A key insight in the Kanban Method is how to measure how much effort is put into a given process, so that it can be improved more efficiently. Most often this will be some kind of pull-based system where each member picks up tasks according to how much time they have available or how many items are currently on their plate.
On some level, all work in this context is assumed to take some amount of time and resources, which for obvious reasons cannot be infinite. As such any assignment of tasks to individuals should also estimate how long that task will take them using the best information available, normally based on previous experience or perhaps an educated guess if no experience exists. In cases where past data does not exist, one way out may be to use the average lead time experienced by the team previously to come up with estimates.
i that knowledge how much time resources are available for this particular project also plays an important role. If you only have one day to complete a task, no matter how many hours it may take you, you simply cannot do it. It’s quite possible that the number of days allotted is greater than or equal to the number of tasks needed done; in fact there is likely at least one task where zero days are allocated because nothing can be done today. This brings us into how long things take on average , often measured as lead time, which I’ll define shortly.
Once we have our tasks along with how long they are expected to take, how do we determine how many people are needed to complete the work? If I have 10 tasks that are expected to take 2 days each, then how long will it take for me to complete them with no one else doing anything?
10 x 2 = 20
This is how much time it takes my team, but what if there were more than just me working on this project? How could I find out how long it would take in total? That’s when you need the concept of Little’s Law .
Little’s Law is a formula used to describe how busy a resource has been over time. It was developed by Dr. John Little in 1963 while he worked at MIT Lincoln Laboratory .
The formula for Little’s Law is:
L = W/P
Work in Process (W)
Average Lead Time (AL)
Number of Customers (N)
The two most important factors to remember for how to estimate with Little’s Law are the average lead time and how many customers there are. This law can help you find the total amount of work your team creates, how long new tasks take on average, how many new tasks you can complete every year, or how long it takes for teams to complete a project. That sounds pretty miraculous doesn’t it? Let me explain how I use this formula personally. I am currently working as an agile coach helping executives learn how to implement Lean / Kanban principles into their work. One common question I get, from all different types of teams using kanban systems, is how to estimate how much work they can complete in a fixed time period. This lead time helps me to calculate how many man-hours the team has available every week.
I will start by explaining how to use Little’s Law to get your average lead time and then how you might use that number for how to estimate in kanban.
Also note: these numbers are just assumptions for how it would play out if there was a constant flow of tasks at a good pace without interruptions or backups. In reality these things do come up, but this should help you benchmark what kind of workload your team could be faced with.
To calculate how many hours your team can complete in a week you need to know how many tasks they pull through the system. Likewise how fast it takes them to process the task (lead time).
If there are 5 people on your dev team with an average lead time of 10 days, pulling 3 items per day, then they could complete 15 tasks total during that week (assuming no interruptions or backlogs).
That is how to estimate – just by knowing how many tasks people want and how long it takes them.
This will allow you to plan ahead for what kind of workload is coming next week, month or even quarter. It also shows how changing one item will affect the overall throughput of the team which is how to scale your Kanban system.
But how do we estimate how long it will take?
The number one rule of estimation is: Never guess!
Estimation techniques vary by what you’re trying to estimate and how accurate the end result needs to be (frequent or infrequent, big or small). The most common and simplest way is estimating in hours (or days). This is simple because we all know how many hours we work on something and hour estimation can go from as precise as +/- 0.5 hrs up to +/- 3 hrs depending on how much risk there is associated with the task (see also: agile estimation techniques) . Most teams using Scrum use story points instead of time, which is a great metric as long as it isn’t treated as if it was time.