Why Cycle Time Distribution Looks Like it Does and How We Can Use It

Troy Magennis

- Cycle time when plotted as a histogram follows certain patterns

- Understanding the odds of 4 people getting to a dinner reservation on-time (1 chance in 16) is the same issue as how delays cause a skewed distribution.

- Its unlikely that one story encounters EVERY type of possible delay, and its also unlikely that a story encounters NO delay

- By reducing a delay cause, you double the odds of delivery. E.g. for 3 people (possible delays) there is a 1 in 8 chance of on-time "delivery"

- Hypothesis: Cycle time follows a Weibull distribution

- Weibull distributions have a shape parameter. 1 = Exponential distribution this matches dev-ops teams , 1.5 matches Agile/Lean/Scrum teams, 2 = waterfall or thereabouts.

- Our goal is to find which delays move the curve to the left. 

- Possible way to compute: Use an algorithm used to join Frequency of delay, Recency of Delay and Length of delay. RFM analysis was comon in the 80's for direct marketing optimization. Can we use for waste reduction?