Do We Need Points To Generate a Release Burn-up Chart?
I attended a presentation recently which talked about issues with estimation. It was pretty interesting, but one thing I thought was particularly interesting was the idea that someone said that they could track the progress of their release by generating a release burn-up chart in terms of the number of stories completed (what I call “card counts”) and it would have just as much information as if we had used story points.
I decided to see if I could replicate the idea. So I went into our database and basically pulled out a spreadsheet of the closure of stories, plotting both the stories completed and the number completed, week by week, over the period of the release (in this case a year). This is the chart I got (note: no manipulation except add titles and legends):
Interesting, right? My knee-jerk reaction when I saw this was “well perhaps we don't need to do point based estimates to allow us to understand progress for a release.” Can anyone say “#noestimates”?
Thinking about this a little further, we are probably seeing a level of the “law of large numbers”. Further, and while the idea that we might not need story points might be correct, I suspect we would not have got to this result without going through the process we did. My feeling is that the reason this is true is that one real value of story point estimating to the team is to teach us how to break down stories into small enough chunks that we can complete them in a Iteration (Sprint). As we get better at making and meeting commitments, we realize that “small” should be even smaller than we thought, and again that thinking is helped through the estimation process.
An effect of having small stories is that card counts and story point velocity end up measuring the same thing. Now, if we could get to the “small story” point without doing story points, we might be able to use the card counts directly to report on progress. I suppose this is an experiment we need to run.