In my coaching work, I have often been asked a question about the planning process for iterations, that until just a few days ago, I would actually brush off!!! I didn’t even realize I was doing this, it is only in retrospect that I see this. This question is simple: “how does a team plan for the improvement efforts that come out of the retrospective when they are supposed to be working at maximum velocity when implementing tasks directly related to the items in the work queue?” Or, more simply, “we don’t have time for process improvements.”
The source of the problem shows up in the normal* structure of an iteration. This structure is as follows:
Step 1: set a goal by selecting work from the top of the work queue.
Step 2: plan the execution of that work by doing a task breakdown and task-level estimation.
Step 3: do the work according to the tasks (this is the bulk of the time in the iteration).
Step 4: do a demo to review the work – use this demo to adjust the work queue if necessary.
Step 5: do a retrospective to review the “how” and come up with action items to make the work more effective next iteration.
Since iterations follow one after another, this structure when rolled up can be re-written in the following manner:
Action (step 3)
Reflection, Learning (step 4)
Reflection, Learning, Planning (step 5)
Planning (step 1 and 2)
Now if we look at the purpose of the steps in comparison to the type of learning activity taking place, we see that there is a big disconnect between the planning that happens in step 5 (the retrospective) and the planning that happens in steps 1 and 2 at the start of the next iteration. Not only that, but reflection and learning are separated into “what” and “how” so are done twice for each iteration.
Frankly, this causes a lot of confusion: I see it in my coaching when teams try to accommodate the two types of planning, and I see it in my training when I teach the structure of the iteration.
So what is to be done? Simplify the structure of the iteration. Instead of thinking of the iteration as a linear block of time that starts with step 1 and ends with step 5, think of it as a cycle that is dominated by Action, and punctuated regularly by Reflection, Learning and Planning. Now, instead of four separate meetings, there is a single meeting every iteration that combines the wrap-up of the previous iteration with the start of the next. Let’s call this the Iteration Transition Meeting or the RLP Meeting or the Adapt Meeting… (suggestions welcome!)
The agenda for this meeting is very simple and contains the essential elements that exist in the “old” style. It starts with Reflection where the facts of the Action are examined: what did we build? how did we build it? how did we feel about what we built? etc. This moves fairly quickly to the Learning component where we try to understand what the facts are telling us: what went well? what needs improvement? what insights can we glean? what new questions are open to us? and of course, what did we learn? Now we are prepared to treat Planning holistically, with no “hidden” activities. We examine our Work Queue, re-arrange it, put new items on it, etc. and do our normal task breakdown and estimation. The difference this time is that both direct business tasks and process improvement tasks are included in the same planning activity. This makes task level (commitment) velocity more truthful!
I will admit that this is not something that I have tried yet. I have seen it done accidentally when people start to ignore the action items from the retrospective in the old iteration structure and instead include them in the task planning. When it happened, I thought it was “messy” because it broke the separation between “what” and “how” oriented activities. Now I think that this might be the best way to do things.
Anyone out there willing to give this a try? If so, I would love to spend some time with you working on it – helping, experimenting, and learning from it.
I should mention a couple important points that didn’t fit in the above narrative. First, this insight happened as a result of discussions over the past week with my father, Garry Berteig who is using agile methods and the learning circle in his classroom environment. Second, Garry has been doing what I just described in his classroom environment; the main difference is that Garry provides a great deal of the Planning to his students by way of the progression of assignments/activities.
One other cool thing: I’m in Frankfurt as I write this on my way to Chennai to deliver a three-day Agile Project Management / ScrumMaster Certification course. I haven’t done international travel since I went to Beijing last summer, and this is the first time I am going outside of North America for work! Fun!