Understanding Your Data: Kanban Analytics
Kanban processes improve by continuously making incremental changes. Effective changes are built upon intelligible and unambiguous data. A picture is worth a thousand words – how many data points does a chart convey?
In this article we’ll take a look at how you can use Kanban analytics to understand your data, spot issues and make your process more efficient.
Cumulative flow diagram
The cumulative flow diagram tracks the total amount of tasks in progress. Each colored band shows a different process state. The values accumulate over time, giving it its “cumulative” name. The CFD shows you how your total work in progress has changed over time and how smoothly tasks are making their way through your process.
There are several pieces of information that you can immediately pull from the CFD. Firstly, the approximate average cycle time – the horizontal difference between the upper and lower bound of a band shows the approximate average cycle time for that stage of your process. Measuring from where a task is pulled into the “Done” band gives the approximate average cycle time for the entire process. If this value does not closely track the average cycle time on your cycle time scatterplot, you are accumulating flow debt.
Next, you can use the cumulative flow diagram to calculate your average throughput over a given time range. Take two points on the “Done” band of the CFD – the gradient between these two points is the average throughput for that time period.
Keep a lookout for a sharp increase in gradient in one or more of the bands – this indicates WIP is expanding and a bottleneck is forming. Another common pattern to watch out for is flat bands, meaning no tasks are being delivered and can signify a blockage. The key to using cumulative flow diagrams for Kanban analytics is to learn to recognize the most common CFD patterns.
Cycle time scatterplot
Cycle time measures the time between the beginning and completing work on a given task – it is one of the most important metrics for Kanban analytics. Lower cycle times mean more work is being delivered and thus higher customer satisfaction. The cycle time scatterplot collects all cycle time data over a period of time – each dot in the graph represents a task. The vertical position of the dot gives the cycle time of the task, while its horizontal position shows the day that the task was delivered.
The horizontal percentile lines on the scatterplot show the percentage of tasks completed within certain cycle time. In the image above, we can see that 70% of tasks are completed within 18 days. This information can be used to forecast likely cycle time on future assignments and define service level agreements. You can also spot problem areas by recognizing common scatterplot patterns.
Cycle time histogram
Histograms are a favorite among project managers for their clear and simple presentation of data. The height of the bars shows the cycle time frequency distribution of your dataset, with frequency on the vertical axis and cycle time on the horizontal axis.
Cycle time histograms make it easy to see average cycle time, highest and lowest cycle times along with the skew of the data. A wide spread of data indicates your cycle time varies significantly and your process is inconsistent. Cycle time results in a tight cluster show a consistent, predictable process.
Limiting work in progress is a key component of the Kanban method. When the WIP limit for a process state is reached, no new tasks are allowed to be pulled in before an outstanding task has left the state. This keeps your team from being overburdened and keeps tasks from being neglected. For Kanban analytics, both the amount and age of work items in progress are important data points.
The aging chart uses the same visual format as your Kanban board, each column represents a stage in your process. As for the cycle time scatterplot, each dot indicates an item of work in progress. The left vertical axis shows how many days a task has spent at that stage. On the right, percentile lines show the percentage of tasks completed within certain cycle time.
Look out for aging WIP in your process. Tasks with an age that is higher than average are a warning sign of potential problems. Whether they are due to accumulating flow debt, a team member having difficulties, or many other reasons – aging WIP is a signal that you should get to the bottom of it. Among other reasons, it’s time to start asking questions.
Throughput Run chart
Another crucial metric in Kanban analytics is throughput. This refers to the number of tasks delivered on a certain day, week, or month. This metric only takes into account completed tasks, no matter how many are currently being worked on.
The throughput run chart is a great way to display the total throughput of the team and present throughput data to stakeholders. We recommend comparing throughput on a regular basis to see how trends build over time.
You can examine your throughput frequency distribution using a throughput histogram. Throughput is shown on the horizontal axis while the vertical axis shows the number of days that this throughput was achieved. It lets you measure your team’s productivity at a glance, shown by the median value.
You can also use Kanban analytics to understand the spread of your data. Throughput values with little spread show a team consistently delivering tasks at roughly the same rate, while a wide spread shows high variability.
The vertical percentile lines on the histogram show the percentage of days with a certain throughput. In the example above, 85% of days had a throughput of 7 items or fewer per day. This past performance data is used as a base for more advanced estimation approaches like Monte Carlo analysis.
Monte Carlo simulations
Delivering a project successfully relies on hundreds of uncertain factors. It’s no wonder that presenting realistic forecasts to stakeholders is such a difficult task. Monte Carlo simulations don’t try to eliminate uncertainty, they harness it and use it to make predictions.
Uncertain factors have a range of possible values instead of a fixed value. The Monte Carlo method runs large numbers of simulations, taking a random value within the possible range for each uncertain factor. Hundreds of thousands of simulations build a probability distribution of what could happen, and how likely it is to happen.
This technique produces highly realistic results for processes with uncertain inputs – it’s used in risk management, financial analysis, and biomedical modeling along with many other fields. Monte Carlo simulations for Kanban analytics use throughput as their driving metric. Your past performance data is used to give the most realistic estimates for your future performance.
Monte Carlo: Delivery Date
“When are you going to be ready?” is the question at the forefront of every client’s mind. Kanban analytics uses Monte Carlo simulations to get the most accurate range of probable delivery dates. The Monte Carlo: Delivery Date chart is ideal for presenting this information to your stakeholders. The peak shows the most likely delivery date, with the tails showing the optimistic and conservative outcomes.
Monte Carlo: Number of Tasks
You can also use Monte Carlo simulations to predict the number of tasks that can be completed within a certain time frame. This is a great way to visualize how likely you are to be ready for a release or how much work you can deliver in a month.
Whatever decisions you make within your organization, they should be data-driven. Tools such as Nave import historical data from your board to build Kanban charts. Over time, you will develop a deep understanding of your process and performance. Stop relying on guesswork – Kanban analytics are your trustworthy advisor to predictable success.
Which charts do you use most frequently to analyze your process? Which is the best to spot problems? How do you present data to stakeholders? Tell us about your experience in the comments!
Meet the Author
Sonya Siderova is a passionate product manager and a driving force behind Nave, a Kanban analytics suite that helps teams improve their delivery speed through data-driven decision making. When she's not catering to her two little ones, you might find Sonya absorbed in a good heavyweight boxing match or behind a screen crafting a new blog post.