In the traditional agile retrospective, everyone gets in a room and puts post it notes on a board. The group then sorts the post its by topic. Then there is discussion about the points brought up. I find there are a few limitations with this approach:
- It requires everyone to be in the room at the same time. This isn’t always easy. (Especially if the team is distributed or partially allocated to other teams.)
- Everyone has to watch the sorting exercise. Throwing more people at this doesn’t make it faster on a big team.
- People get worn out by the long meeting.
Phase 1 – post it generation
I searched if there was an online retrospective website and didn’t find anything. [If anyone reading this knows of one, please post a comment.] Instead of post-its, I created a google docs spreadsheet with the following columns:
- green background – What did we do well?
- red background – What should we do differently?
- blue background – What did we learn?
- purple background – What still puzzles us?
- Division – I wanted to record whether the comments were coming from a programmer or someone in a different part of the team as perceptions were likely to be different. This turned out to be a non-issue as only one non-programmer contributed content. (Which is interesting because when we did lessons learned on the mailing list, we had lots of comments from engineers. This may still happen in phase 3, but at least programmers got to frame the discussion about programming in the brainstorm phase.)
- Optional name – In case there we wanted to clarify something. Only one person choose to be anonymous. This column turned out to help me because a week after I sent out the request to participate, only the more senior members of the team had contributed content. This allowed me to “re-invite” the newer members by name. If everyone was anonymous, I would have just known that turnout was poor but not why.
Phase 2 – post it sorting
Ina corporate environment, I just sort/group the post its. With the high school team, I also edited a little. Just to avoid “I agree with so and so” because the sorted post its don’t have names on them. This took me about an hour. I did not edit the items for length even though some contributes were sentences (which is really long for a post it note.) I also added a couple post its to round out groups and get at missing subtext. I’ve added post-its while grouping in a corporate environment too when the contributed post-its dance around an issue. This is actually the first time I’ve grouped post-its without having an “other” kitchen sink category!
Now that we are ready for phase 3, the spreadsheet has three tabs:
- raw data – what people contributed
- grouped – data from tab 1 grouped by category
- a summary of the categories, # posts its in each category and a summary with some questions for discussion (in a corporate setting, I order the categories by # posts but don’t come up with questions. I did here because phase 3 will be over e-mail)
Phase 3 – discussion
This hasn’t happened yet. I’m sending the spreadsheet out today. I expect it to be like the lessons learned e-mail threads of the past though. Which is fine now that we’ve gotten the topics to talk about out there.