How to quit faster: recognising the right signals before it's too late

Executive overview

Most people quit too late — not because they lack courage, but because sunk costs and emotional bias cloud judgment in the moment. The fix is to make quitting decisions before you're in them.

Annie Duke argues that quitting is a skill, and the best decision-makers exercise it more often than amateurs, not less. The framework: define kill criteria in advance, attach a state and a date, and recruit an outside observer to hold you to them.

Sticking to the wrong thing doesn't show grit — it crowds out the right thing.

Why professionals quit more than amateurs

• Pro poker players fold ~75% of initial hands; amateurs fold less than half.
• Amateurs wait until a hand has "no chance" before folding — by then, it's expensive.
• The sunk-cost trap: "I had too much in the pot" is a reason to keep losing, not a reason to keep playing.
• The relevant question is never what you've already invested — it's whether the next bet is worth making.
• "Quit while you're ahead" is bad advice: worthwhileness, not position, determines when to quit.

The quitting bias in real decisions

• Retail investors cancel take-gain orders early (selling winners too soon) and cancel stop-loss orders to keep gambling on losers.
• Both moves mirror the coin-flip experiment: risk-averse when ahead, risk-seeking when behind.
• The popular aphorisms — "winners never quit," "if at first you don't succeed" — actively encourage the mistakes people already make.

Kill criteria: deciding in advance

• Kill criteria are pre-defined signals that trigger a quitting decision — set before you're emotionally invested.
• Sunk costs weigh too heavily in the moment; advance thinking bypasses that weight.
• A sales team example: sellers generated ~40 early-loss signals (e.g., prospect only talks price, no decision-maker in the room, second meeting not scheduled on the spot).
• Each signal was rated for severity; some were instant kill criteria, others required one follow-up action before killing.
• Deal review then rewarded qualifying out: leadership praised kills that matched criteria, with no postmortem required.
• Result: sellers concentrated effort on deals worth pursuing and stopped rationalizing unwinnable ones.

State and date: the structure of a kill criterion

• The best kill criteria combine a state (what you observe) and a date (when you'll evaluate it).
• Without a deadline, people cycle indefinitely — revisiting the same unhappy job or relationship for years without acting.
• Example: "If they don't schedule a second meeting at the first meeting" (state) + "by end of that meeting" (date) = a usable kill criterion.
• For a job or relationship: pick a quarter, define what "turned around" looks like, define what "still broken" looks like, commit to the decision at that point.
• Acknowledgement: quitting is a luxury not available to everyone — those who do have options should use them.

The quitting coach and outside perspective

• Sunk costs bind the person inside the decision; an outside observer carries none of them.
• Friends and colleagues can see clearly what's not working — but won't say so without explicit permission.
• Ron Conway (SV Angel) uses a structured approach: agree with the founder that things are hard, then ask "how long until we see hard evidence you've turned it around?" and "what would that evidence look like?"
• This stealth-coaches the founder into setting their own kill criteria and granting permission for the follow-up conversation.
• At the three-month check-in: benchmarks hit = keep going; benchmarks missed = shut it down.
• Conway's pride is in getting founders to stop early so they can go on to better work.
• To make this work: the advice-seeker must explicitly grant permission — "tell me what's in my long-term best interest, even if I won't like it."
• Default without permission: friends say "things will be okay" rather than risk hurt feelings.

The hidden downside of goals

• Goals drive persistence — but once set, the finish line becomes fixed and stops being re-evaluated.
• Cognitively, progress toward a goal registers less than distance from the finish line: 16 miles into a marathon still feels like failure if you stop.
• This causes people to keep going past the point of rational sense (or physical safety).
• Fix: add unless clauses to goals — "I'll keep running unless the medical tent says stop," "I'll keep this project unless it's over budget by X% and timeline by 50%."
• Unless clauses make goals flexible without abandoning them.
• Values change is also a valid unless clause: "I'll pursue this if it makes me happy; if I discover it doesn't, I'll re-evaluate."