Human Performance & Leadership
The Post‑Meeting Silence Kills Decision Speed
Amazon's meetings open in total silence, and the reason explains why your decisions keep getting re-argued two weeks later.
2026-09-022 min read
An Amazon meeting starts with everyone reading. The six-page narrative memo goes around, the room stays quiet while people work through it, and only then does anyone speak. Jeff Bezos described the practice in his shareholder letters, along with its companion rule, disagree and commit, which he set out in the 2016 letter and illustrated by greenlighting a show he personally doubted would work. Both practices attack the same enemy from opposite sides: the memo forces the reasoning onto paper before the decision, and disagree-and-commit forces the dissent into the room before anyone leaves it. The silence that costs you is the other one, the quiet after the verdict, when everybody nods and nobody records why.
What that silence exposes you to is not laziness but memory. Baruch Fischhoff showed in 1975 that once people know how something turned out, they can no longer recover what they believed beforehand; the outcome quietly overwrites the earlier estimate, and they insist it was never there. Run that forward two weeks. The decision now sits inside facts nobody had in the room, a customer complaint, a slipped estimate, a competitor's release. Each person re-derives the choice from what they know today, lands somewhere the group did not, and reasonably concludes the group was wrong. Nobody is being difficult. They are reading a verdict whose premises are no longer visible.
A verdict alone cannot be audited, which is why the re-vote is the default setting of every team. Write three things down while the room is still full: the reasons that carried the decision, the strongest objection and why it lost, and the specific fact that would change the answer. The third one ends the loop. Whoever arrives in two weeks with a fresh concern either matches the trigger you named, in which case reopening is correct, or does not, in which case the answer is already on the record.
Key insights
Hindsight bias means an unrecorded decision gets re-argued from premises nobody actually held at the time.
The strongest objection and why it lost is worth more on paper than the decision itself.
Naming the fact that would change your answer turns every later objection into a yes-or-no test.
Why it matters
A decision your team cannot audit will be re-decided, and the second round costs more than the first because the window has moved.
Recorded reasons are what make disagree-and-commit enforceable; without them, commitment has nothing to point back to.
Use this tomorrow
1Take the last decision your team made, write in one sentence the single fact that would reverse it, send that sentence to everyone who was in the room, and count how many reply with a different fact.
2At the end of your next decision meeting, have each attendee state one reason they support the choice before anyone leaves, and count how many of those reasons never came up during the discussion.
Go deeper
Fischhoff's 1975 experiments handed people historical episodes and asked how likely each possible outcome had been. Those told which outcome actually occurred rated it far more predictable than people judging the same episode blind, and they denied that the knowledge had influenced them. This is not ordinary forgetting, where the original belief is retrieved and rejected; the earlier estimate is replaced without the person noticing the substitution. Asking a team to recall its reasoning weeks later therefore returns a reconstruction, confidently delivered, rather than a record.
A different repair operates before the decision rather than after it. Peter Drucker, in The Effective Executive, recounts Alfred Sloan closing a General Motors committee meeting by observing that everyone present was in complete agreement, and proposing to postpone the matter until the group could develop some disagreement and gain some understanding of what the decision was about. Sloan's premise was that unanimity arriving that fast usually means the alternatives were never examined, not that they were examined and rejected. Recording the rationale makes a decision auditable afterward; manufacturing the dissent is what makes it worth auditing.
