© 2017 ANDY CARMICHAEL ALL RIGHTS RESERVED DEVELOPER SOUTH COAST APRIL 2017 Essential Kanban Developer South Coast, April 2017 @DevSouthCoast @andycarmich what you even when you’re not using Kanban! need to know about Kanban…
Jan 28, 2018
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Essential
Kanban
Developer South Coast, April 2017@DevSouthCoast @andycarmich
what you even when you’re not using Kanban!need to know about Kanban…
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
ALTERNATIVE TITLE:
WHAT EVERY SCRUM MASTER NEEDS TO KNOW ABOUT KANBAN
@andycarmich@andyCarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
THE FIRST RULE OF KANBAN
CLUB…
@andycarmich@andyCarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
How to start doing Kanban…
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
KANBAN IS A WAY OF SEEING
IT IS NOT A PROCESS FRAMEWORK(NOR A THREAT TO THEM… UNLESS IT UNCOVERS DEMONSTRABLE IMPROVEMENTS IN YOUR CONTEXT)
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
A method for defining, managing, and improvingservices that deliver knowledge work
A “start with what you do now” approach to change
A way of seeing… a way of changing
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
The Foundational Principles ofKanban?
Change Management:• Start here
• Agree to evolve
• Lead – everywhere by everyone
Service Delivery:• Customer first
• Manage the work not workers
• Evolve policies for outcomes
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
EVOLUTION IS NOT SMALLER CHANGE …
IT’S A DIFFERENT MECHANISM OF CHANGE
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Evolution is not a set of small revolutions…
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
It’s a different mechanism
• The evolutionary process involves • differentiation (copying with deliberate differences
or mutations);
• selecting for fitness; and
• keeping and amplifying “more fit” change, while dampening or reversing “less fit” change
• Technology and processes evolve (at an ever increasing rate)
Beinhocker, Eric D. 2007. The origin of wealth: evolution, complexity, and the radical remaking of economics. London: Random House Business Books.
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
MANAGEMENT IS NOT OBSOLETE
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Lead – everywhere by everyone
• Every Scrum Master a manager; every manager a Servant Leader
• Direction – Management – Leadership (communication of intent as well as objectives)
• From leader-follower relationships to leader-leader relationships
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
MAKE POLICY CHANGES
EXPLICIT
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
General Practices of Kanban
•Make the process… … visual … WiP limited … flow … explicit … responsive (with feedback) … evolve
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Policies (1): The old boarding school rule
http://images4.fanpop.com/image/photos/22400000/The-evil-boarding-school-dark-the-hedgehog-22496901-640-456.jpg @andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Policies (2): When “the law is an ass”
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
MEASURE AGILITY BY…
VALUE DELIVERYLEAD TIME
FITNESS FOR PURPOSE
NOT PRACTICE ADOPTION
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Work in Progress
System Lead TimeDelivery Rate
Understand flow systems
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Limit WiP to achieve “pull” and predictable Lead Time
Deferred commitment and limited WiP are cornerstones of effective demand management
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Manage Flow (to maximise flow of value)
Expedite
Elev
ate
the
con
stra
int
Standard
Fixed Date
Intangible
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
OPTIMISE FLOW RATHER THAN
UTILISATION
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
@andycarmich@andycarmich
Flow Systems follow Little’s Law
In 1961 Dr John Little (studying Queuing Theory) proved that, in a stationary system:
λ = L / W
• λ is the average arrival rate
• L is the average number of items in the queue,
• W is the average time in queue
• Subject to similar assumptions we can apply this to kanban systems:
Delivery Rate = WiP / Lead Time
• The overline indicates the average (arithmetic mean)
For other parts of the process use:
• Throughput: the rate items depart the system under consideration.
• Time in Process (TiP): for the time an item takes from entering to leaving the system (or part of the system) under consideration(or Time in Development, Time in Acceptance, Time in Queue, etc.)
Lead Time and Delivery Rate is what we (service provider and customer)
care about
Lead Time and Delivery Rate is what we (service provider and customer)
care about
Work in Progress is what we can control
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Utilisation, Delivery Rate and Lead Time
• Capacity is finite (and lower than you think!
• Increasing utilisation increases LT (not much at first)
• Increasing utilisation increases DR (at first)
• At a certain point flow is destroyed. DR collapses. LT tends to infinity!
• Near the critical point, flow is fragile. Perturbations have severe, unpredictable and long-lasting effects.
• The critical point depends on variability
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Alan Walker [CC BY-SA 2.0 (http://creativecommons.org/licenses/by-sa/2.0)], via Wikimedia Commonshttps://commons.wikimedia.org/wiki/File%3AThe_M1_on_a_winter's_afternoon_-_geograph.org.uk_-_112140.jpg @andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Resource Efficiency or Flow Efficiency
First focus on flow…then efficiency
“What is Lean” – NiklasModig and Par Ahlstrom
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
value trumps …
flow trumps …
waste elimination trumps …
economies of scale .
The “Lean Decision Filter” - after Mehta, Anderson, Raffo, 2008. @andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
NO TEAM IS AN ISLAND
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Efficient subsystems (teams/services) do not necessarily mean efficient systems
• Chains of services (widening your scope of view)
• Hierarchies of services (higher level systems making commitments than team-level services must deliver)
• Networks of services (blockers in one system are demand in another)
Systems Thinking
Service
Service
• optimise the whole not the parts• plan as a whole, balance demand• make queues visible• flexible resources
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
@andycarmich@andycarmich
A chain of services (e.g. the life of a “story”)
In Progress
wmww
wmww
wmww
Done
wmww
wmww
Sprint BL
wmww
Ready
wmww
Preparing
wmww
wmww
wmww
Selected
wmww
wmww
Pool of ideas
wmww
wmww
wmww
UAT
wmww
Released
wmww
wmww
wmww
Deployed
wmww
Customer
wmww
wmww
Consider this process…What happens
before it?
What happens after?
Don’t just optimise this!
Selection:Now? Never?
Later?
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
A hierarchy of services (e.g. multi-team epics)
• Use feedback loops
• Plan work across teams
• Examine the constraints
• Don’t overload!
Team A
Team B
Team C
X
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Feature Selected
2 - 5
Pool of Ideas
User Story Identified
30
User Story Preparation
In Progress Ready
User Story Development
In Progress
Ready (Done)
Feature Acceptance
8In
ProgressReady
Deploy-ment
5
DeliveredFeature
Preparation
3 - 10In
Progress Ready
Discarded
Epic 511
Epic 606
Epic 287
Epic 329
Epic 439
Epic 562
Epic 478
Epic 431
Story 602-01
Epic 335
Epic 302
Epic 602
Epic 512
Story 602-04
Story 602-02
Story 602-06
Story 602-03
Story 602-05
Story 302-03
Story 302-01
Story 302-07
Story 302-02
Story 302-08
Story 302-06
Story 335-09
Story 335-10
Story 335-04
Story 335-08
Story 335-03
Story 335-01
Story 512-04
Story 512-07
Story 512-02
Story 512-05
Story 512-03
Story 512-06
Story 335-06
Story 335-05
Story 335-02
Story 512-01
Story 335-07
Story 302-09
Story 303-05
Story 302-04
Epic 221
Epic 662
Epic 651
Epic 589
Epic 444
Epic 609
Epic 577
Epic 362
Epic 468
Epic 401
Epic 582
Epic 521
Epic 339
Epic 276
Epic 694
Epic 213
Epic 274
Epic 287
Epic 388
Epic 419
Epic 386
Epic 294
15 15
PolicyBusiness case showing
value, cost of delay, size estimate and
design outline.
PolicySelection at
Replenishment meeting chaired by Product Director.
PolicyAs per
“Definition of Done” (see…)
PolicySmall, well-understood,
testable, agreed with PD & Team
PolicyRisk assessed per
Continuous Deploy-ment policy (see…
wikipedia.org/wiki/ Kanban_(development)
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
A network of services (e.g. dependent services)
• Fulfilling system commitments depends on timely delivery from other services
• Same principles apply within a network of interdependent services as for one
• “Scale by applying Kanban in a context of greater scale”
• Use feedback loops
• Make the queues visible to give the “customer” options
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
No team is an islandOptimise the system not the parts
• Islands of efficiency
• Think in systems
•Build a “bigger tribe”
•Where is the value?
Immelman, Ray. 2003. Great boss dead boss: How to exact the very best performance from your company and not get crucified in the process. Gurnee, IL: Stewart Philip International.
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
DEADLINES CREATE TURBULENCE –
MANAGE TO REAL CHANGES IN COST OF DELAY, NOT ITERATION BOUNDARIES.
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Sprint boundaries and Kanban cadences are not deadlines
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Cadence versus Deadline
• Is the sprint boundary a review point in a continuous process, or a stop-start?
• Is behaviour of the team different approaching the deadline?
• Are most things Done near the end?
• Is there pressure to complete the forecast?(does the forecast become a target?)
Different behaviour when analysing stochastic systems is referred to as regime change. Regime change reduces the forecastability of the system.
Real deadlines are where the cost of delay changes radically. Regime change may be a necessary evil as such dates approach.
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Time (t) influences Value (V)
𝐷𝑒𝑙𝑎𝑦 𝐶𝑜𝑠𝑡 = −𝜕𝑉
𝜕𝑡Δ𝑡
“Cost of Delay” (CoD)
A Cost –e.g. measured in dollars
A rate of change in valuee.g. measured in dollars per week
or “Urgency”?
A sudden drop in the value versus release date curve indicates a real deadline
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
How urgent is it to start this item?
• the ordering of work items (WSJF: Urgency / LT)
• Choosing the class of service that is appropriate for the work item • Policies for more value, less risk• Ensuring some intangible items are
always in play (interruptible if unexpected / irrefutable demand)
• Pay attention to the “last responsible moment” for starting a work item (say, 85% probability point)
Source: @lki_dja
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
FINANCIAL POLICIES DETERMINE
MANAGEMENT BEHAVIOUR
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Financial policies that inhibit business agility
• Capitalisation of Work in Progress
• No recognition of Cost of Delay
• Equating Cost to Value (e.g. “Earned Value”)
• Cost saving versus value delivery
• Annual budgeting
• …
When accountants understand the problem of policies which drive unhelpful behaviour, they can reformulate policies. We cannot do this for them – we cannot achieve behaviour change without them.
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
SUMMARY
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
The Kanban Method is not…• a process or a process framework
• nor an informal mechanism for visually organising work, without inconvenient rules
• nor even a scaling framework, though it is applicable at a wide variety of scales
applicable even when you’re “not using Kanban”
• a way of looking at your work, whatever process you use, and improving customer and business outcomes
• a way to see your work through a different lens – the Lean flow paradigm
• a way to focus on the work (rather than the worker), and the customer’s purpose (rather than the service provider’s)
• a way to continually examine the flow of value and incrementally improve it
the first rule of Kanban Club principles and practices policies – the old boarding school rule; when “the law is an ass” evolution is not smaller change, but different change agility is measured by lead time and value delivery, rather than practices adoption no team is an island flow efficiency versus resource efficiency deadlines create turbulence – manage to real changes in cost of delay, not iteration boundaries Financial policies determine management behaviour
Kanban is…
Themes and memes
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
Physical book available from Amazon and other booksellers. PDF can be downloaded from leankanban.com/guide
Template can be downloaded from goo.gl/Ho5nz8!
@andycarmich@andycarmich
© 2
01
7
AN
DY
CA
RM
ICH
AEL
ALL
RIG
HTS
RES
ERV
EDD
EVEL
OP
ERSO
UTH
CO
AST
AP
RIL
20
17
ReferencesAnderson, David J. 2010. Kanban: Successful Evolutionary Change for Your Technology Business. United States: Blue Hole Press.
Anderson, David J., and Andy Carmichael. 2016. Essential Kanban Condensed. United States: Lean Kanban University. http://leankanban.com/guide
Beinhocker, Eric D. 2007. The origin of wealth: evolution, complexity, and the radical remaking of economics. London: Random House Business Books.
Brad Pitt photo. Fight club, 20th Century Fox. 1999
Greenleaf, R. 1982. Servant leadership: A journey into the nature of legitimate power and greatness. 25th ed. New York: Paulist Press International,U.S.
Immelman, Ray. 2003. Great boss dead boss: How to exact the very best performance from your company and not get crucified in the process. Gurnee, IL: Stewart Philip International.
Little, J. 1961. A Proof for the Queuing Formula: L = λ W. Operations Research, 9(3), pp.383-387.
Mehta, M., Anderson, D. and Raffo, D. 2008. Providing value to customers in software development through lean principles. Software Process: Improvement and Practice, 13(1), pp.101-109.
Marquet, D. 2015. Turn the ship around!: A true story of building leaders by breaking the rules. United Kingdom: Portfolio Penguin.
Modig, Niklas and Pär Åhlström. This Is Lean: Resolving the Efficiency Paradox. 1st ed. Stockholm: Rheologica Publishing.
Reinertsen, Donald G. 2009. The principles of product development flow: second generation lean product development. Redondo Beach, CA: Celeritas Publishing.
@andycarmich@andycarmich