Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Measure and Manage Flow in Practice

Zsolt Fabok
October 18, 2012

Measure and Manage Flow in Practice

The sixth iteration of this talk, presented at Lean Kanban France 2012 (the previous iteration was presented at Lean Agile Scotland 2012)

Zsolt Fabok

October 18, 2012
Tweet

More Decks by Zsolt Fabok

Other Decks in Technology

Transcript

  1. Measure and Manage Flow in Practice @ZsoltFabok http://zsoltfabok.com/ #lkfr12 http://lkfr.org/

    by Zsolt Fabok 2012.10.18 Broke the WIP limit TWICE Still on the team
  2. 5 Stories from the life of a team by using

    real application and data The collected data is the courtesy of Digital Natives
  3. The Cumulative Flow Diagram Done Started Queued lead time cycle

    time WIP backlog time number of work items It offers more than just the WIP...
  4. ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

    ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Q D 3 4 1 2 The simplest way of collecting data:
  5. Actually, “things slow down”, so it was not a good

    idea (solve the right problem instead - systems thinking). End of the detour.
  6. lead time CFD was not much help here... ...because we

    didn’t know much about the nature of the lead time
  7. Distribution of lead times days count 0 3 5 8

    10 13 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 22 28 33 56 average median* *Calculation of medians is a popular technique in summary statistics and summarizing statistical data, since it is simple to understand and easy to calculate, while also giving a measure that is more robust in the presence of outlier values
  8. time spent on implementation (hours) time spent waiting (hours) %

    1 63 98 7 57 90 2 62 97 2 62 97 3 61 96 Some examples of work items with 8-day lead time
  9. Distribution of lead times days count 0 3 5 8

    10 13 15 1 4 7 10 13 16 33 Before average 0 3 5 8 10 13 15 1 3 5 7 9 11 13 22 count days After median
  10. 0 4 8 11 15 31-32 33-34 35-36 37-38 Number

    of rejected work items count week
  11. 0 4 8 11 15 31-32 33-34 35-36 37-38 39-40

    42-43 44-45 Number of rejected work items count week
  12. Sales: “I want to know when the new features can

    hit the market!” Management: “I want to know how much it will cost me!”
  13. All the work items we had so far (~20 work

    items) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ v v v v v v v v
  14. Categorizing them into three groups ~ ~ ~ ~ ~

    ~ ~ ~ ~ ~ ~ ~ v v v v v v v v S M L
  15. The lead time distribution ~ ~ ~ ~ ~ ~

    ~ ~ v v v v M 0 1 1 2 3 3 4 1 2 3 4 5 6 7 8 9 12 13 16 days count
  16. The lead time distribution ~ ~ ~ ~ ~ ~

    ~ ~ v v v v M 0 1 1 2 3 3 4 1 2 3 4 5 6 7 8 9 12 13 16 SLA days count
  17. The lead time distribution ~ ~ ~ ~ ~ ~

    ~ ~ v v v v M 0 1 1 2 3 3 4 1 2 3 4 5 6 7 8 9 12 13 16 SLA days count Expired
  18. The spent time distribution ~ ~ ~ ~ ~ ~

    ~ ~ v v v v M 0 1 2 3 4 5 6 6 7 8 9 10 hours count
  19. The spent time distribution ~ ~ ~ ~ ~ ~

    ~ ~ v v v v M 0 1 2 3 4 5 6 6 7 8 9 10 SLA hours count
  20. #1 We decided that we would force ourselves to keep

    the SLA #2 Nothing changed. Still the same ratio
  21. “If you can not measure it, you can not improve

    it.” Lord Kelvin http://en.wikipedia.org/wiki/File:Lord_Kelvin_photograph.jpg
  22. “If you start measuring something you start optimizing it, and

    I know it's the wrong thing to optimize.” Paul Graham http://paulgraham.com/swan.html
  23. 1. We develop software not models (value) 2. Demand first,

    supply second 3. Observe the system (lead time, throughput) 4. Start measuring, look back if necessary 5. Manage 6. Mind that data expires 7. Goto step 3.