What is Kanban? How do you implement it? What are the benefits? Color in Projects Bucharest, March 2013
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Product Managers
PTCs Requests for estimates of future work are often invisible, have an Testers Developers unpredictable arrival rate & are given priority. Discard rate of estimated future work Emergency work is unplanned & User Acceptance is often 50% or greater! receives highest priority. Arrival rate &
volume are unpredictable. PTCs? What did that acronym mean? Items that did not require coding! Effect is hugely disruptive! Why were they treated as emergencies?
Prioritized Backlog
dja@djaa.com, @djaa_dja
Prioritized Backlog
dja@djaa.com, @djaa_dja
So controlling the unplanned, disruptive Prioritized demand would improve Waiting for predictability! Backlog Test
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
5
Change Requests
Test Ready
Done
Testing
UAT
Deployment Ready
Pull
FF F F F F F
Its important to realize the process for B change. software development did not Pull The kanban system is an overlay on the existing process.Pull It changes C scheduling G and prioritization only
PTCs
PTCs are permitted to break the kanban limit *Blocked to service PTC
dja@djaa.com, @djaa_dja
50 30 10
CRs
The Results
dja@djaa.com, @djaa_dja
5
Change Requests
Test Ready
Done
Testing
UAT
Deployment Ready
Pull
B Pull These are the virtual kanban F J G C F Pull F F Boards are not required to do Kanban! F I F The boardD is a visualization of the F * The first system used database triggers workflow process, the work-in-progress to signal pull. There was no board! and the kanban PTCs
dja@djaa.com, @djaa_dja
Commitment is deferred
Backlog Pool of Ideas
Change Requests
Engineering Ready
Development
Ongoing
Test Ready
Done
Testing
UAT
Deployment Ready
Pull
FF F F F F F
D
G
PTCs
Commitment point
dja@djaa.com, @djaa_dja
Engineering Ready
Development
Ongoing
Test Ready
Done
Testing
UAT
Deployment Ready
FF F F
Reject
PTCs
Deferring commitment and Options have value because the avoiding interrupting is uncertain D future workers for estimates makes E there sense when discard rates discard rate implies is no are G 0% uncertainty about the future high!
The discard rate with XIT was 48%. ~50% is commonly observed.
Discarded
I
dja@djaa.com, @djaa_dja
Requirements Analysis
24 - 48
12 - 24
Committed
Development
4 - 12
Testing
Ongoing
Done
Verification
Min & Max limits insure sufficient options are always available
K L
I
F
Committed Work
Reject
O
dja@djaa.com, @djaa_dja
Q
Commitment point
Replenishment Frequency
Pool of Ideas Engineering Ready Development
Ongoing
Replenishment
Pull
Change Requests
Test Ready
Testing
UAT
Deployment Ready
FF F F F F F
E
The frequency of system replenishment should reflect arrival rate of new information and the transaction & coordination costs of holding a I meeting
PTCs
Discarded
I
dja@djaa.com, @djaa_dja
Delivery Frequency
Pool of Ideas
Change Requests
Engineering Ready
Development
Ongoing
Test Ready
Testing
UAT
Deployment Ready
Delivery
Pull
FF F F F F F
Deployment buffer size can On-demand deployment reduce as frequency of deliveryis D most increases agile!
E
The frequency of delivery should reflect the transaction & coordination costs of deployment plus costs & toleranceI of customer to take delivery
PTCs
Discarded
I
dja@djaa.com, @djaa_dja
eering Ready
Development
Ongoing
Test Ready
Done
Testing
UAT
ment Ready
Pull
FF F F F F F
D
G
E
We are now committing to a specific deployment and delivery date
PTCs
Discarded
I
dja@djaa.com, @djaa_dja
Engineering Ready
Ongoing
The clock starts Test ticking when we Testing accept the customers order,UAT not Ready Development when 3 5 it is placed! 3
Done
Deployment Ready
Pull
FF F F F F F
D
G
PTCs
Lead time ends when the item reaches the first queue.
Discarded
I
dja@djaa.com, @djaa_dja
WIP
Lead Time
Pool of Ideas
Ready To Deploy
dja@djaa.com, @djaa_dja
the systems. The deployment 3 Done Ongoing system uses batches and is 2 separate from the kanban system
Testing
Verification Acceptance
Deployment Ready
Done
F The 2nd commitment is actually a commitment PB for the downstream deployment system
The Kanban System gives us confidence to make that I downstream commitment
GY
DE
D G
MN
P1
AB
Identifying Buffers
Pool of Ideas Engineering Ready Development
Ongoing
Testing
verification Acceptance
Done
Deployment Ready
Done
G
GY
PB DE
P1
I am a buffer!
I am buffering non-instant availability or activity with aa availability or an activity with cyclical cadence
dja@djaa.com, @djaa_dja
Engineering Ready
Development
Ongoing
Test Ready
The clock still starts ticking when we accept the customers order, not when it is placed!
Done
Testing
UAT
Deployment Ready
Done
Pull
FF F F F F F
PTCs
D
G
Discarded
I
dja@djaa.com, @djaa_dja
Pool Enginpercentage of total lead time is of eering adding value (or spent actually Development Ideas Ready
Testing
Verification Acceptance
Done
Deployment Ready
Done
Until then customer orders are merely available options Flow efficiency = Work Time Flow efficiencies of 2% have been F reported*. 5% -> 15% D is normal, P1 > 40% G is good!
100%
Lead Time
Multitasking means time spent E in working columns is often waiting time
MN AB
GY
PB DE
Waiting Working
Waiting
Working
Waiting
Lead Time
* Zsolt Fabok, Lean Agile Scotland, Sep 2012, Lean Kanban France, Oct 2012
dja@djaa.com, @djaa_dja
2 1.5 1 0.5 0
1 14
15
22
29
36
43
50
57
64
71
78
85
92
99
10
11
12
12
13
Days
Mean of 31 days
of The SLA workexpectation is of two types: Change 105 Requests (new features); days with 98 % on-time and Production Defects SLA expectation of 44 days with 85% on-time
dja@djaa.com, @djaa_dja
14
Filter Lead Time data by Type of Work (and Class of Service) to get Single Mode Distributions
Production Defects
Change Requests
Mean 5 days
98% at 25 days
85% at 10 days
dja@djaa.com, @djaa_dja
2
Change Requests
Testing
Verification Acceptance
Done
Deployment Ready
Done
E allocation Consistent capacity should bring some to more consistency MN D AB of each type delivery rate of work
PB DE
Lead Time
Production Defects
G
GY
P1
Separate understanding of Separate understanding of Lead Lead Time each type of work Time forfor each type of work
Lead Time
dja@djaa.com, @djaa_dja
When we need it
With a 6 in 7 chance of on-time delivery, we can always expedite to insure on-time delivery
85th percentile
Commitment point
dja@djaa.com, @djaa_dja
Questions? Break
dja@djaa.com, @djaa_dja
and
dja@djaa.com, @djaa_dja
Kanban Method
A management & cultural approach to improvement View creative knowledge work as a set of services
Encourages a management focus on demand, business risks and capability of each service to supply against that demand
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Dont do this!...
Designs Or Defines
Management
Imposes
Process Coaches
Process
Workers
Follow
dja@djaa.com, @djaa_dja
Kanban Method
Uses visualization of invisible work and virtual kanban systems Installs evolutionary DNA in your organization
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Kanban Kata
dja@djaa.com, @djaa_dja
Feedback Loops
The Kanban Kata
Operations Review Improvement Kata Standup Meeting
dja@djaa.com, @djaa_dja
Standup Meeting
Disciplined conduct and acts of leadership lead to improvement opportunities Kaizen events happen at after meetings
dja@djaa.com, @djaa_dja
Improvement Kata
A mentor-mentee relationship
Usually (but not always) between a superior and a sub-ordinate A focused discussion about system capability
Definition of target conditions Discussion of counter-measures actions taken to improve capability
dja@djaa.com, @djaa_dja
Operations Review
Meet monthly Disciplined review of demand and capability for each kanban system
Implement Virtual Kanban Systems Manage Flow Make Policies Explicit Implement Kanban Kata
Educate your workforce to enable collaborative evolvution of policies & ways of working
based on models of workflow from bodies of knowledge such as Theory of Constraints, Demings Profound Knowledge, Lean, Risk Management such as Option Theory
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Demand
Observed Capability
Demand
Observed Capability
dja@djaa.com, @djaa_dja
Demand
Demand
Observed Capability
Demand
Observed Capability
dja@djaa.com, @djaa_dja
Demand
dja@djaa.com, @djaa_dja
Scaling Kanban
Each Kanban System is designed from first principles around a service provided Scale out in a service-oriented fashion Do not attempt to design a grand solution at enterprise scale The Kanban Kata are essential! Allow a better system of systems to emerge over time
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Scaling up
dja@djaa.com, @djaa_dja
Features
5x
2M ar
9M ar
eb
eb
eb
16 -M ar
23 -M ar
During the middle 60% of the project schedule we Time need Throughput (velocity) to average 220 features Inventory Started Designed Coded Complete per month
dja@djaa.com, @djaa_dja
30 -M ar
2006
2008
10 -F
17 -F
24 -F
Littles Law
Calculated based on known lead time capability & required Changing the WIP limit without Plandelivery based on currently observed rate staffing level ratio capability maintaining and current the working represents a change to the way of practices. Do not assume process working. It is a change to the improvements. process and will produce a change Delivery Rate into the observed common cause If changing WIP reduce undesirable capabilityget of the system effects (e.g. multitasking), new Lead Time sample data (perform a spike) to observe the new capability From observed capability Target Treat as a fixed to variable achieve plan
WIP
dja@djaa.com, @djaa_dja
practices/process exhibited an average 22 up to 25 would WIP of 1 itemRounding per person then we conveniently provide 55/week require 25 people organized in 5for 5 teams with ato WIP limit of 5 items each teams of 5 people complete the 0.4 weeks project on-time
Calculated based on known lead time capability & required At this point perhaps just a little delivery rate black magic and experience may be required. If our current working
WIP = 22
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
No real workflow Visualize project completion through physical position of ticket Visualize business risks
dja@djaa.com, @djaa_dja
Projects-in-progress B
Cash Cows 10% budget Growth Markets 60% budget Innovative/New 30% budget
E
F
G
Color may indicate urgency (or some other risk) H
dja@djaa.com, @djaa_dja
Summary of Benefits
dja@djaa.com, @djaa_dja
Collaboration Benefits
Shared language for improved collaboration
Shared understanding of dynamics of flow
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Organizational Benefits
Improved trust and organizational social capital
Improved organizational maturity
Emergence of systems thinking Management focused on system capability through policy definition Adaptability
dja@djaa.com, @djaa_dja
Thank you!
dja@djaa.com, @djaa_dja
About
David Anderson is a thought leader in managing effective software teams. He leads a consulting, training and publishing and event planning business dedicated to developing, promoting and implementing sustainable evolutionary approaches for management of knowledge workers.
He has 30 years experience in the high technology industry starting with computer games in the early 1980s. He has led software teams delivering superior productivity and quality using innovative agile methods at large companies such as Sprint and Motorola.
David is the pioneer of the Kanban Method an agile and evolutionary approach to change. His latest book, published in June 2012, is, Lessons in Agile Management On the Road
to Kanban.
David is a founder of the Lean-Kanban University Inc., a business dedicated to assuring quality of training in Lean and Kanban for knowledge workers throughout the world.
dja@djaa.com, @djaa_dja
Acknowledgements
Hakan Forss of Avega Group in Stockholm has been instrumental in defining the Kanban Kata and evangelizing its importance as part of a Kaizen culture.
Real options & the optimal exercise point as an improvement over last responsible moment emerged from discussions with Chris Matts, Olav Maassen and Julian Everett around 2009.
The inherent need for evolutionary capability that enables organizational adaptation was inspired by the work of Dave Snowden.
dja@djaa.com, @djaa_dja
dja@djaa.com, @djaa_dja
Appendix
dja@djaa.com, @djaa_dja
Example Distributions
Expedite
Intangible
dja@djaa.com, @djaa_dja
Fixed Date
Standard
dja@djaa.com, @djaa_dja