Metrics in Kanban – Part 1

As with most things in Kanban, we must visualize the metrics in order to know what’s happening. Metrics show how we have improved or where improvement is needed.

It is best to include the team in the process of deciding what metrics to use.

Some of the metrics that can be captured from Kanban are as follows:

Cycle time

What is it?

Cycle time refers to the time a work item takes to go through a part of the process. Usually, it is the time between when the work item is picked up for work and when it is done.

What can we learn from it?

It shows how fast work moves through the process and where it slows down.

Lead Time

What is it?

Lead time shows the time a work item takes to go through the whole process instead of just a part of it, like cycle time.

What can we learn from it?

Since lead time shows the entire process, it highlights many areas that are bottlenecks and need improvement. E.g., if it takes a year to design and 1 week to develop, reducing development to 3 days will not help until the design process is optimized. Also, shortening the cycle time can sometimes be a bad thing, as if the development time is reduced, it affects testing, and the expectation of having next-to-do work items, ready increases, which means requirement gathering, design, and solutioning must speed up and be ready for the development team.

Questions:

• Have cycle time and lead time gone up? What are the contributing factors?
• What should be the ideal cycle time?
• If cycle time of the part of the process is only 30% of the lead time, where the rest of the time spent?
• Is the time measured normal for a work item of this size and type?
• Should we track the cycle time of other parts of the process?

Lead time and cycle time is used for prioritization too. Also it is interesting exercise to break lead time and see where actual time was spent? How much time was spent waiting or being blocked? How much was rework? Analyzing that can help to find bottlenecks and make process improvements.

Note that individual cycle time and lead time is not so interesting. The trend over time is more interesting since it is a variable metric and we need data over a long time to see the trend.

Lead time and cycle time can be visualized using a scatter diagram with a work item identifier on the x-axis and the number of days on the y-axis.

Throughput

What is it?

A number of work items completed per time unit.

What can we learn from it?

Throughput tells us whether your efforts to improve the flow are working or not. For E.g. Are you improving delivery frequency and thereby shortening the feedback loop by reducing WIP or slicing stories or by investing in automation.

Throughput can be visualized as a scatter diagram with time unit on the x-axis and throughput on the y-axis.

Note that throughput can be a balancing metric for lead time improvements. If you are reducing lead time throughput goes down, if you reduce the WIP too much.

Blocked work items

What is it?

A number of work items that hinder the flow of work and their effect.

What can we learn from it?

How blocked work items are effecting lead time? How fast we are unblocking the issue? Are blockers increasing?

Since issues and blocked items are the things that are hindering the workflow, it’s important to be aware of them and clear those as soon as possible. Also it is important to look at the process improvement to reduce the number of blockers. Mostly blockers happen when we are forced to wait for others either to get their input or give our work to them. Conversations about how to resolve those problems like these are easier and more to the point if we are backed by data.

A number of blockers and defects can be visualized using a scatter diagram or a stacked column chart showing the trend, with time unit at x-axis and number of bugs/ defects or number of blockers at the y-axis.

Part 2 coming soon.

Go back to the Essentials page

Reference: Kanban In Action by Marcus Hammarberg and Joakim Sunden