How to Run a Retrospective with More than 50 People

The retrospectives we ran with the tech teams became quite popular in my former company. The rest of the company got interested in running their own retrospective events, so we started running retros with marketing, social media, product launch, content team, films teams, etc.

One complexity I encountered was when I was asked to run regular retrospectives with the team leaders guild (as they work on different areas and most of them had different objectives, they don’t qualify as a team). As they don’t work on delivering an interdependent task, I needed to focus the guild on how to become better leaders and learn together, rather than on how to deliver together.

This is the retrospective model I came up with, how I ran it, and how I improved it after the first couple of sessions.

The structure

I planned to set the stage by discussing the purpose of the retrospective and Norm Kerth’s Prime Directive on retrospective. For gathering insights, I planned to have them write on the board what they want to Celebrate (“Working on project X was great because…“) and what they would Do Differently. From there, I was planning to vote on the first 3-5 most important improvement areas, split the guild in groups and have them come up with actions for each area.

How I ran the retro

1. I put Norm Kerth’s Prime Directive on the screen and had everyone agree with it, I had them say Yes to the rule (or challenge the rule if they had to; needles to say nobody did). (5 min)
2. I split the group in two bigger subgroups. Each group would work on one part of the retrospective: one on the Celebrate wins, the others on Do Differently. In groups they had a sub-process: each person wrote down their ideas, shared and discussed them in groups of 4, then in the bigger group, and arranged them in themes. (15-20 min)
3. When they ended the discussion, I had them add their findings on the board. (5 min)
4. We discussed our insights, making sure that everyone understands the points on the board, ideas were added (if they were missed), and it was clear what each item on the board meant. (15 min)
5. We voted on the most important items in the Do Differently section, prioritising them and clarifying why the top items were more valuable. (5min)
6. We chose first 5 items and generated actions to address them.
7. I split the group in 5 teams, each working to develop an action plan for the top items. The goal was to quickly agree to small experiments and tweaks to try to resolve the issue being discussed; no root cause problem solving in this session. (15 min)
8. We posted the actions on the board and discussed each action, set owners (people from the group proposed themselves for the area they were expert in), and proposed an estimation for completion. (10-15 min)
9. Finally we chose 3 experiments to run until the next retrospective, as agreed by everyone in the room. (5 min)
10. At the beginning of the next retrospective, the team reviewed previous experiments, results, completion, and why it wasn’t completed (if the case).
11. We closed with a gratitude exercise (each person in the room expressed their gratitude to another person in the room). (5 min)

What I iterated on

1. I made the retrospective longer than 1.5 hours (at least two hours, considering the improvements below).
2. The people in the room gave me feedback that they would’ve preferred to work on a different area; e.g. some people form the Celebrate group had improvement ideas, some people from the Do Differently group, had things to celebrate that weren’t added to the board. Next iteration I had both group working on both areas and aggregated their answers.
3. When we started to vote on the Do Differently items, it turned out that the items were not as clear as we thought during our discussing them. So I spent more time to clarify items on the board.
4. Some of the people that worked on actions for one topic didn’t have much input for the area (they didn’t know the topic very well). Next retrospective I had smaller groups (max five people) working on all actions and coming with action plans. The plans were more innovative (and even if we didn’t agree to follow some ideas, they were used by some teams as team improvements).

Overall I created smaller groups, had everyone gather insights and build actions on all dimensions of the discussion (so that ideas are not lost in translation) and increased the time period for the retrospective.