite case, invented but representative: a financial-services firm entering the new year with its change budget cut from roughly £12m to £5m, and six live programmes competing for what remains.
Two were nearly done — a payments upgrade and a regulatory-reporting rebuild — each within a few hundred thousand pounds of delivering benefits already contracted for. Both were Protect: to stop within sight of the line would forfeit almost the entire return for a trivial saving. Two more were early-stage bets on a business-to-business trading exchange whose projected volumes had evaporated along with the market that was meant to use it; both were Stop, and halting them released nearly £3m of the required saving on their own — because their value still to come had collapsed even as their business cases still displayed the old, confident numbers.
That left two in the middle: a customer-data consolidation and a self-service portal, each needing about £1.5m to finish and each defensible on its own terms. The survivable budget could complete one, not both, without starving the protected pair. The portfolio chose the data consolidation — because the portal depended on it, which made the sequence, not the merits in isolation, decisive — and deferred the portal to the following year with its design preserved. The firm finished four things instead of half-finishing six. Had it applied an even forty-per-cent cut across all six, it would in all likelihood have delivered none of them on their original timelines and banked far less benefit for the very same money.
What made the difference was not analytical sophistication; the arithmetic was crude and done on a single page. What made the difference was the willingness to say two things out loud that the even cut allows an organisation to avoid: that two programmes were being stopped outright, and that one defensible programme was being deferred so that another could actually finish. The even cut is popular precisely because it lets no one say either sentence. Triage is unpopular for the same reason — and it is the unpopular sentences, said early and plainly, that preserve the value.
The strongest objections, answered
Two serious arguments push the other way, and a paper that ignored them would be worth little.
The first is the discipline of finishing what you start — the well-founded observation that organisations which abandon programmes mid-flight breed cynicism, waste, and a reluctance to commit next time. This is true, and it is why the Stop category must be executed cleanly and explained, never left to bleed out through quiet de-funding. But the objection mistakes a virtue for a rule. Finishing what you start is sound precisely because most started things are worth finishing; it is not a reason to finish things that have become worth stopping. The honest answer to the cultural cost of stopping is to stop decisively and say why — not to keep spending in order to protect a feeling.
The second is the finance heuristic to cut discretionary spending first and shield committed or run-the-business cost. As triage of the overall budget this is reasonable. As triage within the change portfolio it misleads, because “discretionary” and “low value still to come” are not the same set. A discretionary-looking programme may sit one payment short of a large benefit, while a programme dressed as essential may be a slow march to very little. The discipline the heuristic reaches for is real; the proxy it uses — discretionary versus committed — is simply too crude to sort a portfolio by value, which is exactly what the completion-threshold test exists to do.
Governing the cull
A triage decision is not self-enforcing. Stopped programmes have sponsors who will lobby for revival; protected programmes attract the scope that was cut from elsewhere; and within two quarters the portfolio drifts back toward its pre-cut shape unless the decision is actively held. Three governance commitments hold it:
- Minute the reasons, not just the outcomes. Each Stop and each Judgement call is recorded with the value-still-to-come logic behind it, so that revival must argue against a stated rationale rather than an absence.
- Guard the survivors’ scope. Protected programmes are funded to their agreed completion, not beyond; new demand goes to the next cycle, not into the lifeboats.
- Re-run the sort on a cycle. Triage is a snapshot of a moving picture. A quarterly re-inventory catches the programme whose value still to come has changed, and adjusts before the next crisis forces another blunt cut.
The cost of not deciding
There is a third response, more common than either reflex and more damaging than both: not deciding at all. Faced with an unpleasant sort, a programme manager can let the cut happen by attrition — freeze recruitment, defer decisions, let contractors roll off, and allow every programme to slow in unison until the numbers happen to reconcile. It has the great attraction of requiring no one to be told their programme is being stopped.
It is also the most expensive path available. Attrition cuts by removing capacity, not by removing work, so every programme continues to consume management attention, licences, and fixed cost while delivering later and later. Benefit slips to the right across the entire portfolio at once; nothing is completed and nothing is cleanly stopped, so no saving is actually banked — only deferred spending that will surface again next year. Worse, the programmes that suffer most under attrition are frequently the ones nearest completion, because they are the ones with the least slack to absorb a slowdown. Indecision does not spare the flagship; it quietly starves it along with everything else.
The discipline of triage is, in part, simply the discipline of deciding on purpose rather than by drift. A stated decision can be defended, sequenced, and governed. A drift cannot be defended, because no one ever made it; it can only be discovered, later, in a portfolio that has spent its reduced budget and delivered almost none of it.
Recommendation
When the budget is halved and the plan is due in three weeks, resist the two reflexes that will present themselves as prudent — the even cut and the protected flagship — and triage instead. Sort the portfolio by value still to come and the cost of its loss; fund the protected programmes past their completion thresholds even at the expense of others; stop cleanly the programmes whose remaining spend outruns their remaining return, and say so plainly; decide the contested middle at portfolio level, sequencing the survivors so that benefit lands early. Then govern the decision so that it survives contact with the sponsors it disappoints.
The budget cut is not, in the end, the hard part; it is arithmetic imposed from above. The hard part is refusing to let the cut fall evenly — because an even cut is the one option that damages everything and completes nothing. Triage is harder to do and far easier to defend, and in a year when there is less money to waste, defensibility is very nearly the whole of the job.