Use KANBAN: CFD (Cumulative Flow Diagram) and WIP Area Charts to Measure and Manage Flow

Use CFDs and WIP Area Charts to Measure and Manage Flow

The kanban board is an important tool for managing flow specifically because it visualizes the source of the problem, and lets you limit work in progress where it will be most effective.

When you look for the work that piles up and add a WIP "In short limiting work-in-progress with Kanban encourages higher quality and greater performance.
The act of limiting WIP helps you optimize work capacity by allowing you to pull new work only if capacity is available".

limit to smooth it out, you’re taking steps to increase the flow. The WIP limit works because you’re helping the team focus their effort on the specific part of the project that’s blocking work from flowing.



In fact, that’s all that a WIP limit does it changes which work items the team is currently working on, and causes them to work on the ones that will even out the flow and clear up unevenness and workflow problems before they start to form.

How do you know that you’re actually increasing flow when you add WIP limits?

Once again, we can go back to lean thinking, which tells us that we should take measurements and an effective tool for measuring flow is a cumulative flow diagram, or CFD.

A CFD is similar to a WIP area chart, with one important difference: instead of flowing off of the diagram, the work items accumulate in the final stripe or bar.And while the WIP area charts have stripes or bars that correspond to states in a value stream map, the CFDs (and WIP area charts) have stripes that correspond to the columns on a kanban board.
The CFDs have additional lines on them that show the average arrival rate (the number of work items added to the workflow every day) and average inventory (the total number of work items in the workflow).

The CFD can also show the average lead time (or the amount of time each work item stays in the system). Not all CFDs have these lines; however, they are very helpful in understanding the flow of work items through the system.

The key to managing flow with a CFD is to look for patterns that indicate a problem. The kanban board can show you where the unevenness, loops, and other workflow problems are today, and helps you manage your flow on a day-to-day basis by adding WIP limits.
The CFD lets you look at the way your entire process is performing overtime, so you can take steps to find and fix the root cause of any long-term problems.

Kanban teams use cumulative flow diagrams with stripes that correspond tothe columns on the kanban board. A cumulative flow diagram (CFD) is like a WIP area chart, except that instead of flowing off of the end of the chart, work items accumulate so all of the stripes or bars continue to go up over time

How to build a cumulative flow diagram and use it to calculate the average lead time

To build a CFD, start with a WIP area chart. But instead of gathering data from a value stream map, you’ll gather the data from the number of work items in each column on the kanban board.

Next, you’ll need two additional pieces of data that you’ll add to the diagram every day: the arrival rate and the inventory.



To find the arrival rate for each day, count the number of work items that were added to the first column. To find the inventory for each day, count the total number of work items in every column. Add a dot to theCFD each day for the arrival rate and inventory, and connect them to create two linecharts overlaid on top of the WIP area chart.

Most teams that use CFDs don’t draw them incrementally on a wall; they use Excel or another spreadsheet program that supports charting. One reason aside from ease of managing data is that the spreadsheet can automatically add a linear trendline to the arrival rate and inventory line charts.

These trendlines are very useful, because they can tell you whether or not your system is stable.If they’re flat and horizontal,the system is stable. If one of them is tilted, then that value is changing over time. You’ll need to add WIP limits to stabilize your system, and you’ll be able to tell that the system is stable once those lines flatten out.

This is an example of a CFD that also shows the arrival rate and inventory. The total size of the work in progress at any given time can be found by measuring the difference between the top of the chart and the top of the “done” stripe. The horizontal solid black line shows the lead time for a specific work item in the system. We can’t cal‐ culate the average lead time yet because the system is not stable the long-term inven‐ tory and arrival rates have tilted trendlines, which means they aren’t constant.

If you look at the hand-written notes in Figure, you’ll see that we assigned letters to these values: we used L for the average long-term inventory, λ (that’s a Greek lambda) for the average arrival rate (or the number of work items added every day), and W for the average lead time (or the average amount of time a user is waiting for the team to finish a work item request).

Take a look at the inventory line in the CFD that’s the thick dotted line near the top. It’s trending downward, which tells us that the total inventory is going down over time. It shows that many work items flowed out of the system and were never replaced. But if you look on the bottom, the arrival rate is increasing.

If we keep tracking this project, what will happen? Will the inventory fill up again?

Will there be another release that empties work items out of the system? If more fea‐ tures arrive than leave, then over the long term the inventory will trend upward—and the team will feel it. They’ll slowly have more and more work to do, and will feel like they have less time to do it. It’s that “mired in muck” feeling that happens when the system isn’t flowing.

Luckily, we know how to fix that problem: add WIP limits. The team can use experi‐ mentation and feedback loops to find WIP limit values that work for their system, and if they get them right eventually the rate that work items arrive will balance out with the rate that the team can finish them. The long-term inventory trend will be flat, and so will the long-term arrival rate. And once that happens, the system is stable.



When a system is stable, there’s a simple relationship between these values called Little’s Law—a theorem,you don’t need to know a lot of math to use it:

L = W × λ

In English, this means that if you have a stable workflow, the average inventory is always equal to the average arrival rate multiplied by the average lead time. That’s a mathematical law: it’s been proven and if a system is stable, it is always true. And the reverse is true, too:

W = L ÷ λ

If you know the average inventory and the average arrival rate, you can calculate the average lead time. In fact, calculating the average inventory and arrival rate is pretty straightforward: just write down the total number of work items on your kanban board every day, and write down the number that were added to the first column that day.

We show those on our CFDs using thin dashed or dotted lines. If your system is stable, then after some time you can find the average daily inventory and the daily arrival rate, which we show as thick straight lines. Divide the average inventory by the average arrival rate and you get the lead time.

Stop and think about that for a minute. Lead time is one of the best ways that you have to measure your users’ frustration level: deliver quickly and your users are pleased; take a long time to deliver, and your users grow increasingly frustrated. And long lead times are good indicators of quality problems.



Your lead time is determined entirely by the rate that work items arrive into your system, and the rate that they flow through i and WIP limits give you control over the flow rate.

Source Content = Learning Agile: Understanding Scrum, XP, Lean, and Kanban