A nine-person cross-team session reaches a decision at minute 48 of a 50-minute slot and it is recorded the same afternoon. Two days later two attendees have separately told their own teams they disagreed, and by day four one team has stopped work on it. What failed, and which design of the session made it possible?

The trigger

The decision was taken at the moment the room ran out of time, and agreement was inferred from nobody objecting in the last ninety seconds. Silence at minute 48 measures fatigue and social cost, not consent. In a room of nine, three or four people typically do most of the talking, so the quietest participants are the least likely to have been asked anything directly, and they are exactly the two who later dissented.

Why it propagated

Because the attendees left with no obligation about what they would say next. Each went back to their own team and retold the decision in their own words, with their own caveats attached. The caveat is the part a team remembers, because it is the part that sounds like their manager's judgement rather than someone else's. Within two days, two teams hold a version of the decision that includes the reason not to do it, and neither version is written anywhere.

The recorded decision and the transmitted decision are different artefacts, and only one of them reaches the people who will do the work. Facilitation that ends at the decision record manages the wrong one.

Why detection lagged

Nothing in the process asked anyone to state a position on the record, so disagreement had no channel except the informal one, where it is invisible to the facilitator. The first observable symptom is a delivery signal: a team's work slows or stops. That is four days of lag on a two-day-old problem, and by then reopening the decision costs the credibility of the forum.

The structural fix

End the session with two steps, in this order, before anyone leaves:

1. A round of named positions. Each participant says one of three things: support, support with a reservation, or object; reservations and objections are written down with the person's name and their reason. This converts silence into a position, and it takes about a minute per person, which is why the slot must be planned at 40 minutes of discussion rather than 50.
2. A statement of what each person will tell their team, in one sentence. Discrepancies surface immediately, in the room, at zero cost. This is the step almost nobody runs and it is the one that prevents the hallway version.

Then the record carries the dissents and their revisit conditions, so a dissenter has somewhere legitimate to have been.

The tempting local fix

Longer meetings, or a second meeting to confirm. A 90-minute slot reaches the same fatigue point later, and the confirmation meeting invites relitigation, which teaches the room that holding out is cheaper than objecting. The problem is not duration, it is that assent was never individually requested.

When this is the wrong answer

A two-person decision inside one team does not need a positions round; the overhead is comically out of proportion and the two people will simply talk. The mechanism starts paying when the decision must be implemented by people who were not in the room, which is the same threshold at which a written record becomes worth the effort.

Common weak answers

• "They should have spoken up." Correct and unactionable. The session was designed so that not speaking up was the cheapest option available.
• "Escalate to a decision-maker." Produces compliance and the same hallway retelling, now with resentment attached.