Move fast on two-way doors

On reversible decisions, decide with 70% of the information you wish you had — then adjust.

Why it works

Waiting for certainty on reversible decisions is an error of excessive caution — the expected cost of the delay (missed opportunity, organizational slowness) exceeds the expected cost of acting and reversing. Bezos suggests 70% of the wished-for information is usually sufficient for two-way doors: the missing 30% is rarely decisive, and waiting for it often costs more than a wrong call would.

How to do it

1. For a two-way door, set a decision deadline (often 24–72 hours) rather than leaving it open.
2. Ask: do I have 70% of the information I would ideally want? If yes, decide.
3. Commit to the decision and implement it, with an explicit review point built in.
4. Reverse quickly and without shame if the decision proves wrong — that is the design.

Evidence

Organizational research on decision speed suggests that slower decisions are not reliably better decisions, and that decision speed positively correlates with organizational performance in high-uncertainty environments when decisions are reversible. Eisenhardt’s field study of high-velocity firms found that the fastest decision-makers used more real-time information and more alternatives at once — not less deliberation — which grounds the 70%-information heuristic Bezos articulates in the 2016 shareholder letter. (observational)

Speed advantage depends on decision reversibility and feedback speed; in slow-feedback environments, faster decisions may not allow timely correction.

Sources

• Eisenhardt (1989), "Making Fast Strategic Decisions in High-Velocity Environments", Academy of Management Journal
• Eisenhardt, K. M. (1989). Making Fast Strategic Decisions in High-Velocity Environments. Academy of Management Journal, 32(3), 543–576.
• Bezos, J. (2017). 2016 Letter to Shareholders. Amazon.com, Inc.

Common mistake

Treating the 70% rule as an excuse to decide with 40% of available information rather than as a limit on perfectionist over-deliberation.