Decision Autopsy

Outcome bias is the strongest bias in organisational memory: the bet that failed becomes "obviously reckless," the coin-flip that landed becomes "visionary." The autopsy separates the two questions that always get merged: was it a good decision? and did it get a good outcome? — because only the first is under anyone's control next time.

Required Inputs

• The decision — what was decided, when, by whom, and what the live alternatives were.
• What was knowable at the time — the information, constraints, and time pressure as of the decision date. Be strict: things learned afterward go in a separate pile, and the autopsy will police the boundary.
• The outcome — what actually happened, so the luck accounting has something to account.

The Forensic Frames

• The information test: given only what was knowable then, what would a calibrated outsider have chosen? (The autopsy answers this before re-examining the outcome, to keep hindsight out of the grade.)
• The process test: were alternatives really generated? Was disconfirming evidence sought or only tolerated? Was the reversibility of the choice priced in? Was a kill-criterion set?
• The luck accounting: decompose the outcome into decision quality vs. variance — what portion of the result would replay differently if the world rolled again?
• The lesson filter: the only lessons worth keeping are process lessons ("we never priced reversibility") — outcome lessons ("don't bet on X") overfit to one roll of the dice.

Output Format

1. The two verdicts, separated — Decision: 🟢 sound / 🟡 flawed / 🔴 negligent as made. Outcome: good / bad / mixed. State them side by side; the whole point is that they can disagree.
2. The knowability ledger — table: fact | knowable then? | actually known? | changed the call? Hindsight contamination gets flagged explicitly ("this entered the story after the fact").
3. The luck accounting — one honest paragraph: what fraction of this outcome was variance, with the reasoning shown.