Kill Criteria

A set of signals, defined in advance, that tell you it is time to stop or pivot — each one paired with a pre-committed action. Set out by Annie Duke as the operational payload of a pre-mortem. The signals are useless on their own; what makes them work is deciding the response before sunk-cost and status-quo bias arrive.

The problem they solve

Once a project launches, teams are very slow to kill it. By the time anyone shuts it down — over budget, past deadline — it is clear they should have stopped far earlier. The culprits are well-known biases: sunk cost above all, plus the endowment effect (we over-value what we have built) and identity (stopping feels like failing). Hoping that you will simply act rationally when the warning signs appear is, in Duke’s image, why people climb Everest into a blizzard.

Pre-mortem alone is not enough

A pre-mortem — imagining the project has already failed and listing the early signals that foretold it — rarely changes the plan by itself. Duke’s research with Maurice Schweitzer and Linnea Gandhi found people run the exercise and then proceed largely unchanged. The pre-mortem’s real value is that it generates the signals that become kill criteria.

How to build them

1. Prompt the pre-mortem. Ask the team to imagine a specific failure six months out and name the early signals. Duke’s example to a sales team: ‘You worked an RFP for six months and the deal is dead. What were the early signs?’
2. Pair each signal with a pre-committed action — kill outright, or a specific test followed by kill.

Worked sales-team example:

• RFP written with a competitor in mind → ask directly whether they are working with a competitor and how far along; kill or pursue on the answer.
• Prospect won’t demo, only wants to discuss price → kill. You are a box-checking exercise to beat someone else down on price.
• No decision-maker in the room after the first meetings → offer executive alignment at the next meeting; if declined, kill.

Why pre-commitment is the active ingredient

Deciding the action in advance removes the in-the-moment judgment, when bias is strongest and the temptation to rationalise is highest. It also gives cover: a salesperson who kills a deal against a pre-agreed criterion is executing policy, not losing nerve — which dissolves the usual ‘why did you stop pursuing that deal?’ penalty that keeps doomed work alive.

Source

See Annie Duke on Better Decisions, Kill Criteria, and When to Quit and Quit. Related: Nominal Group Technique, Thinking in Bets.