Zsolt Fabok
April 12, 2013
1.6k

# [ACCU2013] Measure and Manage Flow in Practice

April 12, 2013

## Transcript

1. ### Measure and Manage Flow in Practice @ZsoltFabok http://zsoltfabok.com/ by Zsolt

Fabok 2013-04-12 Broke the WIP limit TWICE Still on the team #accu2013 http://accu.org/index.php/conferences/accu_conference_2013/
2. ### 5 Stories from the life of a team by using

real application and data The collected data is the courtesy of Digital Natives

4. ### WIP point of observation Visualize the situation with Cumulative Flow

Diagram solved
5. ### 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...
6. ### ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Q D 3 4 1 2 The simplest way of collecting data:
7. ### It is not easy to understand and see throughput from

the CFD A quick detour:

11. ### Let’s add more people to the project so that “things

speed up”! /* detour */

detour */
13. ### Actually, “things slow down”, so it was not a good

idea (solve the right problem instead - systems thinking). End of the detour.

16. ### lead time CFD was not much help here... ...because we

didn’t know much about the nature of the lead time
17. ### 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
18. ### 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

20. ### 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

23. ### 0 4 8 11 15 31-32 33-34 35-36 37-38 Number

of rejected work items count week
24. ### 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

26. ### Sales: “I want to know when the new features can

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

items) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ v v v v v v v v
28. ### Categorizing them into three groups ~ ~ ~ ~ ~

~ ~ ~ ~ ~ ~ ~ v v v v v v v v S M L
29. ### 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
30. ### 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
31. ### 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
32. ### The spent time distribution ~ ~ ~ ~ ~ ~

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

~ ~ v v v v M 0 1 2 3 4 5 6 6 7 8 9 10 SLA hours count

35. ### #1 We decided that we would force ourselves to keep

the SLA #2 Nothing changed. Still the same ratio

38. ### “If you can not measure it, you can not improve

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

I know it's the wrong thing to optimize.” Paul Graham source: http://paulgraham.com/swan.html

41. ### The Hawthorne Experiments (Elton Mayo, 1297) can explain the bus

stop effect: Observation may inﬂuence the measurements.
42. ### “You can easily ﬁnd data that perfectly suits your argument.”

Zsolt Fabok (the handsome bloke who is standing in front of you)

44. ### A sad data: “In 2011 drunk driving caused 15% of

road fatalities in the UK” data: http://news.hastingsdirect.com/drink-driving-caused-15-percent-of-road-deaths-in-2011/ image: http://www.carrentals.co.uk/blog
45. ### If 15% of the road fatalities is caused by drunk

drivers, then 85% is caused by sober drivers. Therefore we can [mistakenly] conclude from data that drunk driving is safer. image: http://www.volker-doormann.org/cavesokr.htm
46. ### So let’s advise people to have at least one beer

before driving... image: http://www.consumeraffairs.com/news/2012/study-shows-big-drop-in-teen-drinking-and-driving-since-1991.html Don’t do that! This is a bad advice based on an even worse conclusion!

48. ### 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.