The Retros That Stopped Producing Actions: Rebuilding a Monthly Ritual
Four consecutive retrospectives produced 37 sticky notes and 2 completed actions. The team had a ritual with no output, and the ritual itself had become the thing people complained about rather than the work.
The team was a 9-person platform group inside a retail technology company. Retrospectives ran monthly, lasted 60 minutes, and had been facilitated by the same person since the team formed.
How 37 notes became 2 actions
Notes were grouped into themes, themes were discussed, and the session ended with a list of things the team should do differently. The list had no owner and no date, so it functioned as an aspiration rather than a commitment.
Two actions were completed in four months. Both were completed by the person who had raised them, without any follow-up.
The narrowing rule
The first change was arithmetic: one action per retrospective, with an owner and a date, chosen before the session ended. One action, not five. The team resisted because it felt like dropping good ideas, and the counter-argument was that the previous approach had dropped all of them.
The chosen action had to be something the team could change without asking permission. That constraint eliminated about half the candidates immediately.
The first three months
Three actions were chosen and three were completed: a written definition of done, a limit of two concurrent reviews per person, and a fifteen-minute rule for blocking questions. All three were small, and all three changed behaviour within a fortnight.
Completion was possible because each action was small enough to fit in a week and specific enough to be checked.
What the format lost
Discussion shortened. With one action and a 45-minute budget, the session spends about 30 minutes on the topic that produced the action and 15 on everything else. Two team members said the conversations had become narrower and less useful for surfacing things that were not yet problems.
The facilitator added a ten-minute open section at the start with no action requirement, which restored some of the range without producing a second list.
The follow-up mechanism
Each action was added to the team’s board as a card with the owner’s name. At the next retrospective the first five minutes reviewed the previous action, and the review was factual: done, partially done, or not started.
The word partially was eventually removed, because it was used for actions that had not started. The team replaced it with a two-option check.
The retrospective that produced nothing
One session failed to choose an action, because the discussion kept returning to a problem the team could not change without a decision from outside. The team recorded the block and wrote a one-paragraph note to the engineering manager instead of an action.
That note produced a change three weeks later, which the team counts as a success even though it never appeared on the board.
Who facilitates
Facilitation rotated for the first time in the fourth month. The rotating facilitator changed the questions asked, which changed what surfaced. Two of the four actions chosen in that period came from topics the previous facilitator had never raised.
Rotation costs about twenty minutes of preparation per session and has been worth it.
Nine months on
Actions completed rose from 2 in four months to 11 in nine. The retrospective is still monthly and still 60 minutes, and attendance has stopped being an issue.
The action that outlived the team
One action, the limit of two concurrent reviews per person, was adopted by two other teams and is now written into the engineering handbook. It came from a session that a facilitator described as unremarkable, which is the ordinary way a useful change arrives.
None of the 37 sticky notes from the earlier period produced anything that outlived the quarter.
Thirty-seven notes once produced two completed actions. The sessions now produce one action each and most of them finish, which is a smaller ambition and a much shorter list of things the team meant to do.
