As a product manager, improving your team performance is always at the forefront of your mind. Without knowing how much your team delivers results, it’s impossible to evaluate your performance and give accurate delivery estimates to your customers. Kanban throughput measures your team’s capacity to deliver and enables you to make data-driven decisions about your future performance.
What is throughput?
Throughput in Kanban refers to the amount of work delivered over a certain period. No matter how many work items your team has in progress, this metric ignores anything unfinished.
Consider a workflow where throughput is calculated on a weekly basis. In five weeks, this team delivers 5, 7, 3, 5, and 8 tasks respectively. Therefore, average throughput is calculated as (5 + 7 + 3 + 5 + 8)/5 = 5.6 or 6 tasks per week if we round up to the nearest whole number.
Tracking throughput
Kanban throughput data is generally displayed using a bar chart or a histogram.
The throughput run chart shows the actual throughput on a daily/weekly/monthly basis. It is a way to display the total throughput of the team and present throughput data to stakeholders. It is very useful when the chart displays the average throughput for the time period in order to compare actual throughput to average throughput on a day to day basis.
Throughput histograms visualize the frequency distribution of your throughput data. This allows you to see the skew of the data, the width of its spread, the median/mean/mode values, and how all of these change over time. The vertical percentile lines show the probability of repeating a certain throughput based on the performance of your team in the past.
The width of the data spread indicates the predictability of your process. A tight cluster shows that your team is consistently delivering work at around the same pace. Average throughput indicates your team productivity. You can see if your productivity is increasing or decreasing by observing how trends develop over time.
Evaluating team performance with throughput
The throughput histogram allows you to make quick assessments of team performance. The throughput for the team can be easily evaluated by the median number of tasks that were completed per unit of time (day, week, month). Over time, you can identify trends in your team’s task delivery – ideally, the throughput should increase or stay at similar levels.
Decreases in the number of tasks delivered could indicate your team is encountering problems. If the median number of tasks that were completed in a day for the quarter is 10 and the team’s median throughput for the last month of the quarter is just 7, this means that team productivity has decreased.
Improve your team throughput
Set WIP limits
The three flow metrics – throughput, cycle time, and WIP – depend on each other. The equation known as Little’s Law describes their relationship:
WIP = Throughput * Cycle Time
By changing one, you will almost certainly affect one or both of the others. Delivering work faster means decreasing cycle times by limiting WIP. Little’s Law only applies to stable systems – systems where all the Little’s Law assumptions are valid. If this is the case, Little’s Law will work without making any drastic changes to the team such as hiring or firing staff.
Limiting WIP also reduces multitasking. While multitasking allows you to work on more things at once, it actually reduces productivity. Switching between tasks carries at least a 10% penalty per switch – the higher the number of tasks in play at once, the higher the overall switching cost.
The ideal WIP limit for a team prevents multitasking but leaves no team member with nothing to do. A good starting point is setting your WIP limit slightly lower than the number of people in your Kanban team.
Encourage your team members to collaborate and “swarm” on outstanding tasks. Throughput only measures delivered work, not work in progress – getting team members to resolve blockers and work together to complete tasks keeps throughput consistently high.
Look for the weakest link
The throughput of a multi-stage process is limited by the throughput of the slowest step. This is known as the theory of constraints. The first priority when looking to improve overall team throughput is to identify and improve or eliminate this constraint.
“Constraints” can be anything impeding performance – common examples include shortages of staff or equipment, company procedures, and informal workplace norms. Some things to look out for when identifying constraints are:
- Large WIP increases in any process state
- Idle tasks with no one working on them
- Cycle time scatterplot showing outlier tasks
- Frequent expedite/emergency tasks if using Classes of Service
- Team feedback on constraints during Kanban meetings
Predict future performance
Tracking throughput provides a huge advantage to product managers – an accurate way to estimate future performance. Accurate delivery predictions allow for long-term planning, build trust and improve client relationships.
For long-term and larger-scale Kanban estimates, we recommend using Monte Carlo simulation to take into account uncertainty within your process.
The key to keeping Kanban throughput high – and customers happy – is to focus on getting work done rather than doing more work. Keep your team focused on a small number of tasks at a time by using WIP limits and measure their true capacity to deliver.
Has reducing your WIP limit impacted your team productivity? How have you increased your team Throughput over time? Have your delivery predictions become more accurate? Tell us about your experiences in the comments!
