Kanban Flow Metrics – Forecasting Replaces Estimations

One of the biggest questions to answer in project management is – When will it be done? And surely, if we could say that with confidence, project planning, setting budgets, and presenting ideas to the stakeholders would look much different.

While we do not have a crystal ball (or at least one that actually works), we do have something that can put you close – Kanban flow metrics. Based on probabilistic forecasts and your team’s data, Kanban flow metrics can elevate your understanding of the processes and help forecast the possible results more easily.

Let’s explore what the flow metrics stand for and how they can help you become better at Kanban planning and project management.

What are the Flow Metrics?

The concept of Flow metrics was first introduced in 2018 by Dr. Mik Kersten and then further discussed in his book Project to Product: How to Survive and Thrive in the Age of Digital Disruption with the Flow Framework. Which proposed looking at your value creation process as a flow, hence the name – Flow Framework.

The flow framework itself is aimed at optimizing the product value stream. It is done by eliminating roadblocks and allowing your process to flow freely and without unnecessary interruptions.

To achieve this, Dr. M. Kersten proposes dividing all of the work you do into 4 flow item types and tracking 5 specific flow metrics. By doing so, the team can:

1. Monitor what type of work dominates their workflow,
2. Track the metrics for each group individually as well as together,
3. Identify where the flow is interrupted.

Using all of the above information allows us to eliminate issues and improve the process itself.

Flow metrics

The Flow framework specifies 5 flow metrics for the teams to track. Recently, there have been some teams that started advocating for a 6th metric to be added – accuracy. Which can complement the first 5 metrics defined in the original approach.

Here are the 5 flow metrics in more detail:

Tracking these metrics allows us to review the value stream and ensure there are little to no interruptions to the flow. For teams that track performance for each flow item type, this gives additional information on how to improve process efficiency in various aspects of their business.

Flow metrics in Kanban?

If you are familiar with Kanban, the flow metrics and the overall concept of tracking the flow may sound quite familiar. In fact, it seems a lot of inspiration for flow metrics has been drawn from this Agile approach.

However, when we talk about Kanban flow metrics, the terminology differs a little from the ones discussed above. Even though the general idea is quite similar.

One of the key concepts in Kanban is process visualization. Usually done on Kanban boards, it allows teams to see how their items move through the process from the initial idea or request to being completed. In other words, it helps teams monitor their flow or process of value delivery.

Kanban boards differ in their complexity depending on the team’s processes and help identify immediate signs of delays or bottlenecks. However, that is not all there is to the approach. To go beyond the surface level, Kanban teams track various metrics that help them understand the performance better and make decisions on how the flow can be improved.

Sounds familiar? It is.

There are 4 core Kanban flow metrics that teams use to evaluate their performance.

Teams using Kanban boards, usually tend to separate their item statuses into at least 3 larger categories – Planning, In Progress, and Done. These same categories are then usually utilized to calculate the Kanban flow metrics. For example, the cycle time of an item will often be equal to the time that item spent in the ‘In Progress’ section of the board.

While the team can choose to set different start and end points for their calculations, this approach is often still the most popular.

Kanban flow metrics

Now that we know about the 4 Kanban flow metrics, let’s dive in a little deeper and understand how they work.

WIP (Work In Progress) marks how many items are currently ‘In Progress’, and should not be confused with WIP limits.

WIP limits define the maximum number of items that should be in progress at any point. WIP gives us the actual current number on the board. Ideally, the WIP number should be lower than that of the WIP limit. However, this is often not the case, with teams taking on more than they are able to handle.

Cycle time accounts for all the time between when the item is started (pulled into progress) and done.

This means it includes the wait times, reviews, weekends, and everything else that happens in between as well as doing the actual work. This metric represents the duration of your process in real-time, not business days or touch time.

Work item age helps us understand how long the task has already been worked on.

It can help identify outliers, as well as aid in forecasts about when that task may be finished. By knowing the average cycle time and work item age, we can estimate how much longer typically we would have to wait for that item to be delivered.

It can also be used to identify items that have been blocked for a long time or forgotten.

Throughput helps us answer the question of how many items will the team finish during a certain period of time.

It can aid in estimating, how fast new items will be coming into different process steps. For example, if your throughput is 15 items, it could be argued, that your testing team should be able to handle testing 15 items during that specific period of time.

Kanban flow metrics calculation

As mentioned above, start and finish dates are used to calculate all of the Kanban flow metrics.

For those using Kanban board software, such metrics are usually calculated automatically. And those relying on physical Kanban boards will have to do such calculations by hand. Either way, it is important to know how the calculations are done, to get a better grasp on their meaning and value.

To calculate the Kanban flow metrics, each item must have a timestamp of when it was started and when it was completed. Based on this information and the current progress of the items in your process, you will be able to calculate all of the metrics discussed above.

• WIP = all items that have a ‘Started’ timestamp, but not a ‘Completed’ timestamp
• Cycle time = ‘Completed’ timestamp – ‘Started’ timestamp +1 (If an item was started on Monday, and finished on Tuesday, its cycle time will be 2 days.)
• Work item age = ‘Current’ timestamp – ‘Started’ timestamp +1 (If it is currently Wednesday and the item was started on Monday, the work item age is calculated as 3 days)
• Throughput = number of items that got the ‘Completed’ timestamp on each day (calculated for each day separately, then pooled if needed)

Probabilistic approach

As you can see, the Kanban flow metric calculations above are based on single-day or single-item values.  But what about the scenario, where you need to know data for multiple items? This is especially true for the cycle time and throughput metrics. Where you may find yourself asking questions like:

• How long does it take for a team to deliver an item?
• How many items can be delivered during a certain period of time?

You may be tempted to say – okay let’s just calculate the average. However, doing so will include everything in your estimates. Even the outliers. This means that if one item took considerably longer than the rest, the average will be affected by it greatly. Giving you a false understanding of the team’s performance and capabilities.

To avoid such scenarios, Kanban flow metrics apply the probabilistic approach. Instead of simply relying on the averages, teams are basing their decision on the probability of how likely something is to happen.

Sounds complicated? Well, it did to me at first.

And the best way to get around this is by looking at examples.

Example

In the one below, we have data for 6 tasks that the team has finished and their cycle times. I know, 6 tasks seem like very little data, but this is done for the simplicity of the explanation.

In this case, if we were to calculate the average cycle time, it would be (3+4+3+4+20+4)/6 = 6,3 days. 

Now that we look at 5 of those tasks, we can immediately see that their cycle time is considerably shorter and the average is skewed by the single outlier. 

When applying the probabilistic method, we look at certain percentiles of completed items instead of adding up all of them. For example:

• If we take the 50th percentile, we select 50% of our data pool (in this case 3 tasks) with the shortest cycle times. So, by looking at the table above, we can say that there is a 50% chance that any item in our current backlog (once started) will be completed in (3+3+4)/3 = 3,3 days.
• If we take the 80th percentile, we select 80% of our data pool (in this case 5 tasks) with the shortest cycle times. In our example, we could then say, there is an 80% chance that any item in the backlog (once started) will be delivered in (3+4+3+4+4)/5 = 3,6 days.

To help you visualize how such a calculation is done for a larger amount of items, let’s take a look at this scatterplot diagram from Daniel Vacanti.

We see the cycle times of all completed items during a certain period of time. Dots mark the item and the vertical axis marks the cycle time length. While the dotted lines mark various percentages of probability ranging from 50% all the way up to 95%.

As you can see, the 50th and 85th percentiles both hold quite a large number of items. And we start to see the true outlier items with the 95th percentile. While each team and their deliveries are different, there usually is a larger gap between item cycle times as the percentiles go up.

The probabilistic approach to analyzing Kanban flow metrics allows us to decide what kind of risk we want to take with our forecasts. It also gives us the opportunity to track and analyze more data and understand the flow of value just a little better.

When will it be done?

So, now that we have covered both – the flow and Kanban flow metrics, it is time to circle back and tackle the most important question – When will it be done?

While we do not have that crystal ball just yet, applying all the techniques mentioned above gets you as close as possible. By analyzing your team’s data and previous performance you can draw experience-backed estimations and forecasts for the future.

Sounds promising, but your head already hurts just by imagining how much data there is to be analyzed? Well, we have something that may be able to help you.

A ‘When will it be done’ layer for your tasks in Teamhood.

Since Teamhood already holds all the data needed for Kanban flow metric calculations, we decided to do them for you. All you have to do to get your forecasts is:

• Turn on the layer
• Select the time period
• Select your desired accuracy
• The forecasts will be displayed immediately

So all you have to do is update your task board regularly and the forecasts will be delivered for you.

Curious to try it out? Set up a free account or watch this video to understand how it works.

Visualize your Kanban process and get automated forecasts.

Easy, flexible, fast.

Get Strated