SlideShare a Scribd company logo
1 of 37
for leading
Continuous Improvement
Kanban Metrics
in practice
@BattistonMattia
About me
● from Verona, Italy
● software dev & continuous improvement
● Kanban, Lean, Agile “helper”
● Sky Network Services
Mattia Battiston
@BattistonMattia
mattia.battiston@gmail.com
Ciao!
Why are we here?
OUR
EXPERIENCE
WHY
HOW
IMPROVING
LESSONS
LEARNT
FORECASTING
Kan...what?
a little knowledge of Kanban helps
(limiting WIP, lead time, value vs waste, queues, batches, etc.)
Why do we need metrics?
#1: drive continuous improvement #2: forecast the future
But I thought metrics were bad....
Typical problems:
gaming
dysfunctions
Good vs Bad metrics
● look at improving the whole system ● reward/punish individuals
“95% performance is attributable to
the system, 5% to the people”
W. Edwards Deming
● feedback about state of reality ● used as target
● leading (let you change behaviour) ● lagging (tell you about the past)
● all metrics must improve ● local optimisations
Our system
Iteration-Based
On-demand
Direct
How do we collect the data?
The Spreadsheet
Inputs: story details; start time and duration of each state
Public version: https://goo.gl/0A9QSN
For you to copy, reuse, get inspired,
etc.
All the maths you need
● Min, Max
Normal: data is distributed
around a central value
e.g. height of UK population
Skewed: data has a long tail
on one side (positive or
negative)
e.g. income of UK population
(positive skew)
Lead time of stories follows
skewed distribution
● Average (mean)
avg(1,2,2,2,3,14) = (1+2+2+2+3+14)/6 = 4
● Median: separates the high half from the low half. Less impacted by outliers
median(1,2,2,2,3,14) = 2
● Mode: value that occurs more frequently
mode(1,2,2,2,3,14) = 2
● Standard Deviation: measures the amount of dispersion from the average. When
high, values are spread over a large range.
stdev(1,2,2,2,3,14) = 4.5; stdev(1,2,2,2,3,5) = 1.2;
● Percentile: percentage of elements that fall within a range
50% perc(1,2,2,3,7,8,14) = 3; 80% perc(1,2,2,3,7,8,14) = 7.8;
● Normal Distribution vs Skewed Distribution:
Cumulative Flow Diagram
Description: Each day shows how many stories are in each state
n.stories
days
Cumulative Flow Diagram
Ideal CFD: thin lines growing in parallel at a steady rate -> good flow!
Cumulative Flow Diagram
● Objective: retrospect (but needs a good facilitator)
CFD used for Retrospective
● Objective: demonstrate effectiveness of changes
changed WIP limit in DEV from 3 to 2
Cumulative Flow Diagram
● Objective: decide what you should work on today
● Objective: forecasting: rough info about lead time, wip, delivery date (although
they’re easier to use when tracked separately)
WIP
Lead Time
CFD Patterns
(taken from CFD article by Pawel Brodzinski)
growing lines: indicate large WIP + context switching.
action: use WIP limits
stairs: indicates large batches and timeboxes
action: move towards flow (lower WIP,
more releases, cross-functional people)
flat lines: nothing’s moving on the board
action: investigate blockers, focus on finishing, split in
smaller stories
single flat line: testing bottleneck
action: investigate blockers, pair with testers,
automate more
typical timeboxed iterationdropping lines: items going back
action: improve policies
metrics for
Delivery
Time
Control Chart
Description: For each story it shows how long it took. Displays Upper and Lower control
limits; when a story falls out of limits something went wrong and you should talk about it.
stories
leadtime(days)
Cycle/Lead Time stats + History
Description: Stats to get to know your cycle time and lead time. They let you predict “how
long is the next story likely to take?”. Visualize trends of improvement
Lead Time distribution
lead time (days)
n.storiesthattookthatlong
Description: For each lead time bucket (#days), how many stories have taken that long.
Useful to show as a percentage to know probability.
50%
80%
Story Health
Description: Indicates if the story is in good health or if we should worry about it. Based
on lead time distribution
50-80% >90%80-90%0-50%
0-4 gg 5-7 gg 8-10 gg >10 gg
Cycle Time vs Release Prep. Time
stories
days
Description: For each story shows how long it spent in the iteration and in release
preparation (Context specific). Used to discuss cost vs value of release testing.
metrics for
Predictability
Iteration Throughput
iteration
no.storiescompleted
Description: Number of stories that get done for each iteration
Rolling Wave Forecasting
Description: visualise in the backlog the likelihood of stories getting done in the next 2
weeks
Arrivals Rate
Description: how often a story is started, aka pulled into our system (arrival). This is how
often you can change your mind about what to do next
Points vs Lead Time
leadtime(days)
story points
Description: Shows low correlation between estimated points and actual lead time
Disney Stations
Description: Like queueing at Disneyland. “How long in here? How long from here?”
Task Time
Description: Shows how long tasks usually take (context specific). Gives you an idea of
how long a story will take based on n. of tasks
metrics for
Quality
Bugs percentage
Description: Percentage of bugs over stories. Also expressed as “1 bug every X stories”
metrics for
Continuous
Improvement
Flow Efficiency
Description: Shows how long stories have spent in queues - nobody was working on
them. Shows how much you could improve if you removed waiting time.
Time in status
timespentinstate(days)
story
Description: for each story visualise how long it spent in each status (absolute and
percentage). Shows trends of where stories spend more time
Retrospective
Resources
Books
Metrics
● Data driven coaching - Troy Magennis
● Seven Deadly Sins of Agile Measurement - Larry Maccherone
● The Impact of Lean and Agile Quantified - Larry Maccherone
● Kanban at Scale: A Siemens Success Story - Bennet Vallet
● FocusedObjectives@Github - Troy Magennis
● Visual feedback brings key Agile principles to life - Bazil
Arden
● How visualisation improves Psychological Safety - Bazil
Arden
Forecasting
● Cycle Time Analytics - Troy Magennis
● Top Ten Data and Forecasting Tips - Troy Magennis
● Forecasting Your Oranges - Dan Brown
● Using Maths to work out Potentially Deliverable Scope -
Bazil Arden
● Forecasting Cards - Alexei Zheglov
Story Points
● Story Points and Velocity: The Good Bits - Pawel
Brodzinski
● No correlation between estimated size and actual time
taken - Ian Carroll
Lead Time
● Analyzing the Lead Time Distribution Chart - Alexei
Zheglov
● Inside a Lead Time Distribution - Alexei Zheglov
● Lead Time: what we know about it, how we use it - Alexei
Zheglov
● The Economic Impact of Software Development Process
Choice - Troy Magennis
More
● Flow Efficiency - Julia Wester
● Cumulative Flow Diagram - Pawel Brodzinski
Thank you!
@BattistonMattia
mattia.battiston@gmail.com
really, really appreciated! Help me improve

More Related Content

What's hot

What's hot (20)

The Lego Kanban Game
The Lego Kanban GameThe Lego Kanban Game
The Lego Kanban Game
 
Align, Inform, Inspire: Measuring Business Agility and SAFe® with Flow Metrics
Align, Inform, Inspire: Measuring Business Agility and SAFe® with Flow MetricsAlign, Inform, Inspire: Measuring Business Agility and SAFe® with Flow Metrics
Align, Inform, Inspire: Measuring Business Agility and SAFe® with Flow Metrics
 
Agile metrics for predicting the future
Agile metrics for predicting the futureAgile metrics for predicting the future
Agile metrics for predicting the future
 
Forecasting with Less Effort and More Accuracy (Agile Camp NY 2018)
Forecasting with Less Effort and More Accuracy (Agile Camp NY 2018)Forecasting with Less Effort and More Accuracy (Agile Camp NY 2018)
Forecasting with Less Effort and More Accuracy (Agile Camp NY 2018)
 
Agile Scrum Estimation
Agile   Scrum EstimationAgile   Scrum Estimation
Agile Scrum Estimation
 
Agile Metrics...That Matter
Agile Metrics...That MatterAgile Metrics...That Matter
Agile Metrics...That Matter
 
Agile effort estimation
Agile effort estimation Agile effort estimation
Agile effort estimation
 
The Kanban Retrospective
The Kanban RetrospectiveThe Kanban Retrospective
The Kanban Retrospective
 
STATIK (Systems Thinking Approach to Introduce Kanban) Workshop Meetup
STATIK (Systems Thinking Approach to Introduce Kanban) Workshop MeetupSTATIK (Systems Thinking Approach to Introduce Kanban) Workshop Meetup
STATIK (Systems Thinking Approach to Introduce Kanban) Workshop Meetup
 
Kanban boards step by step
Kanban boards step by stepKanban boards step by step
Kanban boards step by step
 
Intro to Kanban - AgileDayChile2011 Keynote
Intro to Kanban - AgileDayChile2011 KeynoteIntro to Kanban - AgileDayChile2011 Keynote
Intro to Kanban - AgileDayChile2011 Keynote
 
Paper Boats Simulation - Facilitation Guide
Paper Boats Simulation - Facilitation GuidePaper Boats Simulation - Facilitation Guide
Paper Boats Simulation - Facilitation Guide
 
Agile Estimating Technique
Agile Estimating TechniqueAgile Estimating Technique
Agile Estimating Technique
 
Agile Transformation in Telco Guide
Agile Transformation in Telco GuideAgile Transformation in Telco Guide
Agile Transformation in Telco Guide
 
Agile metrics
Agile metricsAgile metrics
Agile metrics
 
Modern Agile - Joshua Kerievsky
Modern Agile - Joshua KerievskyModern Agile - Joshua Kerievsky
Modern Agile - Joshua Kerievsky
 
OKR Canvas - Ágiles 2018
OKR Canvas - Ágiles 2018OKR Canvas - Ágiles 2018
OKR Canvas - Ágiles 2018
 
Kanban India 2022 - Keynote - Todd Little | Turbocharge your Scrum with Kanban
Kanban India 2022 - Keynote - Todd Little |  Turbocharge your Scrum with KanbanKanban India 2022 - Keynote - Todd Little |  Turbocharge your Scrum with Kanban
Kanban India 2022 - Keynote - Todd Little | Turbocharge your Scrum with Kanban
 
The Four Keys - Measuring DevOps Success
The Four Keys - Measuring DevOps SuccessThe Four Keys - Measuring DevOps Success
The Four Keys - Measuring DevOps Success
 
PopcornFlow: Continuous Evolution Through Ultra-Rapid Experimentation
PopcornFlow: Continuous Evolution Through Ultra-Rapid ExperimentationPopcornFlow: Continuous Evolution Through Ultra-Rapid Experimentation
PopcornFlow: Continuous Evolution Through Ultra-Rapid Experimentation
 

Viewers also liked

Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
azheglov
 

Viewers also liked (20)

Real Life Clean Architecture
Real Life Clean ArchitectureReal Life Clean Architecture
Real Life Clean Architecture
 
Kanban Metrics in practice at Sky Network Services
Kanban Metrics in practice at Sky Network ServicesKanban Metrics in practice at Sky Network Services
Kanban Metrics in practice at Sky Network Services
 
Metriche Kanban in pratica a Sky UK [ITA]
Metriche Kanban in pratica a Sky UK [ITA]Metriche Kanban in pratica a Sky UK [ITA]
Metriche Kanban in pratica a Sky UK [ITA]
 
CONTINUOUS IMPROVEMENT: HELL ON EARTH? (KATHERINE KIRK) - LKCE13
CONTINUOUS IMPROVEMENT: HELL ON EARTH? (KATHERINE KIRK) - LKCE13CONTINUOUS IMPROVEMENT: HELL ON EARTH? (KATHERINE KIRK) - LKCE13
CONTINUOUS IMPROVEMENT: HELL ON EARTH? (KATHERINE KIRK) - LKCE13
 
Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
Lean Kanban Central Europe 2014: Beyond VSM: Understanding and Mapping Your P...
 
Metrics As A Learn And Change Agent
Metrics As A Learn And Change AgentMetrics As A Learn And Change Agent
Metrics As A Learn And Change Agent
 
Learned Helplessnes
Learned HelplessnesLearned Helplessnes
Learned Helplessnes
 
Agile DC Lead Time
Agile DC Lead TimeAgile DC Lead Time
Agile DC Lead Time
 
Growth Hacking for Product Managers
Growth Hacking for Product ManagersGrowth Hacking for Product Managers
Growth Hacking for Product Managers
 
Scrum vs ScrumAnd vs ScrumBut: Which one are you doing? :: Agile Tour London ...
Scrum vs ScrumAnd vs ScrumBut: Which one are you doing? :: Agile Tour London ...Scrum vs ScrumAnd vs ScrumBut: Which one are you doing? :: Agile Tour London ...
Scrum vs ScrumAnd vs ScrumBut: Which one are you doing? :: Agile Tour London ...
 
Learned Helplessness
Learned HelplessnessLearned Helplessness
Learned Helplessness
 
Servant leadership un neutered 2016 09 21
Servant leadership un neutered 2016 09 21Servant leadership un neutered 2016 09 21
Servant leadership un neutered 2016 09 21
 
Inside Lean Kanban (#lkuk14 keynote)
Inside Lean Kanban (#lkuk14 keynote)Inside Lean Kanban (#lkuk14 keynote)
Inside Lean Kanban (#lkuk14 keynote)
 
2014 02 kanbans not-so-hidden agendas 2
2014 02 kanbans not-so-hidden agendas 22014 02 kanbans not-so-hidden agendas 2
2014 02 kanbans not-so-hidden agendas 2
 
Creating Resilient, Robust, & Antifragile Organizations
Creating Resilient, Robust, & Antifragile OrganizationsCreating Resilient, Robust, & Antifragile Organizations
Creating Resilient, Robust, & Antifragile Organizations
 
Kanban values exercise, 2016 edition
Kanban values exercise, 2016 editionKanban values exercise, 2016 edition
Kanban values exercise, 2016 edition
 
Enterprise Services Planning - Effective Middle Management
Enterprise Services Planning - Effective Middle ManagementEnterprise Services Planning - Effective Middle Management
Enterprise Services Planning - Effective Middle Management
 
Key Note - PMI Congress Poland - The Role of the Project Manager with Kanban
Key Note - PMI Congress Poland - The Role of the Project Manager with KanbanKey Note - PMI Congress Poland - The Role of the Project Manager with Kanban
Key Note - PMI Congress Poland - The Role of the Project Manager with Kanban
 
STATIK, Kanban's Hidden Gem (#lkce14 talk)
STATIK, Kanban's Hidden Gem (#lkce14 talk)STATIK, Kanban's Hidden Gem (#lkce14 talk)
STATIK, Kanban's Hidden Gem (#lkce14 talk)
 
Creating Space to Be Awesome -- Offentlig Chef
Creating Space to Be Awesome -- Offentlig ChefCreating Space to Be Awesome -- Offentlig Chef
Creating Space to Be Awesome -- Offentlig Chef
 

Similar to Kanban Metrics in practice for leading Continuous Improvement

Release planning using feature points
Release planning using feature pointsRelease planning using feature points
Release planning using feature points
Madhur Kathuria
 
agilefinancenarrated-190414134302.pptx
agilefinancenarrated-190414134302.pptxagilefinancenarrated-190414134302.pptx
agilefinancenarrated-190414134302.pptx
ssuser1ecf25
 

Similar to Kanban Metrics in practice for leading Continuous Improvement (20)

Agile estimation
Agile estimationAgile estimation
Agile estimation
 
Agile Estimation
Agile EstimationAgile Estimation
Agile Estimation
 
Agile metrics - Measure and Improve
Agile metrics - Measure and ImproveAgile metrics - Measure and Improve
Agile metrics - Measure and Improve
 
Speak To The Business! Agile Metrics That Inform Rather Confuse the Business
Speak To The Business! Agile Metrics That Inform Rather Confuse the BusinessSpeak To The Business! Agile Metrics That Inform Rather Confuse the Business
Speak To The Business! Agile Metrics That Inform Rather Confuse the Business
 
I love the smell of data in the morning (getting started with data science) ...
I love the smell of data in the morning (getting started with data science)  ...I love the smell of data in the morning (getting started with data science)  ...
I love the smell of data in the morning (getting started with data science) ...
 
Minimal Viable Agile
Minimal Viable AgileMinimal Viable Agile
Minimal Viable Agile
 
Agile estimating 12112013 - Agile KC Dec 2013
Agile estimating 12112013 - Agile KC Dec 2013Agile estimating 12112013 - Agile KC Dec 2013
Agile estimating 12112013 - Agile KC Dec 2013
 
When will it be done? (Lean Agile Forecasting)
When will it be done? (Lean Agile Forecasting)When will it be done? (Lean Agile Forecasting)
When will it be done? (Lean Agile Forecasting)
 
Practical Agile Analytics: Reduce uncertainty and stop making such a big deal...
Practical Agile Analytics: Reduce uncertainty and stop making such a big deal...Practical Agile Analytics: Reduce uncertainty and stop making such a big deal...
Practical Agile Analytics: Reduce uncertainty and stop making such a big deal...
 
NoEstimates@iNatuix
NoEstimates@iNatuixNoEstimates@iNatuix
NoEstimates@iNatuix
 
Release planning using feature points
Release planning using feature pointsRelease planning using feature points
Release planning using feature points
 
Machine Learning Project - Default credit card clients
Machine Learning Project - Default credit card clients Machine Learning Project - Default credit card clients
Machine Learning Project - Default credit card clients
 
Story Points considered harmful – a new look at estimation techniques
Story Points considered harmful – a new look at estimation techniquesStory Points considered harmful – a new look at estimation techniques
Story Points considered harmful – a new look at estimation techniques
 
Agile 2014 Software Moneyball (Troy Magennis)
Agile 2014   Software Moneyball (Troy Magennis)Agile 2014   Software Moneyball (Troy Magennis)
Agile 2014 Software Moneyball (Troy Magennis)
 
When will it be done? - Agile Camp Chicago
When will it be done? - Agile Camp ChicagoWhen will it be done? - Agile Camp Chicago
When will it be done? - Agile Camp Chicago
 
Data driven coaching - Deliver 2016
Data driven coaching - Deliver 2016Data driven coaching - Deliver 2016
Data driven coaching - Deliver 2016
 
Data Driven Decision Making - Chapter 1. Team Performance
Data Driven Decision Making - Chapter 1. Team PerformanceData Driven Decision Making - Chapter 1. Team Performance
Data Driven Decision Making - Chapter 1. Team Performance
 
agilefinancenarrated-190414134302.pptx
agilefinancenarrated-190414134302.pptxagilefinancenarrated-190414134302.pptx
agilefinancenarrated-190414134302.pptx
 
The agile forecast joe tristano southern fried agile 2018_ final
The agile forecast joe tristano  southern fried agile 2018_ finalThe agile forecast joe tristano  southern fried agile 2018_ final
The agile forecast joe tristano southern fried agile 2018_ final
 
7. space the estimation aid for bringing agile delivery predictability - p...
7. space   the estimation aid for bringing agile delivery predictability  - p...7. space   the estimation aid for bringing agile delivery predictability  - p...
7. space the estimation aid for bringing agile delivery predictability - p...
 

Recently uploaded

TECUNIQUE: Success Stories: IT Service provider
TECUNIQUE: Success Stories: IT Service providerTECUNIQUE: Success Stories: IT Service provider
TECUNIQUE: Success Stories: IT Service provider
mohitmore19
 
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
Health
 

Recently uploaded (20)

Optimizing AI for immediate response in Smart CCTV
Optimizing AI for immediate response in Smart CCTVOptimizing AI for immediate response in Smart CCTV
Optimizing AI for immediate response in Smart CCTV
 
HR Software Buyers Guide in 2024 - HRSoftware.com
HR Software Buyers Guide in 2024 - HRSoftware.comHR Software Buyers Guide in 2024 - HRSoftware.com
HR Software Buyers Guide in 2024 - HRSoftware.com
 
Software Quality Assurance Interview Questions
Software Quality Assurance Interview QuestionsSoftware Quality Assurance Interview Questions
Software Quality Assurance Interview Questions
 
How To Troubleshoot Collaboration Apps for the Modern Connected Worker
How To Troubleshoot Collaboration Apps for the Modern Connected WorkerHow To Troubleshoot Collaboration Apps for the Modern Connected Worker
How To Troubleshoot Collaboration Apps for the Modern Connected Worker
 
TECUNIQUE: Success Stories: IT Service provider
TECUNIQUE: Success Stories: IT Service providerTECUNIQUE: Success Stories: IT Service provider
TECUNIQUE: Success Stories: IT Service provider
 
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
 
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
 
ManageIQ - Sprint 236 Review - Slide Deck
ManageIQ - Sprint 236 Review - Slide DeckManageIQ - Sprint 236 Review - Slide Deck
ManageIQ - Sprint 236 Review - Slide Deck
 
Right Money Management App For Your Financial Goals
Right Money Management App For Your Financial GoalsRight Money Management App For Your Financial Goals
Right Money Management App For Your Financial Goals
 
BUS PASS MANGEMENT SYSTEM USING PHP.pptx
BUS PASS MANGEMENT SYSTEM USING PHP.pptxBUS PASS MANGEMENT SYSTEM USING PHP.pptx
BUS PASS MANGEMENT SYSTEM USING PHP.pptx
 
Exploring the Best Video Editing App.pdf
Exploring the Best Video Editing App.pdfExploring the Best Video Editing App.pdf
Exploring the Best Video Editing App.pdf
 
%in Bahrain+277-882-255-28 abortion pills for sale in Bahrain
%in Bahrain+277-882-255-28 abortion pills for sale in Bahrain%in Bahrain+277-882-255-28 abortion pills for sale in Bahrain
%in Bahrain+277-882-255-28 abortion pills for sale in Bahrain
 
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
 
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
 
A Secure and Reliable Document Management System is Essential.docx
A Secure and Reliable Document Management System is Essential.docxA Secure and Reliable Document Management System is Essential.docx
A Secure and Reliable Document Management System is Essential.docx
 
Pharm-D Biostatistics and Research methodology
Pharm-D Biostatistics and Research methodologyPharm-D Biostatistics and Research methodology
Pharm-D Biostatistics and Research methodology
 
The Real-World Challenges of Medical Device Cybersecurity- Mitigating Vulnera...
The Real-World Challenges of Medical Device Cybersecurity- Mitigating Vulnera...The Real-World Challenges of Medical Device Cybersecurity- Mitigating Vulnera...
The Real-World Challenges of Medical Device Cybersecurity- Mitigating Vulnera...
 
Microsoft AI Transformation Partner Playbook.pdf
Microsoft AI Transformation Partner Playbook.pdfMicrosoft AI Transformation Partner Playbook.pdf
Microsoft AI Transformation Partner Playbook.pdf
 
10 Trends Likely to Shape Enterprise Technology in 2024
10 Trends Likely to Shape Enterprise Technology in 202410 Trends Likely to Shape Enterprise Technology in 2024
10 Trends Likely to Shape Enterprise Technology in 2024
 
AI & Machine Learning Presentation Template
AI & Machine Learning Presentation TemplateAI & Machine Learning Presentation Template
AI & Machine Learning Presentation Template
 

Kanban Metrics in practice for leading Continuous Improvement

  • 1. for leading Continuous Improvement Kanban Metrics in practice @BattistonMattia
  • 2. About me ● from Verona, Italy ● software dev & continuous improvement ● Kanban, Lean, Agile “helper” ● Sky Network Services Mattia Battiston @BattistonMattia mattia.battiston@gmail.com Ciao!
  • 3. Why are we here? OUR EXPERIENCE WHY HOW IMPROVING LESSONS LEARNT FORECASTING
  • 4. Kan...what? a little knowledge of Kanban helps (limiting WIP, lead time, value vs waste, queues, batches, etc.)
  • 5. Why do we need metrics? #1: drive continuous improvement #2: forecast the future
  • 6. But I thought metrics were bad.... Typical problems: gaming dysfunctions
  • 7. Good vs Bad metrics ● look at improving the whole system ● reward/punish individuals “95% performance is attributable to the system, 5% to the people” W. Edwards Deming ● feedback about state of reality ● used as target ● leading (let you change behaviour) ● lagging (tell you about the past) ● all metrics must improve ● local optimisations
  • 9. How do we collect the data?
  • 10. The Spreadsheet Inputs: story details; start time and duration of each state Public version: https://goo.gl/0A9QSN For you to copy, reuse, get inspired, etc.
  • 11. All the maths you need ● Min, Max Normal: data is distributed around a central value e.g. height of UK population Skewed: data has a long tail on one side (positive or negative) e.g. income of UK population (positive skew) Lead time of stories follows skewed distribution ● Average (mean) avg(1,2,2,2,3,14) = (1+2+2+2+3+14)/6 = 4 ● Median: separates the high half from the low half. Less impacted by outliers median(1,2,2,2,3,14) = 2 ● Mode: value that occurs more frequently mode(1,2,2,2,3,14) = 2 ● Standard Deviation: measures the amount of dispersion from the average. When high, values are spread over a large range. stdev(1,2,2,2,3,14) = 4.5; stdev(1,2,2,2,3,5) = 1.2; ● Percentile: percentage of elements that fall within a range 50% perc(1,2,2,3,7,8,14) = 3; 80% perc(1,2,2,3,7,8,14) = 7.8; ● Normal Distribution vs Skewed Distribution:
  • 12. Cumulative Flow Diagram Description: Each day shows how many stories are in each state n.stories days
  • 13. Cumulative Flow Diagram Ideal CFD: thin lines growing in parallel at a steady rate -> good flow!
  • 14. Cumulative Flow Diagram ● Objective: retrospect (but needs a good facilitator) CFD used for Retrospective ● Objective: demonstrate effectiveness of changes changed WIP limit in DEV from 3 to 2
  • 15. Cumulative Flow Diagram ● Objective: decide what you should work on today ● Objective: forecasting: rough info about lead time, wip, delivery date (although they’re easier to use when tracked separately) WIP Lead Time
  • 16. CFD Patterns (taken from CFD article by Pawel Brodzinski) growing lines: indicate large WIP + context switching. action: use WIP limits stairs: indicates large batches and timeboxes action: move towards flow (lower WIP, more releases, cross-functional people) flat lines: nothing’s moving on the board action: investigate blockers, focus on finishing, split in smaller stories single flat line: testing bottleneck action: investigate blockers, pair with testers, automate more typical timeboxed iterationdropping lines: items going back action: improve policies
  • 18. Control Chart Description: For each story it shows how long it took. Displays Upper and Lower control limits; when a story falls out of limits something went wrong and you should talk about it. stories leadtime(days)
  • 19. Cycle/Lead Time stats + History Description: Stats to get to know your cycle time and lead time. They let you predict “how long is the next story likely to take?”. Visualize trends of improvement
  • 20. Lead Time distribution lead time (days) n.storiesthattookthatlong Description: For each lead time bucket (#days), how many stories have taken that long. Useful to show as a percentage to know probability. 50% 80%
  • 21. Story Health Description: Indicates if the story is in good health or if we should worry about it. Based on lead time distribution 50-80% >90%80-90%0-50% 0-4 gg 5-7 gg 8-10 gg >10 gg
  • 22. Cycle Time vs Release Prep. Time stories days Description: For each story shows how long it spent in the iteration and in release preparation (Context specific). Used to discuss cost vs value of release testing.
  • 24. Iteration Throughput iteration no.storiescompleted Description: Number of stories that get done for each iteration
  • 25. Rolling Wave Forecasting Description: visualise in the backlog the likelihood of stories getting done in the next 2 weeks
  • 26. Arrivals Rate Description: how often a story is started, aka pulled into our system (arrival). This is how often you can change your mind about what to do next
  • 27. Points vs Lead Time leadtime(days) story points Description: Shows low correlation between estimated points and actual lead time
  • 28. Disney Stations Description: Like queueing at Disneyland. “How long in here? How long from here?”
  • 29. Task Time Description: Shows how long tasks usually take (context specific). Gives you an idea of how long a story will take based on n. of tasks
  • 31. Bugs percentage Description: Percentage of bugs over stories. Also expressed as “1 bug every X stories”
  • 33. Flow Efficiency Description: Shows how long stories have spent in queues - nobody was working on them. Shows how much you could improve if you removed waiting time.
  • 34. Time in status timespentinstate(days) story Description: for each story visualise how long it spent in each status (absolute and percentage). Shows trends of where stories spend more time
  • 36. Resources Books Metrics ● Data driven coaching - Troy Magennis ● Seven Deadly Sins of Agile Measurement - Larry Maccherone ● The Impact of Lean and Agile Quantified - Larry Maccherone ● Kanban at Scale: A Siemens Success Story - Bennet Vallet ● FocusedObjectives@Github - Troy Magennis ● Visual feedback brings key Agile principles to life - Bazil Arden ● How visualisation improves Psychological Safety - Bazil Arden Forecasting ● Cycle Time Analytics - Troy Magennis ● Top Ten Data and Forecasting Tips - Troy Magennis ● Forecasting Your Oranges - Dan Brown ● Using Maths to work out Potentially Deliverable Scope - Bazil Arden ● Forecasting Cards - Alexei Zheglov Story Points ● Story Points and Velocity: The Good Bits - Pawel Brodzinski ● No correlation between estimated size and actual time taken - Ian Carroll Lead Time ● Analyzing the Lead Time Distribution Chart - Alexei Zheglov ● Inside a Lead Time Distribution - Alexei Zheglov ● Lead Time: what we know about it, how we use it - Alexei Zheglov ● The Economic Impact of Software Development Process Choice - Troy Magennis More ● Flow Efficiency - Julia Wester ● Cumulative Flow Diagram - Pawel Brodzinski

Editor's Notes

  1. thank you everyone for coming to my talk. Other sessions are great Can I ask: who here has an idea of what Kanban is? who here already knows what a CFD is? Great, you are in the right place!
  2. My name is Mattia, I’m from a small lovely town in Italy called Verona. You might know it as the city of Romeo and Juliet Background of software development, now team leader. I don’t consider myself a real coach but I like to help teams improve (run workshops, retros, talks, etc.) I work at Sky Network Services, we’re the department in Sky that deals with the Voice and Broadband. The way I explain it to my wife is “we make the internet work”. Oh btw we’re hiring
  3. this presentation is a slightly modified version of one that I do internally for teams that want to learn about Kanban metrics I use it to share our experience about why we like using metrics, what metrics we use and how, and how we collect the data
  4. not mandatory, but it helps if you have an idea of what kanban is and its values (e.g. limiting WIP) if you need help on this consider going to the London Lean Kanban Day, last year it was great and I’m sure this year it’s going to be even better!
  5. Why do I even need metrics? what problem are we trying to solve? #1 - improve (hints on where you should improve, validate experiments); Kanban is all about continuous improvement -> start with what you do, and use data to improve basically you are constantly running experiments and validating them with your data #2 - forecast future, move away from estimates and use historical data to predict the future Forecasting is a big topic that deserves to be discussed on its own, for this presentation I’m focusing on #1 (but if you want to know more we can have a chat later, or look up Troy Magennis, Kanbandan, Dimitar Bakardzhiev
  6. the typical reaction when you start talking about metrics is “how about no, thanks” The argument is usually that metrics can be gamed and they will cause dysfunctions (you’ll destroy the system but the metrics will say you’re doing great) example: velocity; we have to complete 10 points in each iteration -> double the points for each story! So yes, we are stepping in a high risk area when we talk about metrics
  7. That’s why we distinguish between good and bad metrics: good metrics are about improving the system as a whole, rather than rewarding/punishing individuals. We recognize Deming’s rule about the performance of a system (95/5). They always work with a systemic approach, team metrics absolutely not used as target! they are a feedback mechanism. Only for internal use, no exposure to management, PMs, etc. they are usually expressed as trends rather than single numbers; single numbers tends to become targets, trends instead can tell you a more generic “you are improving” they are leading to let you act on them, rather than lagging when you can’t do anything about it
  8. we use Jira but that has been setup with a very generic process to fit all teams; we’re pretty much a physical shop; we have physical boards to represent the real workflow and we either print cards or use post its this is the main board of my team, representing the main part of our process we represent our WIP limits with placeholders -> next = 3, dev = 2, test = 3; an empty placeholder is a pull signal we’ve got the next two stories to work on in next, then dev, functional testing (with “waiting” as a buffer), and waiting for a cut our point of commitment is when a story enters Next, so that’s when our Lead Time starts CLICK for “Iteration based” apps, stories have to wait in “waiting for cut” until the end of the the even iteration (2-4 weeks). Then we do a cut of the application and it goes through release testing, then to production. the “on demand” apps we can do the cut as soon as we want, so we try to do it as soon as possible and release as soon as possible “direct” stories are small things that have to be done (emails, reports, etc.) and go straight to done we treat these three as 3 different work item types, based on either different workflow or different speed in the same workflow This is very important because for most metrics you want to differentiate between work item types. They will have different lead time and different demand In particular for “On demand” and “Direct” we calculate the lead time from Next to Done But for “Iteration-based” we only count the time from Next to Cut and call it “Iteration Cycle time”, as the rest of the time is fixed time
  9. collecting the data is really simple. We record transitions; we stamp the card each time it moves from one state to the next this piece of information is enough to get most of the metrics we use then we regularly put this information in a spreadsheet, where we record the work item type and the transitions. You might decide to track some extra pieces of information, e.g. for bugs we want to record the environment it was found in. It’s up to you btw there are formulas to only count working days you’re probably wondering why we don’t use a tool. for collecting the data: make sure you are recording reality (real workflow) and you can change the data (e.g. if you forget to update it) Jira is quite bad, can’t update the dates for displaying and analysing the data: Jira is rubbish. Other tools do something, but in my experience you still want access to the raw data and go crazy with a spreadsheet. You will want to reorganise data, rearrange, split it differently, play with it...tools don’t have enough flexibility to do data mining. If you use a tool, make sure you have access to the raw data
  10. Our spreadsheet: the only input is some details about the story, what state it’s in, and then for each state when it enters that state and how long it stayed in there Collect CFD data every day - it’s inferrable but it makes it easier All the rest is calculated or inferred (organized in various sheets) Our spreadsheet has grown in complexity over time because of many experiments, eventually I will redo it and publish it
  11. I left this slide in but I’ll skip it, you can read it later if you want; the math involved is really easy, and all the formulas are already in excel
  12. Probably the most famous Kanban chart for each day it shows how many stories are in each state
  13. can be used in retrospectives and root cause analysis to look at history (but needs good facilitator)
  14. can be used as leading, but it might be just easier to look at the board keep queueing states as thin as possible don’t let any state grow too much alert when you don’t see flow (when chart is not going steadily up)
  15. one of the most famous (with CFD); usually done with dots, but it was easier to use columns (because of non-numeric data) Objective: retrospect: talk about stories that took longer or shorter than expected, and improve your process or policies; as you improve you should see trends of improvement leading: see stories approaching the limit and decide if you want to act on it Tips & Traps: use percentiles instead of std. deviation for the limits (std. dev only if you have a normal distribution) hard to talk about problems all the time (too busy to improve)
  16. we calculate some stats about our cycle time (or lead time). Objective: forecasting. Lets you answer questions like “how long do stories usually take? what are the chances that the next story is going to take longer than 10 days?” Tips and traps: distinguish between “all time” and “last X months”. I usually look at the last 5 or 6 months (the process is constantly improving, older data is not representative anymore) shows trends of improvements, but doesn’t really tell you why
  17. on x-axis you’ve got days, on y the number of stories that took that long. You should find a skewed distribution. You can draw the curve that interpolates the data - it’s called Weibull distribution Get the probability of each bucket, and if you sum them you can find that 50% of stories take 6 days or less, 85% of stories take 10 days or less So next time you ask “how long will the next story take?” you can decide how certain you want to be and pick a number. “how long will the next story take with a 80% confidence?” If I want a story to be done by a particular date, I know it needs to be in Next at least 10 days before the fixed date to have 83% confidence Objective: forecasting (on a story level, but this is the data you’d use to do a montecarlo simulation) Tips & Traps: long tail is symptom of high standard deviation (high variability) multiple peaks are often hiding multiple work item types from the shape of the weibull you can draw some conclusions
  18. concept of Health of a story, based on how long the story has been in progress for based on the lead time distribution I know that 50% stories take up to 6 days, so I consider that green. After 6 days it becomes yellow, we start worrying about it. And then red and black. Objective: leading. what should I work on today? It’s a way to escalate problems and raise alarms Tips and traps: remember to do it per work item type (can’t expect different work items to take the same amount of time)
  19. this is quite specific to our context for those stories that are iteration-based we show how long they wait in “release preparation” You can see that release testing takes up most of the time. That’s the effect of having a big batch, with all the dysfunctions that come from it. We put this together with the low value that we get from it, which is number of bugs found, to decide that it would have been crazy for new projects to follow the same approach That’s why for new applications we moved to a flow approach context specific, but it’s an example of how you can use data to drive your argument
  20. throughput: how many stories are done in a particular amount of time? we use the iteration as cadence, so “how many stories are done in two weeks?” Objective: how many stories should I plan in next iteration? are we going faster? Traps: If you split by work item type you might have iterations where you did nothing of that particular work item type; so the average is kind of weird depending on what project we’re working on, if it only involves iteration-based applications I would look at one throughput or another
  21. Shows how often a story is started Also called takt time in lean This is how often you can change your mind about what to do next (instead of deciding every two weeks) Arrivals and departures should be balanced
  22. this is a controversial one, people tend to have strong feelings for or against story points for stories of 1 points, they took from 2 to 10 days; stories of 2 points took 2 to 20 days; etc. very low correlation between story points and actual lead time this worked as a shock factor and we decided to stop wasting time with planning poker, or fingers in the air Now we do story breakdown and use historical data (e.g. lead time distribution) to forecast how long they’ll take
  23. like at disneyland “how long do I wait from here?” we use 50th and 80th percentile to show how long stories will take from here, and how long they’re going to spend in here Tips and Traps: remember, it’s only valid on a “per-work item” basis (can’t mix) useful to take decisions in the middle of the iterations
  24. another context specific metrics. When stories are in next we create a list of tasks for the story to agree on the scope and the acceptance criterias we monitored how long tasks take for a while, and now we can predict quite accurately how long a story is going to take based on the n. of tasks highly accurate for dev time (+ creating tasks makes scope clearer) Objective: forecast: how long is a story going to take, based on n. of tasks? leading: how much is left to do on this story? should we swarm? or should I rather start another story? Tips & Traps: high correlation between dev and n. tasks very low correlation between test and n. tasks (makes sense) also helps with defining scope of stories
  25. one simple way of keeping track of quality: count number of bugs; we express it as “n. of bugs per stories” so that we can keep them into account when we plan in future
  26. one of the best incarnation of lean mindset. Shows how long stories have spent in queue states, therefore no one was actively working on them (pure waste in lean) It’s a demonstration of deming’s rule - just by removing wait time we could improve our performance of 50% How do you reduce wait time? probably reduce WIP, have true cross-functional people, attack sources of variability Tips: represent queue states on your board by using red labels
  27. shows where stories spend most of the time; interesting to compare it with what the team perceives are the states that take longer. Clearly shows that development is only a small fraction of time; is this our intended process, or are we here just by chance? does this reflect the importance of each state? e.g. if we think development is the main state, are we happy with it being just a small fraction? lots of potential but hard to use for the risk of people feeling blamed
  28. GOOD drive changes to process: tells you what you should improve on and gives you directions for what to change validate experiments and arguments: can see if a change is having the effect you wanted, or find good arguments for your point enable forecasting helps answer “what should I work on today” infinite learning possibilities, only your imagination is the limit google spreadsheet as tool worked very well WATCH OUT resist the temptation to automate everything, or even worse create a custom tool! You will keep changing charts, metrics, reorganize them, etc. Use intelligent automation, just to make manual steps easier. example: you could capture the CFD data automatically, but I still prefer to have a look and copy/past them when I’m happy they’re right don’t obsess over precision, you’re looking for trends rather than precision. Number won’t always 100% precise but it doesn’t matter only use the last period of time, for example the last 6 months. Data older than that might not be accurate anymore, or might not reflect your current process people still know it better: if you have a reason to think that reality is different from what a metric is saying, you’re probably right; it’s worth investigating, you’re probably on to learning something new! PROBLEMS it was hard to get the team on board, and I actually never succeeded. NIM team is particularly difficult, people tend to have strong opinions and you have to find the right moment when they’re willing to listen. This is true in general. Be prepared to do it yourself, don’t expect people to help you until they think it’s helping them. It still works, you can look at the metrics and decide what’s the next change you’ll introduce. Just make sure you’re never forcing anything on the team, that’s an instant fail it’s hard to make these metrics speak: what do they mean? how do I interpret a particular chart? how do I read this? It’s hard to make them user friendly enough. So again, you can still use them as a management tool to drive the changes you introduce, but you need to make then easier to use if you want people to look at them as soon as you have work item types with different process the complexity explodes; I don’t know how to fix this, you need to make your metrics even more user friendly difficult to make it visible; ideally these metrics should be printable as some kind of dashboard, so before standup and after every iteration someone refreshes them and prints them. But I never managed to do it as soon as you open the spreadsheet you get an information overload that often scares people. Again, need to improve the usability. Add instructions, help, etc. should organise the metrics better, probably by their usage (example: daily vs iteration, lead time vs predictability) write down when important changes or important events happen, so that when you’re looking at the past you know “that’s when we changed the WIP limit in dev, have we improved?” it doesn’t matter what the numbers are saying, you’ll never be able to convince people with just numbers. You need to translate that to a feeling, make them feel a problem and then they’ll listen. example: it doesn’t matter that all metrics are telling you that you have too much WIP, people will be scared of WIP limits no matter what