Kanban Metrics you won’t hate – Part 2: Control Chart

This is the second article about Kanban Metrics. The first article was about the Cumulative Flow Diagram, used to measure the throughput time for items in your backlog. The time it takes to go from “In Progress” to “Done” and then projected over a period of time.

But what are the ways you can use a Control Chart?

  1. Analyze your team’s performance
  2. Measure the effect of a process change
  3. Provide stakeholders information on your team’s performance
  4. Creating SLA’s for your team

Normally you should have a constant cycle time for your team. If you have this effects can be seen directly. You can use your time during the kaizen/retrospective to analyze the root of the inconsistency. Let’s look at 2 situations:

  1. You have an increasing cycle time. During the kaizen/retrospective we could see that we had changed our D.o.D. which causes the increasing cycle time.
  2. You have an erratic cycle time. Normally you should have consistent cycle times for work items with the same story points or within the same classes of service. Check the report for consistency. If there are differences on small and large story points, examine them during the kaizen/retrospective and improve future estimations or classifications.

What is common cause variation?
These are the items that fall between our limits and are done in a normal time.

What is special cause variation?
All items that fall outside the limits, are special cause variations. These are the items you first have to look at and try to solve them.

So what does this metrics look like. Find an example here:

Looking a bit deeper into it, we have the blue line, which is the average line of our items. We have an UCL (Upper Control Line) and LCL (Lower Control Line) defining the constraints of common cause variations. These 2 lines are calculated at a distance of 3Sigma (Normal distribution) from mean. All in between are normal. No necessary action is needed. These items are done within a normal time. But the items above the UCL are to be avoided. The ones below the LCL can be reproduced.

When looking at special cause variation, also have a look at the ones that are significantly below the lower control limit and ask yourself: “What were the conditions that made this one go faster? Can we create those conditions for other items?”

For more information on Kanban metrics

  1. Cumulative Flow Diagram
  2. Control Chart

And if you want to know how to use them on your personal Kanban, see the book "Personal Kanban in a Nutshell"