Management & Thinking
Pre-mortem
事前失敗分析法 · Source: Gary Klein
Stress-testing a plan before committing — finding failure paths you haven't seen yet
Core Concept
Assume the decision has already been made — and it failed completely. Now work backwards: why did it fail? Forcing yourself into a "failure has happened" mindset often reveals weaknesses invisible during optimism.
✓ When to use this
Before making big-bet, hard-to-reverse decisions — fundraising commitments, product launches, contract signings, team restructures. The more excited you and the team feel, the more this matters.
✗ When not to use this
Skip this for small reversible day-to-day choices — it becomes over-analysis. Also unsuitable in pure exploration where there is no concrete plan yet to "fail."
Questions you will be asked
Using this framework, you will work through —
- 1.What is the decision or plan you're considering?
- 2.Now assume: one year later, this decision has failed completely. What do you see?
- 3.List three to five specific reasons it could fail
- …and 3 more
Worked example
Expand to see what a filled-in run looks like
›
Worked example
Expand to see what a filled-in run looks like
Situation
考慮辭掉穩定的工程師工作、全職投入自己的 SaaS side project（已經有 200 個付費使用者，月營收約 6 萬）。家裡有 18 個月生活費。
1. What is the decision or plan you're considering?
在 12 個月內把 side project 做成正職、不再回去上班。
2. Now assume: one year later, this decision has failed completely. What do you see?
一年後我看到自己又回到求職市場，存款只剩 4 個月，履歷上多了一段「創業失敗」需要解釋；產品停在 250 個使用者沒長大。
3. List three to five specific reasons it could fail
1) 沒有持續的銷售管道，客戶來自一波運氣；2) 我低估了一個人做客服 + 開發的疲勞；3) 競品更新速度比我快；4) 沒設定明確的「該收手」門檻；5) 配偶承受太多風險，關係惡化。
4. Which of these failure reasons are you least prepared for right now?
第 2 點——我從沒長期一個人扛多角色，這部分我完全沒準備。
5. Knowing this weakness, what would you do differently right now?
辭職前先用 3 個月「假日 + 每週請 2 天假」測試獨自運作的負擔感；同步建立第二個獲客管道（內容行銷）再辭。
6. After this exercise, has your decision changed?
修改後繼續：把全職時間從「立刻」延後到「驗證 3 個月後」。
Use it inside ChatGPT / Claude
Paste the prompt below and the AI will walk you through this framework, one question at a time.
你現在是一位幫使用者做 Pre-mortem 的決策教練（基於 Gary Klein 的方法）。 你的任務是引導使用者假設他正在考慮的決策已經失敗，並倒推失敗原因，找出尚未準備好的盲點。 依序問下列六題： 1) 你正在考慮的決策或計畫是什麼？ 2) 假設一年後這個決策徹底失敗了，你看到了什麼？（要他描述具體景象，不要說「可能會」） 3) 列出 3–5 個導致失敗的具體原因。 4) 這些原因中，你目前最沒準備好應對哪一個？ 5) 知道這個弱點，你現在會做什麼不同的事？ 6) 這個練習後，你的決策方向改變了嗎？ 互動規則： 1. 一次只問一題，等使用者回答後再進入下一題。 2. 使用者答完所有題目前，不要做總結或下結論。 3. 若答案太抽象、太籠統，請追問一次具體例子或數字後再繼續。 4. 全部答完後，輸出三段：(a) 摘要使用者的關鍵判斷；(b) 你看到的盲點或張力；(c) 一個具體下一步行動建議。 5. 不要替使用者做決定，只把判斷攤開讓他自己決定。
Related Frameworks
Management & Thinking
Inversion
Finding "how to guarantee failure" in order to reverse-engineer "how to guarantee success"
Management & Thinking
Second-Order Thinking
Complex decisions with cascading effects, strategic choices, any "obvious" decision — because first-order outcomes everyone sees are often already priced in
Management & Thinking
One-way Door vs. Two-way Door
Calibrating how much care a decision deserves — avoiding over-deliberation on reversible choices and under-deliberation on irreversible ones
FAQ
How is pre-mortem different from post-mortem?
The time frame is reversed. Post-mortem analyzes "why did it fail" after the fact; pre-mortem assumes future failure and reasons backward to what to watch now. Gary Klein's research shows pre-mortems surface 30% more risks than standard risk assessment — the "assume it failed" framing flips people from defender to detective.
What if the team is too excited and won't voice concerns?
This is exactly what pre-mortem solves. Tactics: (1) don't ask "what could go wrong" (gets pushed back) — ask "imagine it's failed badly two years from now; name three reasons" — forcing the failure frame; (2) collect anonymously; (3) reward problem-finding — explicitly say "the person who names the most lethal risk gets recognized" — flipping the incentive from "don't say" to "want to say."
Should I do this for small decisions too?
No. The cost is 30–60 team-minutes — worth it for decisions where failure cost vastly exceeds that time: product launch, job change, marriage, founding a company, buying a house. Doing it for "what's for lunch" is theater. The test: if this decision goes wrong, how long to recover? Days — skip. Months/years — do it.
Related studies