What is the relation between Little’s Law and Lead Time?

What is the relation between Little’s Law and Lead Time?

The Little`s Law is widely known as a forecasting technique in Agile world. However, it has several requirements to use it properly. Let`s explore more about Little`s Law and how to use it correctly, so it guides you well in your forecasts. 

The Little’s Law: Average Lead time = Average WIP / Average Delivery Rate.

WIP (Work in Progress) – a number of items in the system overall or in certain stages (columns on the Kanban board).  

To use Little’s Law it is important to meet several requirements:  

  • items entering and leaving the Kanban system are of the same type;
  • the lead time of items (tickets) in the process is not increasing or decreasing (the system is stable);  
  • the WIP is fluctuating consistently between a max limit and a lower but consistent value;
  • all the tickets, that entered the system, must get to its end (no bounced tickets, or tickets that get stuck inside forever).  

That is why using Little’s Law may be a bit tricky:

Even if it looks simple and obvious, you can only use it with a thin tail lead time distribution (Maturity Level 3-4 according to Kanban Maturity Model). The other prerequisites are also not typical of a lower maturity implementation. WIP limits, a stable system, and a negligible amount of aborted or abandoned tickets – are all attributes of a maturity level 3 or 4 implementations. Little´s Law lures us in with its simplicity. It looks attractive to low-maturity organizations with rudimentary Kanban flow systems, but there is danger in using it for forecasts when the prerequisite conditions are not met.  

However, if applied correctly the Little’s Law formula has proven its usefulness in understanding how flow systems (Kanban systems) behave when WIP changes. Therefore, the benefit of trimming the tail is that it enables you to use Little´s Law as a convenient forecasting technique.  

It is important to understand that Little´s Law is a function of averages.

That is why it is necessary to be able to calculate an average for each variable, with a reasonably small sample data set. We believe that the threshold for a reasonable sample is 70 to 100 data points (tickets). This will allow forecasting an average with less than 10% error.  

The Little’s Law formula communicates the abilities of the system – the more tickets/tasks you put in it – the longer your lead time will be. WIP limits encourage you to work on the tickets/tasks one by one reducing multi-tasking. The biggest benefit of WIP limits is that tickets spend less time waiting, and less waiting time results in a thinner tail. WIP limits stabilize the system and hence, stable upper and lower limits of variation, which are the necessary prerequisites for the Little´s Law equation. It is unwise to use Little´s Law if you don´t have WIP controls in place.  

How to use the Little’s Law:  

You can use Little´s Law for basic forecasting to define how long the current WIP will take until it's ready given the previous experience. For instance, if it took 20 days to complete 10 work items, using the formula you can predict how many days (on average) you will need to complete 12 WIP items. Under most, but not all, circumstances, Little´s Law behaves linearly, and hence if 10 items took 20 days, 12 items will probably take 24 days. The equation tends to be more predictable and linear when modeling increases in WIP rather than decreases. We plan to explain the very technical mathematical reasons later in other materials, such as a book on Enterprise Services Planning.  

Back to blog

Leave a comment

Please note, comments need to be approved before they are published.