Deciding What Not to Do: A Practical Guide to the Stop-Doing List

Every organisation has a mechanism for starting things. Almost none have a mechanism for stopping them.

That asymmetry is the whole problem. Starting is celebrated, sponsored, and attached to someone’s objectives. Stopping is awkward, politically costly, and belongs to nobody. So the initiative count only ever goes up, capacity stays fixed, and the amount of force behind any individual bet gets thinner every year until nothing moves.

Overload is rarely a resourcing failure. It is a decision failure, and it is fixable with a process.

Why stopping is structurally hard

Four forces, worth naming so you can design against them.

Sunk cost. Eighteen months and a budget already spent feel like reasons to continue. They are not. The only relevant question is the return on the next euro and the next month, not the ones already gone.

Ownership. Someone’s objectives, someone’s team, occasionally someone’s promotion case is attached to the thing. Asking them to propose killing it is asking them to argue against their own interest.

Absence of a forum. There is a meeting for approving new work. There is usually no meeting whose explicit purpose is removing work. Decisions that have no forum do not get made.

Optionality theatre. “Let us keep it running at low intensity in case.” A project on life support consumes attention, coordination cost, and headcount at a level that never shows up in the budget line, while delivering nothing. Half-funded is usually worse than either fully funded or dead.

The four kill criteria

Apply these to every active initiative. One clear hit is a discussion. Two is a decision.

1. It does not map to the strategy

If an initiative does not build one of the two or three capabilities your strategy said you must have, it is consuming resource that belongs somewhere else. This assumes, of course, that you have a strategy and not a budget with ambition attached. If you are not sure which you have, start with strategy is not planning.

In a typical first pass, 20 to 40 percent of active work maps to nothing. That is not a scandal; it is accumulated sediment from previous strategies that nobody removed.

2. It would not be approved today

The cleanest test available. If this initiative were proposed fresh this morning, with today’s information, today’s costs, and today’s competitive position, would it get funded ahead of the alternatives?

If the honest answer is no, the only thing keeping it alive is that it already exists. That is inertia, not judgement.

3. It cannot show a mechanism, only a correlation

Ask what specifically changes in customer behaviour or unit economics because of this work, and how you would know. If the answer is a metric that moves for many reasons, or an attributed number with no counterfactual, you are funding an association rather than a cause. The same discipline that separates real from imagined performance in advertising applies here: see how to measure retail media incrementality for what a real causal claim looks like.

4. It is starved

Below a threshold of resource, most initiatives cannot succeed regardless of merit. A two-year transformation with 0.4 of a full-time engineer is not a slow project; it is a stalled one that has not been admitted. If you would not double the resource, you should probably stop it. Concentration beats spread almost every time.

The process: a quarterly stop review

Ninety minutes, once a quarter, with the same seriousness as a budget review.

Step 1. Inventory everything. Every active initiative, project, report, meeting series, tool subscription, and recurring commitment. One line each: owner, resource consumed (in person-days per quarter, not vague fractions), and the outcome it claims. The inventory alone is usually the shock. Most leadership teams cannot name everything their organisation is doing.

Step 2. Score against the four criteria. Done independently before the meeting, by the leadership team, not by the owners. Owners defend; that is what ownership does. Independent scoring removes the ritual of defence from the scoring step.

Step 3. Sort into four buckets.

• Double. Strategy-critical and currently starved. These are where the freed resource goes.
• Keep. Working, correctly resourced, leave alone.
• Starve deliberately. Explicitly reduced with a stated end date. Not drifting; scheduled.
• Stop. Ends this quarter.

Step 4. Set a floor. Commit in advance that at least 10 percent of current effort will move to Stop. Without a floor, every review concludes that everything is important. The floor forces ranking, which is the actual work.

Step 5. Reallocate immediately. This is the step that gets skipped and it determines whether anyone participates honestly next quarter. If stopping something means a team simply loses capacity, nobody will ever nominate their own work. If freed capacity visibly moves to the Double bucket, the review becomes a way to get resourced rather than a way to get cut. The mechanics of that move are in how to allocate resources like a CEO.

Making a stop decision stick

Stopped things have a habit of resuming quietly. Three practices prevent it.

Write a closure note. Half a page: what we tried, what we learned, what would have to be true for us to revisit. This does two things. It converts a kill into a documented learning, which lowers the political cost enormously. And it gives the next person who proposes the same idea something to read.

Name the reversal trigger. “We would restart this if the partner channel exceeded 15 percent of revenue.” A stated trigger turns a permanent-feeling loss into a paused option, which is far easier for the owner to accept and far more honest than pretending the idea was worthless.

Remove the infrastructure. Cancel the meeting, archive the channel, close the dashboard, revoke the licence. Work resumes through its leftover scaffolding. If the weekly sync still exists, the project still exists.

The individual version

The same logic runs at personal scale, and it is where most operators feel it first. Take your calendar for the last four weeks and mark each recurring commitment: does this exist because it produces a decision or an outcome, or because it once did? Then apply criterion two. If you were building your week from scratch today, would this meeting be in it?

The answer is usually no for somewhere between a fifth and a third of a senior calendar. That reclaimed time is the cheapest capacity available to any organisation, and it costs nothing but the discomfort of one conversation per removed commitment.

What good looks like

An organisation with a functioning stop discipline has three visible traits: the number of active initiatives is roughly flat year over year rather than climbing, leadership can name the top five priorities without consulting a document, and killing something is a normal event that produces a written note rather than a political incident.

None of that requires a new framework. It requires a forum, four criteria, a floor, and the willingness to say out loud that a profitable, well-run, genuinely nice piece of work is not what this company should be spending its next quarter on.

The full accountability checklist and the operating rhythm that keeps decisions like this recurring are in Think Like a CEO.