I came back to the NCAA team at an interesting point in our development cycle. It’s the tail end of Production and we are only 6 weeks away from declaring the product Alpha. For Electronic Arts, Alpha is Feature Complete: all new features that have been implemented and a certain level of stability exists with the game.
Throughout Production we are writing requirements that we call Tests for Completion (TFCs). It is the expectation that the developers who are creating the feature, whether they are artists, designer, or programmer, make sure that they feature they are working on fulfills these TFCs. At the end of our sprints, all the TFCs we commit to at the beginning of the sprint need to pass a review by our QA department. Invariably not all of them get complete and if any of them miss, we try and get them complete as soon as possible right after the end of the sprint. However, sometimes there’s a few that slip even more beyond that sprint and can start hitting up against Alpha. In order for our QA department to declare the game Alpha, all TFCs must be 100% complete.
In order to give us some buffer room, we designate two sprints as “contingency”. Those two sprints are not included in the scope of the project. The two sprints are: one right before the Holiday break (late December) and the other one right before Alpha. We choose late December because there are a lot of developers that like to take extra time off around the break. Overall, these two sprints give us some time in order to iterate on features and hit Alpha on time.
NCAA Football is divided into four sub-teams, we call Pods. Coming onto the team three weeks ago, I’m pretty much taking up the charge to Alpha. My first thought was, “Will each Pod make it to Alpha?” The team had what they called a Master Burndown, which combined the points and velocity of all the Pods. This is great, but all it only takes one feature to hold up Alpha. So, I used some of the techniques I talk about in my Hansoft Basics presentation, to calculate a per-Pod velocity based on data in Hansoft. That ended up giving me data that we were probably going to be booked around a week and a half into the final contingency sprint. Good, but not great.
From this point going to Alpha, we have to keep a very close eye on things that are changing on the backlog. Figure out what is being added and what is being removed. It’s delicate balancing act of wanting to get as many features into the product as possible, quality of the product, and the number of hours we are going to have to crunch during Alpha. I’m hoping we come out completely balanced.