A New Copy!

Today we tackle Chapter 3 of  Daniel S. Vacanti’s Actionable Agile Metrics for Predictability. Chapter 3 is titled Introduction to Little’s Law.  Little’s Law is incredibly clever and potentially life-changing if you are overly fixated on size.  Buy your copy today and read along!

We originally wrote about Little’s law in September 2014. Little’s Law brings WIP, Cycle time and Throughput (metrics discussed in chapter 2) into a relationship that can help deliver information that can be used to answer the basic questions of predictively. The basic configuration of Little’s law is stated as:Average cycle time = average work in progress / average throughput

Little’s Law is a relationship of averages.  Basic algebra allows the formula to be solved for any variable if the other two variables are known; however, for the results to fit the real world different assumptions are required for different variations of the formula. The different variations include:

Average Cycle Time = Average Work in Progress / Average Throughput (EQ1)

Average Throughput = Average Work in Progress / Average Cycle Time

Average Work in Progress  = Average Cycle Time * Average Throughput

The major assumption that equation makes is that system being measured is under steady state.  In a steady state, the process has very little change0-driven variability.  For example, an organization that had just learned Scrum has probably not achieved steady state, therefore the measurement data would highly variable. Small changes are driven by continuous process improvement typically don’t push a process out of a steady state. Looking at the relationship of the variables in equation 1, we can predict that if WIP increases, i.e. if throughput holds steady, cycle time will increase.  Stated differently, according to Little’s Law, if cycle times are too long, consider reducing WIP (continually starting work is a really bad idea).  Equation 1 requires two other assumptions.  The second assumption is that the process has been under observation (measurement) long enough to know that is in a steady state condition.  The third is that consistent unit of measure is being used.

Vacanti explains that if we are focusing Little’s Law on throughput, which is at the heart of predictability in both waterfall and Agile projects, we need to adjust the assumptions. Vacanti identifies two scenarios. The first is the scenario in which WIP goes to zero at some point for example at the end of the project or the end of a perfect sprint. The additional assumption is that anything that enters the process to be worked on will eventually be completed and leave the process.   The second scenario is the exact opposite, WIP does not go to zero.  The added assumptions Vacanti identifies are:

  1. Average input and output are equal.
  2. All work started is completed.
  3. WIP should roughly be the same at the beginning and end of the period being measured.
  4. Average age of WIP is not increasing
  5. WIP, cycle time and throughput are measured using consistent units of measure.
  • Assumptions 1 and 2 reflect conservation of flow.
  • Assumptions 3 and 4 equate to system stability.
  • The 5 assumption is all about the math.

The big takeaway is the that the size of the work item does not matter.  Little’s Law is based on the relationship between averages.  I applied Little’s Law to predicting to forecasting the throughput of a team doing features, tasks, and production support issues. Applying the Little’s Law to the resulted in a prediction with a fit with r squared value greater than 0.9.    More later in the week on using Little’s law and flow metrics!


Previous Installments:

Week 1: Introduction and Game Plan
Week 2:  Flow, Flow Metrics, and Predictability
Week 3  The Basic Metrics of Flow


Actionable Agile Metrics for Predictability: An Introduction by Daniel S. Vacanti


Get your copy and begin reading (or re-reading)!