henrik kniberg: lean from the trenches keynote @ agileee

Post on 20-Aug-2015

2.083 Views

Category:

Documents

6 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Lean from the Trenches Oct 6, 2012

Agile Eastern Europe, Kiev

Henrik Kniberg Agile/Lean coach

www.crisp.se

2 Henrik Kniberg 2

3 Henrik Kniberg

3

4 Henrik Kniberg

4

5 Henrik Kniberg 5

Kanban

Scrum

XP

Lean principles

6

<dislaimer>

No solutions. Only examples.

</disclaimer>

Henrik Kniberg 6

7

Once upon a time…

7

8

PUST – Polisens Utrednings STöd

Henrik Kniberg 8

2011 2010

9 Henrik Kniberg 9

Timeline

Q3 Q4 Q3 Q4 Q1 Q2 Q1 Q2

2009 2010 2011

1.1 1.2 1.3 1.5

Project launch

10 people

30 people 60+ people

1.0

First pilot

1.4

Nationwide

10

Slicing the elephant

Henrik Kniberg 10

PUST

1.0

1.2

1.3 1.4

1.5

Region Östergötland, Uppsala, etc

Crime types (weapon, drunk driving, shoplifting, etc)

Integrations

1.1

11 Henrik Kniberg 11

PUST Project

”Customer”

Acceptance test user group

Onsite user

Pilot users

User involvement

12

Team structure & ”Daily cocktail party”

12

13

Team structure - before

Henrik Kniberg 13

3 Development

teams

Test team

Requirements analyst team

?

% #

!

!

?

% #

!

!

14

Improved team structure

Henrik Kniberg 14

3 Feature teams

System test team

Requirements analyst team

15 Henrik Kniberg 15

”Daily cocktail party” 9:15 – 10:15

16 Henrik Kniberg

16

Feature team 1 Feature team 2 Feature team 3

9:30 – 9:45 9:15 – 9:30 9:30 – 9:45

9:45 – 10:00

Test sync

Requirements sync

Dev sync

9:45 – 10:00

Project sync

10:00 – 10:15

17

The project board

17

18 Henrik Kniberg 18

Next 10 features Ideas Features Development System

test

User acceptance

test

FLOW

Production

19 Henrik Kniberg 19

20

Planning (2w)

week 1 week 2 week 3 week 4 week 5 week 6 week 7 week 8

Retrospectives (2w)

Release (2m)

Demo & systest (continuous)

Cadences

21

22 Henrik Kniberg 22

23

Impediments & escalation

Henrik Kniberg 23

Feature blocked

General impediments

asdfasdfasdf As police!I need to scan!So that !

Can’t test!!Waiting fo

r!

barcode reader!

Jim!March 12!

2011-03-01!

Police car = urgent (patch)

24

Scaling the boards

24

25 Henrik Kniberg 25

Next 10 features

Ready for sys test

Dev in progress

Sys test progress

Feature swimlanes

26

Henrik Kniberg 26

Dev Team 1 Dev Team 2 Dev Team 3

Next 10 features

Dev in progress

Ready for sys test

27 Henrik Kniberg 27

Definition of ”ready for

development”

Definition of ”ready for

system test”

28 Henrik Kniberg 28 Henrik Kniberg 28

29

How we handle tech stories

29

30

Tech stories

Henrik Kniberg 30

Next 5 tech stories

Next 10 features

Ready for Development

31

Example: the 7 meter class

Henrik Kniberg 31

32 Henrik Kniberg 32

”Oh no, bottleneck in System Test!

FLOW

33 Henrik Kniberg 33

Bottleneck

”Let’s stop building new

features”

”... and focus on test automation!”

Question of the week: How can we best

contribute to system test today?

34 Henrik Kniberg 34

Everyone doing tech

stories

35

How we handle bugs

35

36 Henrik Kniberg 36

Test Fix %&@#!

Release

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8

Release

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8

Test at end:

Test

%&@#!

Fix

Fix Time saved!

Test

Test continuously:

Before: Test at end

Now: Test continuously

37

Bug fixing process

Henrik Kniberg 37

Bug found!

More important than any of the current top 30?

No

Write sticky-note, find developer,

fix now!

Yes

Replace one of the other

top 30 bugs with this one

Yes

Ignore it

No

Don’t log it. Fix it NOW!

Blocker?

38

Three input queues

Henrik Kniberg 38

Next 5 tech stories

Next 10 features

Next 5 lower priority bugs

39 Henrik Kniberg 39

Recurring bugs

40

How we continuously improve the process

40

41

Retrospectives every 1-2 weeks

Henrik Kniberg 41

42

Too much, too fast!

Henrik Kniberg 42

43

Limiting the rate of change

Henrik Kniberg 43

”A4” improvement proposal

”A4” improvement proposal

”A4” improvement proposal

”A4” improvement proposal

44 Henrik Kniberg 44

Proposal:  More  Customer-­‐Valued  Stories  Why?  What  Problem  Are  We  Trying  to  Solve?  •  Hard  to  get  an  overview  of  the  project  board  

from  customer  perspec7ve,  many  stories  are  so  small  that  they  can’t  be  delivered.  

DescripAon  /  FAQ  A  story  that  goes  into  development  must:  1.  Be  size  S  or  M  2.  Be  as  customer  valuable  as  possible,  as  long  as  we  don’t  

break  the  size  rule.  

The  requirements  team  ensures  that  each  card  under  ”Ready  for  Development”  is  a  customer-­‐valued  story  (regardless  of  size).  However,  before  it  enters  development  it  must  be  S  or  M.    Ques7on:  What  happens  if  the  story  is  L,  and  must  be  delivered  as  a  whole  before  it  is  valuable  to  the  customer?  •  Break  it  down  to  smaller  stories  (new  cards)  which  are  size  M,  

but  with  highest  possible  customer  value  per  story.  •  Visually  group  these  stories  by  wri7ng  the  name  of  the  higher  

level  feature  in  big  blue  leSers  at  the  top  of  each  card.  

Who  Is  Affected  By  The  Change?  •  Requirements,  development,  and  test  teams.  

What  Are  the  Change  ImplementaAon  Steps?  •  Update  Defini7on  of  Done  for  ”Ready  for  

Development”,  add  ”the  story  is  valuable  to  the  customer”.  

•  Go  through  the  board  &  iden7fy  stories  that  are  too  small  to  be  valuable.  Combine  these  into  bigger  stories  (as  long  as  they  don’t  exceed  Medium).  

Remove Confiscation

CONFISCATION

S

Confiscation L

Example:  Register

Confiscation

CONFISCATION

M

45

Limiting the rate of change

Henrik Kniberg 45

”A4” improvement proposal

”A4” improvement proposal

”A4” improvement proposal

”A4” improvement proposal

46

How we capture and use process metrics

46

47

Process metrics

"   Velocity = feature per week "   Throughput time = weeks per feature

48 Henrik Kniberg 48

Ready for acceptance test

(this week so far)

Ready for acceptance test

(previous weeks)

Velocity per week

Velocity – stories per week

49 Henrik Kniberg 49

49

# of features delivered to acceptance test

Week

Velocity-based release planning All of these will

be done

Some of these will be done, but not all

None of these will be done

50 Henrik Kniberg 50

51 Henrik Kniberg 51

Bed time!

Timeline (5 minute intervals)

Total work

Play time!

52 Henrik Kniberg 52

53

Velocity improvement

Henrik Kniberg 53

Q1 3 features per week

Q2 6.5 features per

week

# of features delivered to acceptance test

Week

54

Throughput time – weeks per feature

Henrik Kniberg 54

Next 10 features

Ready for acceptance

test Throughput time (elapsed days)

55 Henrik Kniberg 55

Elapsed days

Feature

56

Throughput time improvement

Henrik Kniberg 56

Q1 6-14 weeks per feature Q2

3-6 weeks per feature

Elapsed days

57

Surprising insight

Henrik Kniberg 57

”Small” & ”Medium” features take roughly same

amount of time

Large:  59  days  

Medium:  37  days  

Small:  31  days  Average throughput time

Elapsed days

Feature

58

How we track the high level goal

58

59 Henrik Kniberg 59

High level goal

Milestone

60

Death March Detection using gut feel

Henrik Kniberg 60

5 = certainly 4 = probably 3 = barely 2 = probably not 1 = forget it

Do you believe the current goal is achievable?

61 Henrik Kniberg 61

62

Wrapup

62

63

Incremental delivery Setting the project up for success Co-location

Customer involvement

64

Creating a culture of continuous improvement

Henrik Kniberg 64

Communication

Data

Clarity

65

Perfection is a direction, not a place

Henrik Kniberg

The important thing is not how you work.

The important thing is how you improve the way you work!

top related