The Ninety Seconds That Decide Whether Your Retro Gets Honest

Twelve people join a retro call. The first person to speak sets the temperature for the next hour. If that person is your most senior engineer and they open with "honestly, fine, the sprint was fine," you have just lost half the room's real answers.

That is the problem the opening mood check solves, and it is why the format matters more than the question. A one-question survey collects everyone's answer at the same moment, before anyone hears anyone else. Voting does not do this, because votes attach to options someone else wrote. Topic columns do not do it either, because the first three cards frame what counts as a legitimate thing to be unhappy about.

What to run: a single question, scale of one to five, answered privately, revealed as a distribution. "How was the last two weeks for you?" Nothing cleverer. I have tried weather metaphors and emoji and animals-you-feel-like, and they produce charming answers that are hard to compare across sessions. A number is boring and legible.

The reveal is the actual meeting moment. You show the spread and say nothing for a few seconds. On a good sprint you get a cluster at four with one two, and the person who answered two now has cover. They are not the sole complainer in a happy room, they are a data point the room can see. I have watched people volunteer things in that gap that they would not have written on a card with their name next to it. That is the vulnerability you were hoping for when you asked everyone to "be open" at the start of the retro, except it is structural instead of aspirational.

Two failures to know about. First, if you show a spread and then move straight to the agenda, you have trained people that the check is decorative and next time everyone answers three. You have to spend at least one question on it: "there's a one and a five here, would either of you say more?" Second, do not name the low scorer. If nobody claims the one, let it sit. An unclaimed low score is information about your psychological safety, and outing the person is the fastest way to guarantee the next one is also unclaimed.

The whole thing costs ninety seconds. What you get is a room where the quiet people have already spoken once, which makes the second time much cheaper. Keep the question identical across retros and you also get a series you can look back at, which is a different use worth its own conversation.