brain/
conceptpsychology

Kill criteria: precommit the quit while you're not in it

Notes

Kill criteria: precommit the quit while you're not in it

One-line summary: The worst time to decide whether to quit is while you are in the thing. Write the failure and success signals at the start, turn them into if-then quitting rules, and manage to those rules — not only to outcomes. "Would I buy it today?" does not work.

What it is

Duke's name for a turnaround time applied to ordinary projects. Kahneman (via Duke): "the worst time to make a decision is when you're in it" (cupcake in front of you after you've decided to eat healthier).

Barry Staw's finding, as she recounts it: instructing people to "treat the decision like it's fresh" / "what would you do if you had to buy it today?" made no difference. What did help was, at the initial choice, listing the benchmarks that would show the allocation succeeding or failing — then making the allocation. That is a kill criterion: a signal specified in advance, while you are not yet defending a sunk identity.

Worked examples in this interview:

  • mParticle sellers. Independently and asynchronously, ~40 sellers listed early signals of a lost deal. Common ones: first meeting is only about price; cannot get a decision-maker in the room. Turned into rules: price-only → kill immediately; no executive → offer executive alignment next meeting, kill if refused. Management consequence: two ways to win — close the deal or follow the kill criteria — so quitting is a graded success, not a pass-fail loss.
  • Ron Conway / SV Angel. When a founder says they can turn it around, Conway agrees, jointly specifies what "turned around" looks like in two months, and precommits: miss the benchmarks → return capital. He usually thinks they should shut down that day; two months is still cheaper than two years (Astro Teller's $2M vs $9M).
  • Quitting coach. Kahneman's is Thaler: "someone who loves you but doesn't care much about hurt feelings in the moment." The coach is not in it — no sunk cost, endowment, identity, or career risk on the original allocation.

Kill criteria are the advanced, on-your-own-timeline version; the coach is the same idea on someone else's timeline.

Why it matters to psychology

This is implementation-intention / Ulysses-contract work aimed at escalation of commitment. It also changes the reinforcement of quitting: if the only rewarded outcome is "closed the deal / summitted," grit wins by construction. Dual success metrics (win or clean kill) are a behavior-change design, not a slogan.

Evidence

Implications

  • Write kill criteria before the first meeting / first allocation / summit day, asynchronously if a group (she points to How to Decide for why).
  • Stop using "would I buy it today?" as a rationality ritual; it does not clear cognitive debris.
  • Managers who only score closed-won will punish good quits. Score the kill.

Open questions

  • How specific must a kill criterion be to survive motivated re-interpretation once you are in it?
  • Does a quitting coach work without the two-month benchmark contract, or is the precommitment doing the work?

Related

Referenced by
Concepts
Bankroll management: loss limits as the lesser irrationalityMonkeys and pedestals: do the hard uncertain thing firstOptionality: two-way doors make starting possiblePre-mortem analysisQuitting as a skill, not a vice
Entities
Annie Duke