Some years ago (circa 2007) Yuval Yeret found a secret weapon for fighting Zombie Scrum: Kanban and Flow.
Read the complete article here: https://www.scrum.org/resources/blog/kanbanflow-scrum-masters-not-so-secret-weapon
five conditions under which Brendan Wovchko has found Kanban to be a better fit than Scrum:
- Low Tolerance for Change
- Obvious at All Levels
- Fluid Priorities
- Small Teams
- Complex Collaboration
Read the complete blog-post here: https://www.mountaingoatsoftware.com/blog/when-kanban-is-the-better-choice
In this article you can read how Christian Hofstetter has applied flow metrics to run a retrospective with the goal to gain more insight and deciding what to do to improve.
You can find the article here: https://medium.com/the-liberators/how-to-do-a-retrospective-using-flow-metrics-612bf48bdeba
The Professional Scrum with Kanban (PSK) course has now been out for more than 6 months at Scrum.org. As one of the first few trainers who wanted to teach this course when it came out, I find that it is a great way to combine the Scrum framework with Kanban as a strategy to deliver value to your customer.
Out of the many topics that we talk about in this class, I’ve found that the use of throughput instead of velocity/capacity to be a positive change. I’ve taught the regular Professional Scrum Master (PSM) course for about 6 years now and when I get to the Sprint Planning slides, we usually extend the conversation around velocity and capacity. I pull up my complementary slide deck around relative estimation, poker planning, charts to track velocity and we spend an additional hour on this topic. I answer questions around the meaning of story points, how they should be understood, tracked and used in multi-team Agile project.
In the PSK class, this conversation is completely different. When I get to the Sprint Planning slides, I point out that the throughput history is used as an input to the Sprint Planning. With a few examples, I show how easy it can be to get from your electronic Agile tracking system (Ex: Jira) or on your physical Kanban board.
I then get a new set of questions from students which I find a lot more interesting. The conversation goes quickly around the variation in the size of the Product Backlog Items (PBIs) that are taken by the team at Sprint Planning. I can also tie it back to Little’s Law where limiting work in progress will increase throughput, thus helping students see throughput linked to limiting work in progress. There are very few questions around understanding throughput. Students find it is a metric that makes sense to the business compared to story points.
While our industry has talked about poker planning and story points since almost the beginning of agility in 2001, I think it is more than time that the conversation at Sprint Planning shifts to historical throughput instead of using velocity. Maybe one day in the software engineering museum, we can see a deck of poker planning cards next to a set of punch cards.
Looking at differences between Scrum & Kanban can help us see which will work better for us.
- Scrum requires planning the sprint. You can plan in Kanban but it’s normally isn’t done.
- Scrum requires cross-functional teams. Kanban doesn’t. While making it more flexible if also may miss the opportunity for team structure improvement.
- Scrum requires starting with its roles, practices, events & artifacts. Kanban allows you to start where you are & provides a transition model for improvement.
- Scrum improves by removing impediments. Kanban improves by focusing on shortening cycle time.
Teams that don’t like to be told what to do may resist Scrum. Kanban requires more discipline from the team than Scrum.
Factors to consider when deciding which to use:
- culture– including resistance to being told what to do &attachment to roles
- nature of work being done
- ability to create cross-functional teams
Note that executives can better relate to Kanban’s focus on flow. Combined with its insistence on visibility, executives can better understand the importance of managing workload.
In few cases is one clearly superior to the other. Taking a blend of the two often makes sense. Doing this is not difficult.