When Portfolio Leaders Should Stop a Project

“We have spent too much to stop now” sounds like prudence. It is often a refusal to make the decision that is actually required.

The money already spent may be material for financial reporting, public accountability, relationships, and learning. But if it cannot be recovered, it cannot make the next pound, hour, or unit of leadership attention more valuable. The portfolio question is forward-looking: given what we now know, which use of the resources still available creates the best outcome—including the option to stop responsibly?

Stopping is therefore not automatically a verdict on the people who approved or delivered the work. It is an investment decision made under changed evidence. A project may need to continue, reset, pause, or stop. What matters is whether leaders compare those options honestly, rather than using past effort to protect the original choice.

HM Treasury’s Green Book is explicit in its public-sector context: sunk costs—costs already incurred and unable to be changed—should not affect what decision makers do next. Appraisal should compare the future costs, benefits, risks, and opportunity costs of the options still available. This is not a command to ignore history. It is a discipline for preventing history from being mistaken for a future benefit.

Separate sunk cost from the value that remains

Three terms need to be kept apart.

Term	What it means	What it should do in a stop decision
Sunk cost	Money, time, or effort already spent and not recoverable	Record it for accountability and learning; do not use it to prove that continuing is worthwhile.
Remaining value	The benefits, assets, capabilities, obligations, or avoided harms the next investment could still produce	Estimate and challenge it using current evidence, not the original business case alone.
Opportunity cost	The value forgone by using scarce funding, people, leadership attention, capacity, or time here rather than elsewhere	Compare the project with realistic alternative uses of the constrained resource.
Exit cost	The future cost of stopping: contract close-out, transition, redundancy, service continuity, decommissioning, data, asset, and reputation consequences	Include it in the forward comparison; a responsible stop is not cost-free.

The meta-analysis of the sunk-cost effect documents a substantial research literature on how prior investment can influence economic decisions. That does not prove that a particular sponsor or programme is biased. It does explain why a formal future-only case is useful when the original decision is personally, politically, or financially visible.

In public appraisal, opportunity cost means the value of the next best use of a resource. The same logic applies in a corporate or nonprofit portfolio, even though the relevant value measures differ. Funding may be constrained. So may specialist engineers, customer-access windows, executive attention, regulatory capacity, or operational-change tolerance. A project can still have a positive case in isolation and be the wrong use of the scarce resource in the portfolio.

A recovery plan is not an investment decision

Delivery teams are expected to recover work when a problem appears. They should investigate root causes, narrow scope, change a sequence, renegotiate a dependency, or improve a control. That is recovery management.

A portfolio leader has a different job: to decide whether the organisation should make the next investment at all. A convincing recovery plan does not, by itself, answer that question. It may show a path to a milestone while leaving unresolved whether the milestone still leads to a valuable outcome, whether the risk is acceptable, or whether another initiative deserves the capacity more.

This distinction becomes important when a project drifts into “recovery theatre”: increasingly detailed plans describe how the same objective might still be delivered, but nobody has refreshed the objective, the benefit case, the external conditions, or the alternative uses of the remaining resources.

The appropriate question is not, “Can the team get back to green?” It is, “Does the option to continue, with a credible recovery plan, still outperform reset, pause, and stop?” The answer may be yes. It should be supported by a new case rather than by loyalty to the old one.

Compare four future options

Avoid a false choice between “continue unchanged” and “cancel.” Most consequential reviews should consider four options.

Option	Appropriate when	Evidence the decision maker needs
Continue	The strategic problem, remaining benefit case, feasibility, and risk tolerance still hold, and recovery action is credible	Remaining costs and timeline; refreshed benefits; dependency and capacity evidence; recovery milestones; conditions that would trigger another review
Reset or reframe	The original route no longer works but a materially different scope, sequencing, operating model, or benefit path may still create value	A new objective and option case; revised feasibility and governance; funding and transition implications; proof that “reset” is more than a relabelled delay
Pause	Uncertainty is temporary and a real option can be preserved at a proportionate holding cost	What is being preserved; what it costs; expiry/restart conditions; service and people controls; a firm review date and authority
Stop	Remaining value is no longer sufficient, the strategy has changed, risk is unacceptable, affordability fails, or a better use of resources dominates	Exit and transition plan; contract, people, service, asset, data, and reputation implications; residual value/learning transfer; released capacity and funding decision

The National Audit Office’s work on projects leaving the UK Government Major Projects Portfolio makes the useful point that cancellation is not necessarily a poor outcome when objectives, value, or policy conditions have changed. It also says that cancelled projects still need to yield lessons for future decisions. In other words, cancellation is neither proof of leadership nor proof of failure. It is one possible portfolio response to current evidence.

Reset deserves particular care. The NAO’s lessons on resetting major programmes note that changing circumstances can mean a programme can no longer achieve its intended outcomes, or can do so only at excessive cost. A reset may be necessary. But it must establish a refreshed route to an outcome—new scope, objectives, controls, costs, and evidence—not merely grant the original plan a new deadline.

Set triggers before the decision becomes personal

Do not wait for a public failure to reopen the case. At approval, and at each material re-baseline, define review triggers: conditions that require the accountable authority to compare future options again. They are prompts for judgment, not automatic cancellation thresholds.

Useful triggers include:

• a change in the strategic objective, user need, policy, market condition, or external dependency;
• a material change in the remaining cost, schedule, benefit, affordability, risk exposure, or regulatory/safety/control position;
• failure of a critical assumption or a recovery milestone;
• evidence that a benefit owner, adoption path, or operating model no longer exists;
• a new initiative that competes for the same constrained capacity and has a stronger case;
• a contractual, supplier, data, or technology event that changes the feasible options.

When triggered, commission a short future-only option case. It should identify the decision, the options, remaining costs and benefits, unmonetised consequences, risks and uncertainties, key assumptions, switching conditions, exit costs, opportunity cost, and recommendation. The Green Book’s caution that a single metric is not sufficient for a balanced value-for-money judgment is relevant here: a score or benefit-cost ratio may inform a choice, but cannot settle strategy, risk, distributional consequences, capability, or timing on its own. HM Treasury, The Green Book

Add an independent challenge. The reviewer should not be asked whether the delivery team is working hard enough. They should test whether the remaining evidence supports the recommended option, whether the alternatives are real, and whether the future costs of stopping or continuing have been omitted. The NAO argues that assurance for high-risk projects should be independent, outcome-focused, evidence-based, and capable of triggering intervention. NAO, Assurance for high risk projects

Stop a project without abandoning its obligations

“Stop” is a portfolio decision. It is not an instruction to walk away from commitments.

The exit plan should make five things explicit:

1. People and stakeholders: who is affected, how decisions are communicated, what work or skills transfer, and what support or consultation is required.
2. Service, customer, and benefit continuity: whether a live service, interim process, customer promise, regulatory duty, or benefit owner needs a controlled handover.
3. Commercial and legal position: contract rights, liabilities, supplier notice, intellectual property, assets, data, records, and dispute risks—assessed by the appropriate specialists.
4. Technical and operational safety: secure decommissioning, access removal, data retention/disposal, operational ownership, and controls for anything left running.
5. Learning and residual value: studies, designs, prototypes, supplier knowledge, data, relationships, or capabilities that remain useful to another option.

The NAO case on the first UK carbon-capture demonstration competition illustrates why this nuance matters. The competition was cancelled on value-for-money and affordability grounds after four years. The audit reported that completed engineering and design studies still contributed knowledge for later work, while criticising the lack of timely, detailed consideration of commercial risk during continuation decisions. It is a historical, sector-specific case—not a template for ordinary projects—but it shows that stopping an investment can retain learning value while exposing weaknesses that should be corrected in the next decision.

Use a stop-decision protocol

Download the project stop-decision protocol. Use it for a proportionate review, not as a ritual designed to rationalise a choice already made.

1. Trigger the review from a material change or a pre-agreed condition.
2. Prepare a future-only comparison of continue, reset/reframe, pause, and stop. Keep past spend visible for accountability, but out of the future-value calculation.
3. Test remaining value, strategic fit, feasibility, risk, opportunity cost, and exit consequences. Identify the assumptions that would change the answer.
4. Obtain independent challenge and the specialist input required by the investment’s contracts, people, data, service, safety, regulatory, financial, and risk context.
5. Record the accountable authority, options considered, evidence, decision, dissent, conditions, released capacity, transition owner, and review date.
6. Transfer the assets and learning that remain valuable. Do not rewrite the original decision with hindsight or turn the review into a blame exercise.

The decisive moment is not when a project is obviously failing. It is when its future case has changed enough that continuing without a new comparison has become the riskier choice. Portfolio leaders should make that comparison early, calmly, and in a form that allows a responsible stop to be understood as disciplined allocation—not defeat.

Sources and editorial limits

• HM Treasury, The Green Book — current UK public-sector appraisal context for forward-looking option assessment, sunk costs, opportunity cost, uncertainty, and balanced judgment.
• NAO, Projects leaving the Government Major Projects Portfolio, NAO, Resetting major programmes, and NAO, Assurance for high risk projects — public-audit context for cancellation, reset, learning, and evidence-based assurance.
• NAO, Carbon capture and storage: lessons from the competition for the first UK demonstration — a bounded public case used to illustrate cancellation, learning value, affordability, and commercial-risk considerations.
• Ruger, On the sunk-cost effect in economic decision-making — research context for bias awareness, not proof that any individual project decision is biased.

The four-option comparison, trigger set, and stop-decision protocol are Strategica synthesis. This article does not prescribe a universal stop threshold or replace any formal authority, contract, statute, accounting rule, employment process, safety obligation, regulatory requirement, or specialist advice.