Outgrace
OutgraceThe Leadership Magazine, home
StrategyHow-to guide

Kill criteria for a new venture, set before the money moves

Write the stop rule, its owner and its review date before the first payment. Then every tranche of budget carries a condition that everyone can read.

Most new ventures are approved with a budget, a plan and a launch date, and no written point at which someone agrees to stop. When the money runs low, or an early result disappoints, the decision falls to whoever argues hardest that one more quarter will fix things. A board that has never written a stop rule has, in effect, decided to continue until someone runs out of patience. This guide is for the people who sign off new ventures: the chief executive, the finance lead, the chair of an investment committee, and the senior official who releases public money to a project. It shows how to write kill criteria, with a measure, a threshold, an owner and a review date, before the first payment leaves the account.

Before you start

Three things must be true before you draft a single criterion. First, one person must have the authority to stop the venture, and that person should not be the one who proposed it. Second, the budget should be released in stages, so that each stage ends in a decision rather than a drift. Third, the people approving the venture should agree on its purpose in one sentence, because a stop rule written against a vague purpose will be argued away at the first bad number. If any of these is missing, fix it first. Criteria attached to a funding decision with no named stopper are decoration.

A word on sources. This method is practitioner judgement. The sources we could check cover UK central government appraisal guidance and research on entrepreneurship training. Neither sets out a stage-gate model or the wording of kill criteria, so treat what follows as a working method to test against your own approvals, not as an established standard.

  • One named person can stop the venture without a further board vote.
  • The first tranche is small enough to lose without damaging the rest of the business.
  • The venture's purpose fits in one sentence, and every approver has read it.
  • Someone outside the sponsoring team has been asked to review the case.
  1. Step 1

    Write the decision first

    Before any measure, write down the decision the criteria serve. For most ventures it is one of three: fund the next tranche, hold the budget until a named event, or stop. Put that decision in the approval paper as a sentence with a date beside it. What usually goes wrong is that the team writes a list of metrics and only later discovers that nobody agreed what a bad number would change.

  2. Step 2

    Choose three to five measures that can fail

    Pick measures that could come back bad: paying customers from a named group, unit cost at a stated volume, a regulatory approval, or a partner's signed commitment. Page views and pilot enthusiasm rarely fail in a way that matters. Keep the list short, because a long list lets the team point to the one number that looks good.

  3. Step 3

    Put a number, a date and an owner on each

    Each criterion needs a threshold, a date by which it will be measured, and a named owner who reports the figure. Write the threshold so that a stranger could check it. For example, 'at least 400 paying customers by the end of the first quarter' is testable, while 'strong early traction' is not. The usual failure is a shared owner, and no team ever reports a bad number about itself.

  4. Step 4

    Fix the review dates before the first payment

    Set the dates for each stage review in the approval paper, not in a calendar invitation that can drift. Tie each tranche of budget to the review that comes before it. Schedule reviews early enough that a miss can change the next spend. A common failure is a review set for the month after the money has already gone out.

  5. Step 5

    Agree what a miss triggers

    For each criterion, write one response: stop, re-plan with a capped budget, or hold the next tranche until a named event. Make the response automatic wherever you can. If the response is left to the review meeting, the meeting will argue for the sponsor and the threshold becomes a suggestion. In our experience, a miss with no written response usually ends in more time rather than a decision.

  6. Step 6

    Flag the spending you cannot reverse

    Go through the budget and mark any commitment that cannot be unwound within a quarter: a lease, a multi-year supplier contract, a hire with a long notice period. Where you can, place those commitments after the review that tests the venture's core assumption. Ask the finance lead to price the flagged items twice, once at the expected cost and once in the event of a stop, so the board sees how much of the budget a stop would really recover.

  7. Step 7

    Record the criteria in the minute

    Write the criteria, owners, dates and responses into the board or committee minute that approves the funding, not only into the paper behind it. Minutes are what a new chair reads first and what an auditor will ask to see. Ask the company secretary, or the equivalent officer, to confirm that the minute states who can stop the venture.

Where it goes wrong

The most common mistake is a stop rule that the sponsor can change alone. If the sponsor can move a threshold after the first miss, the criterion is a forecast, not a condition. The second mistake is measuring too late. A criterion that reads the outcome after most of the budget is spent tests nothing useful. The third is treating a stop as a personal failure of the person who proposed the venture, which teaches owners to hide bad numbers. A stop is a decision the approvers made in advance, and the record should say so.

There is a subtler risk, and the research is careful about it. The National Bureau of Economic Research published NBER Working Paper 34755, 'Learning to Quit?', by Esther Bailey, Daniel Fehder, Eric Floyd, Yael Hochberg and Daniel J. Lee. It reports a multi-year, multi-site randomised field experiment with innovation-driven entrepreneurs. The NBER's bulletin says entrepreneurship training generally aims to raise founders' chances of success, and that training may also help founders judge when a venture should stop receiving time and resources. It adds that effective training may shut some startups down sooner while improving outcomes for others. The bulletin is equally clear that this sooner-closure effect is a possibility the study tests, not a reported finding. For a leader, the lesson is to write criteria that protect good ventures from a hasty stop as well as bad ventures from a slow one. A stop rule without a definition of success will do only one of those jobs.

Public bodies face a version of the same gap. The 2025 review of the Green Book shortened the document, merged most annexes into the core text, and reframed value for money and benefit-cost ratios. Business case guidance has moved to a separate gov.uk website, and the five case model guidance has been replaced by separate programme and project business case guidance. The sources we checked do not show the Green Book page setting out kill criteria, so a public body should look for its stop points in its own programme and project business case guidance. The Green Book describes itself in one plain line:

The Green Book is the UK government guidance on appraisal.

The Green Book, gov.uk, gov.uk
Who should own a kill criterion if the sponsor and the best analyst are the same person?

Separate them. The sponsor should not be the person who reports the figure that decides whether the venture continues. Where a small company cannot separate the roles, have the chair or a non-executive director receive the number directly from finance, and ask them to record in the minute whether the threshold was met.

How many criteria are enough?

In our judgement, three to five will cover the core assumptions: demand, cost, and one external dependency. Beyond that, the team tends to lean on whichever measure looks best. Each criterion should be checkable by someone outside the team.

What happens if a measure is not available on the review date?

Treat a missing measure as a miss unless the approvers agree a new date in the minute. Otherwise the review simply moves, and the venture keeps the budget it was meant to lose.

Does a public body have to set stop points?

Whether a given body must do so depends on its own rules and guidance, which this guide does not set out. Check the current Green Book and business case pages on gov.uk, and ask your finance and assurance team how stop points should be recorded. Where funding terms are involved, take advice from a qualified adviser.

This week, pick one venture that is in your budget or waiting for approval. Write its purpose in one sentence and list the three measures most likely to fail. Ask the named owner for a threshold, a date and a response for each. If the owner cannot give you a number, the venture is not ready for funding. Bring the draft to the next committee and ask the chair to minute who can stop it.

Newsletter

The Briefing

One email a week: the leaders we profiled, the ideas worth stealing, and nothing else.

Weekly, free, one click to leave. Read our privacy policy.