The Road to a Fairly Predictable System

Abstract: I see agile software development methodologies (Scrum and XP) and modern management methods (Kanban) as tools for predictability. The development organisation would like to know what is coming and whether it will be the right thing to do. On the other side, the management would like to know when the new features can hit the market. Both want predictability but in a different way. We are on the road to a fairly predictable system that can accurately tell us what the next right thing to do will be and also when it is going to be ready. In my talk, I'll share my experiences with you, focusing on practices - from portfolio Kanban to risk management - we do to be more predictable.

Transcript

1. @ZsoltFabok http://zsoltfabok.com/ #llkd14 http://lanyrd.com/2014/llkd14/ by Zsolt Fabok 2014.04.28 The Road to a Fairly Predictable System
2. @ZsoltFabok and /or #llkd14
3. Objective: Reduce “MURA” (unevenness) in the project’s life by increasing predictability
4. Courtesy of Pawel Brodzinski: http://www.slideshare.net/pawelbrodzinski/fixing-portfolio-management
5. The Schedule Portfolio Board
6. Done Week 11 Week 12 Week 13 # ~ 2 / 3 # ~ 0 / 1 # ~ 1 / 1 # ~ 3 / 3 Week 14 Week 15 now # ~ ~ ~ team blue’s item # ~ ~ ~ team red’s item team yellow’s item # ~ ~ ~ # ~ 4 / 5 # ~ 6 / 6 # ~ 4 / 9 # ~ 4 / 4 # ~ 1 / 2 # ~ 0 / 3 # ~ 0 / 4 # ~ 0 / 5 # ~ 0 / 3 # ~ 1 / 4
7. Done Week 12 # ~ 3 / 5 Week 11 Week 13 Week 14 Week 15 portfolio board # ~ 1 / 2 team red’s board Next Design Implementation Review Done # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ 0 / 3 # ~ ~ ~ # ~ ~ ~
8. The changes happen on the team level boards. The portfolio boards only provide a different perspective.
9. The Release Board
10. releases done # ~ 1 / 10 # ~ 2 / 5 # ~ 4 / 5 # ~ 2 / 3 # ~ 3 / 3 Experim ental
11. releases schedule portfolio team Releases Done Done Week 11 Week 12 Week 13 Week 14 Next Implement Review Done # ~ 2/ 3 # ~ 5 / 7 # ~ 1 / 2 # ~ 4 / 6 # ~ ~ ~ # ~ ~ ~ depends updates
12. Predictions based on history
13. On the portfolio level
14. days 0 3 5 8 10 13 15 1 2 3 4 5 6 7 8 9 1011121316 count For example, lead time and WIP distribution for a team on the portfolio level 0 2 5 7 1 2 3 4 count WIP
15. On the team level
16. Courtesy of digitalnatives.hu
17. All the work items we had so far (~20 work items) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ v v v v v v v v
18. Categorizing them into three groups ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ v v v v v v v v S M L
19. 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
20. 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
21. 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
22. The spent time distribution ~ ~ ~ ~ ~ ~ ~ ~ v v v v M 0 1 2 3 4 5 6 6 7 8 9 10 hours count
23. The spent time distribution ~ ~ ~ ~ ~ ~ ~ ~ v v v v M 0 1 2 3 4 5 6 6 7 8 9 10 SLA hours count
24. Courtesy of digitalnatives.hu
25. A big jump forward
26. If we find anything on any of the boards or in any of the measurements, it becomes a risk...
27. The Risk Kanban Board
28. It started with a cost of delay flavored Kanban System.
29. Courtesy of prezi.com
30. # ~ ~ ~ 8 Done 4 # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ 5 2 time impact Ongoing
31. time impact time impact time impact time impact level 1 level 2 level 3 level 4 cost of delay: t2 t1 cost of delay(t1) < cost of delay(t2) Based on Jeff Anderson’s cost of delay functions; http://bit.ly/1fiO9bf
32. time impact time impact time impact level 3 level 3 level 2
33. time impact time impact level 2 level 2 level 1 time impact
34. Review your board during the stand-up meeting:
35. # ~ ~ ~ Done # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ time impact Ongoing # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ priority The Risk Kanban Board
36. Managing risks
37. source: http://www.4dxbook.com
38. source: http://www.4dxbook.com
39. # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ time impact Make your worst risk a WIG (widely important goal)
40. Example Level 2 risk: “team 2 is never on time, and that risks the main release at the end of the year.”
41. WIG: “team 2 will be able to keep his weekly commitments by the 3rd quarter.” w31 w32 w33 w34 w35 input (demand) output (throughput) 14 13 13 15 24 15 23 12 17 10 Lag measure w31 w32 w33 w34 w35 items connected to the portfolio board Lead measure
42. This is a good example, when the lead measure has no effect on the lag measure. w31 w32 w33 w34 w35 19 25 23 20 24 15 23 12 17 10 w31 w32 w33 w34 w35
43. Summing up
44. releases schedule portfolio team risks Releases Done Done Week 11 Week 12 Week 13 Week 14 Next Implement Review Done # ~ 2/ 3 # ~ 5 / 7 # ~ 1 / 2 # ~ 4 / 6 # ~ ~ ~ # ~ ~ ~ # ~ ~ ~ Level 1 Level 2 Level 3 Level 4 # ~ ~ ~ # ~ ~ ~ Done updates
45. Further reading http://zsoltfabok.com/blog/2012/08/waste-in-software- development/ http://zsoltfabok.com/blog/2014/03/risk-kanban-board/ http://zsoltfabok.com/blog/2014/04/schedule-portfolio- board/ http://zsoltfabok.com/blog/2014/02/be-more- predictable/ http://zsoltfabok.com/blog/2014/01/the-purpose-of-agile/
46. Thank you very much for your attention! @ZsoltFabok http://zsoltfabok.com/