How to Kill a Project

Most organisations have at least one project that everybody privately knows is in trouble. It still has meetings. It still appears in the quarterly update. People are still working on it, although perhaps fewer than before. The language around it becomes more careful. Nobody quite says it is succeeding, but nobody makes the decision to stop either.

Sometimes the problem is obvious. The expected demand never appeared, the economics no longer work or the strategic context changed. More often, the project has simply become difficult to justify with confidence. That is where organisations get stuck.

Starting something usually has a clear moment of decision. Stopping rarely does. Work continues one week at a time until continuation becomes the default, and the original decision to invest quietly turns into an assumption that investment should continue.

A project does not need to fail dramatically to deserve ending. It simply needs to stop earning its place.

Why stopping is harder than starting

Stopping is often described as a sunk cost problem. That is part of it, but the harder forces are usually social and organisational. Someone backed the project. Perhaps a senior leader championed it publicly. A team invested months of effort. Customers may have been told it was coming. Ending the work can easily feel like a judgement on the people involved rather than a judgement on the current case for investment.

Evidence also rarely arrives cleanly. There may be some usage, one enthusiastic customer, one encouraging metric or a plausible explanation for why the expected result has not appeared yet. Weak projects survive surprisingly well on fragments of positive evidence because complete failure is rare.

There is also a simpler problem. Projects usually have someone responsible for delivery, but responsibility for deciding whether the work should continue is often much less clear. A sponsor, product owner or programme lead may be accountable for getting the work done without anyone being explicitly accountable for asking whether it still deserves to be done.

Those are not always the same responsibility.

When nobody clearly owns the decision to continue, reshape or stop, momentum tends to make the decision instead.

Make continuation a decision

One of the simplest ways to improve stopping is to change the default. Instead of asking whether anything has gone badly enough to justify ending a project, ask whether the project has earned another period of investment.

A useful question is simple.

Knowing what we know now, would we start this again?

The point is not to punish projects for failing to meet their original assumptions perfectly. Most meaningful work changes as evidence arrives. The question is whether the current version of the opportunity still deserves the people, money, attention and continuing ownership it requires.

There are usually three legitimate answers.

Continue when the evidence still supports the case and the next period of investment is justified.

Reshape when the underlying opportunity still matters but the current solution, scope or level of commitment no longer does.

Stop when the case is no longer strong enough to justify another unit of capacity.

Making those outcomes explicit matters because too many reviews quietly treat continuation as the normal answer and everything else as an exception. A good review does the opposite. It makes continuation something that has to be earned in the present tense.

Separate the project from the person

Stopping becomes much harder when people believe their reputation is attached to the outcome. A leader who knows that ending a project will be interpreted as admitting they made a bad decision has every incentive to find one more argument for keeping it alive. A team that expects to be judged by whether its work survives will defend the work long after the evidence has weakened.

That is expensive because it turns learning into something threatening. Good judgement does not mean always being right at the beginning. It means making the best call you can with incomplete information, then being willing to change that call when reality gives you a better answer.

Judgement is not conviction.

If an organisation says it wants experimentation but punishes people when an experiment leads to stopping, it does not really want experimentation. It wants success stories. The same applies to leadership decisions. A decision can have been sensible when it was made and still be wrong to continue today.

Strong organisations protect the person while staying rigorous about the work. That distinction creates room for people to change their minds without needing to rewrite the past first.

Buy evidence, not time

Weak projects often survive because leaders are waiting for decisive evidence. Unfortunately, evidence rarely arrives wearing a label that says “stop now”. Customer behaviour may be mixed. Market conditions may still be moving. The team may reasonably argue that another quarter would reveal more.

Sometimes that is exactly the right call.

The important question is what the additional investment is buying.

If another month, quarter or phase will test a specific assumption, close an important evidence gap or reveal whether a different shape of the work performs better, continued investment may be justified. The organisation is spending capacity to learn something that could materially change the decision.

If the answer is simply “we need more time”, the organisation may be buying delay rather than evidence.

Time does not strengthen a case by itself.

Be explicit about what another period of investment will tell you that you do not already know.

Before extending a weak project, be explicit about what you expect to know afterwards that you do not know today, and what you will do differently depending on the answer. If nobody can describe that, the additional investment is difficult to justify.

Compare future value, not past effort

Stopping conversations often become dominated by what has already been invested. Months of work, a feature that is nearly finished, a customer promise, a senior sponsor or a team that has worked hard are all real considerations.

They are not, however, the right comparison for the next decision.

The relevant question is not whether the project was worth everything already spent on it. That money and time have already gone. The question is whether the future value of continuing is stronger than the future value of the best alternative use of the same capacity.

That changes the conversation considerably.

The engineers still assigned to the project cannot work on something else. The leadership attention being consumed cannot be spent elsewhere. The complexity being added will have to be maintained. The support burden, customer communication and future decisions all travel forward with the work.

The organisation is never choosing between continuing and nothing. It is choosing between this project and every credible alternative use of the same people, money and attention.

Opportunity cost becomes much harder to ignore once the alternative is visible. If stopping releases capacity, make the likely next use concrete wherever possible. Perhaps a stronger customer problem moves forward. Perhaps complexity is removed. Perhaps quality improves. Perhaps the organisation creates enough space to examine a more consequential bet properly.

The case for continuing should be stronger than the case for what continuing prevents.

Make stopping authority explicit

Another uncomfortable question is whether the person closest to delivery should also be the person who decides whether the work continues.

Sometimes the answer is yes. Sometimes it is not.

The person responsible for delivering a project often has the richest understanding of it, which makes their judgement important. They may also carry the greatest emotional, reputational and practical investment in seeing it survive. Neither fact makes them incapable of making the decision, but the potential tension should be recognised rather than ignored.

For consequential work, it should be clear who recommends, who contributes evidence and who ultimately has authority to continue, reshape or stop. That person also needs permission to use the authority in practice.

A stopping rule without someone able to act on it is only a preference.

End the obligation, not just the roadmap item

Once the decision is made, stopping properly matters. Removing a line from the roadmap is not the same as ending the commitment the organisation created.

Record the decision and the reasoning while people still remember what happened. A short account of what you believed, what changed and why the decision changed is usually enough. The aim is not to create an elaborate post-mortem. It is to preserve the judgement so the same idea does not return six months later without any memory of what was learned.

Affected people should hear the decision quickly. Customers may need communication. Internal teams may have built plans around the work. Commercial promises may need to be revisited. Dependencies elsewhere in the product or organisation may need to be unwound.

Anything already built also needs an explicit future. Some things should be deleted. Others should be archived. A small number may need to remain in a defined maintenance state. “Leave it running for now” is not neutral. Anything that continues to exist usually carries support, security, compatibility, data, integration and ownership costs with it.

Stopping is complete only when the continuing obligation is understood.

The final step is to capture what the work taught you. The most useful learning is rarely that execution could have been better. Look for the assumption underneath the project. Was the customer problem weaker than expected? Was the proposed behaviour change unrealistic? Was distribution the real constraint? Did the economics depend on adoption that never came? Did the strategic context change enough to alter the original case?

The learning may be worth keeping even when the project is not.

Build the stopping conditions earlier

The best time to make stopping easier is before the project begins. Agree what the work needs to prove, and be clear about the intended customer or business change rather than only what will be delivered. Write down the assumptions that matter most and identify the evidence that would strengthen or weaken the case.

Decide in advance what could lead to more investment, a change in shape or a stop. Name who can make that call and set a point when the work will have to re-earn its place.

This does not mean predicting every possible outcome upfront. It means creating enough clarity that the organisation can respond to evidence without renegotiating the rules once everyone is emotionally, politically and operationally invested.

Stopping becomes much harder when the criteria are invented after the result is known.

What good stopping looks like

A healthy organisation is not one that rarely stops things. It is one that can stop something without turning the decision into a crisis. People understand why the work ended. The learning survives. Customers and teams are treated properly. Any remaining obligations are explicit. Capacity is genuinely released. Nobody needs to rewrite history so the original decision looks foolish.

That capability matters even more as building becomes cheaper. When starting is expensive, friction stops many weak ideas before they become commitments. As that friction disappears, more ideas can make it further into the organisation and into the world. Better building capability therefore needs to be matched by better stopping capability.

Starting cheaply is only safe if we can also stop well.

The ability to end weak work is part of the ability to make better bets.