As a project manager one of my current tasks is risk management, and I have to focus especially on schedule risk. Therefore I was looking for something that can show me quickly the current status of the project and this kind of risk. I started to use a portfolio Kanban board that covers the whole project, but my board is different from Pawel Brodzinski’s resource oriented board:
I’m using an old idea from Chris Matts - a work item is put into a column that tells us when the it is supposed to be ready -, and the idea of organisational level Kanban boards. If this is week 13, team yellow is doing fine, whereas teams red and blue are in trouble.
Besides the nice overview the board points to the most problematic parts of the project regarding schedule risk: the oldest item (team red’s work item in week 11) and the team with the highest number of delayed work items (team blue).
The completeness attribute tells me something else: a work item with a 3/5 completeness (number of done items / number of planned items) won’t be finished this week if today is Friday, however a 2/3 might be finished. Moreover, team red has started to work on something that is supposed to be ready by week 15, but they haven’t finished their week 12 commitment. Team yellow might be delayed later: so far - in 3 weeks - they finished 6 of their work items (1/1+2/2 from done, 2/3 from week 13) and they committed to deliver 8 by week 14. That is very unlikely to happen.
These numbers are not accurate enough but they are a good basis of a discussion. Looking at the assigned team’s lead time and throughput can tell us more about the situation (more about this in a later post). Information about the completeness should automatically come from the assigned team’s board:
It is still early to say much more about this approach. I have two urgent things to solve. First, I cannot see dependencies between teams which would be important. Second, somehow I need to gather more data from teams and show it to them on the portfolio level, such as lead time, takt time and throughput, which would provide a good basis for predictions.
comments powered by Disqus