While I regularly change all other retrospective exercises, the action planning technique I’ve been using for the past two years has worked so well that I don’t want to change it. That has not always been the case.
Action planning in my earlier retrospective often had some problems. Some of the action items were just good intentions without enough concrete details to make them actionable. These items focused on some longer-term goal that couldn’t be achieved during the next sprint. The result of this was that nothing actually happened. “Improve communication” is a good example for such an action. Other action items had the opposite problem. These items were very concrete but lacked a clear goal. For example, one action that came out of many retrospectives was to change the time of the daily scrum. We easily accomplished that action, but didn’t move any closer toward our overall goals as a result. If actions aren’t tied to goals, the team gets stuck in short-term thinking and leaves the longer-term, harder improvements simply because they cannot do them within the next sprint.
To solve these problems, I ask the team to generate all actions in a specific format, as shown below.
Long-term goal: Have test automation on acceptance-test level
Now-Action: Pete will automate one test using Fit
This format helps the team consider a long-term goal for every action. It also helps them create very concrete actions to move the team a step closer to the long-term goal. The now-action has to be one that can be implemented in the next sprint and must be something the team can accomplish itself. The question to ask is, “What can we do?” This does not mean that the action or the goal has to be within the team’s authority. For example, the team might need new, expensive test equipment. The team might not have budget authority to make such a decision. In this situation, the long-term goal could be, “Have enough test equipment so there is no waiting waste”. The now-action might be, “Make a cost analysis for purchasing test equipment and share this with the person authorized to approve the purchase.”
Every sprint retrospective starts with a review of the previous now-actions. If the now-actions are not completed, we invest time, perhaps the whole retrospective, discussing why not and how we could change that. If the now-actions from retrospectives are not complete, then it’s probably no use to even have retrospectives since no improvements are being made.
Action generation and prioritization
Having an action format is of little value if we do not have a method for generating those actions. The first step in action generation is to have every team member individually generate as many actions as possible. Every action is written in the established format (goal and now-action) and written on an index card. This activity is timeboxed to about ten minutes. After the individual actions are generated, we divide the group into pairs. The individuals in each pair explain each other’s actions to each other. The pair selects the most important actions out of their combined actions, trying to limit the total to a handful of actions, for example five. In this phase they can, of course, generate new actions. This activity is also timeboxed, though it’s often finished before the end of the allotted time. Next, the pairs join with another pair to form groups of four and repeat the process, this time sharing only the most important actions they selected from the previous activity. The foursome further pares down the chosen actions, to for example four. This process continues until there is a whole team discussion. At that point the team selects the actions that have slowly emerged to be the most important.
At the end of the selection process, I write all the actions on one flipchart sheet and hand it over to one of the team members. This sheet is then hung in the team’s workspace as a visual reminder. Those actions not selected can be discarded.
For example, in a retrospective with eight people, the following steps would be taken:
- 10 min – Individual action generation.
- 10 min – Pairs select the top five of their shared actions
- 10 min – Groups of four select the top four of their ten shared actions
- 10 min – The whole group selects the top three of their eight shared actions
The process is based on having team sizes that are a multiple of two; however, with some creativity it works for any size group. It might mean that in the first step there is one group of three or in the second step one group of six and one group of four. (The calculations did give me and my co-facilitator a headache in a group of around forty.)
The advantage of this action generation technique is that it combines individual action generation with group action generation. It then slowly, step-by-step, creates consensus on the actions. Everyone tends to stay involved in the process mostly because they have all been involved in the process from the beginning. Also by slowly increasing the group sizes, the more silent personalities tend not to be excluded.
While we generate as many actions as we can, we ultimately select only a few, typically three, to be done in the next sprint. Selecting too many actions is a common mistake in action planning. The effect is a loss of focus and a huge amount of time spent on tracking the actions.
Sprint retrospectives should be linked so that they build on each other and focus on long-term, continuous improvements. One way to link retrospectives is to bring the actions from the previous retrospective into the current one and discuss whether they are complete or not.
Another way of linking retrospectives is to keep the long-term goals from every retrospective, typically by writing them on a flipchart sheet. In every retrospective, then, we display the long-term goals and use them as input for the action generation.
Can we generate new now-actions for these long-term goals? Do we have any new long-term goals that we need to add to the list? Every now and then we go through the list to see which goals are still valid and which ones we achieved. We remove all invalid or complete actions.
I’ve used the action planning technique presented here in countless retrospectives over the past two years. Without fail, it has resulted in a small number of agreed-upon actions. Whatever other retrospective exercises you use can serve as input for the action planning. This action planning exercise should happen at the end of a retrospective. I hope that sharing this technique will help other teams improve their own retrospectives.