How-How Diagram: A Practical Approach to Problem Resolution
When do you need it by? Business Agility Metrics
1. @martinaziz
Martin Aziz
Business Change Consultant
martinaziz@gmail.com
When will it
be done?
When do
you need it
by?
Business-Agility Metrics in
an Agile World
Toronto Agile Conference 2018
2. Do customers always ask
“when will it be done?”
Likelihood
to ask it
Low
High
3. Making Sense of a Metrics Dashboard
Sprint Velocity
by team
Happiness IndexTest Coverage
Number of teams
Doing Scrum
How long it tookStory Point Totals
Level of Automation Feature Usage
4. Your customers maybe cares about
one or two of these.
Sprint Velocity
by team
Happiness IndexTest Coverage
Number of teams
Doing Scrum
How long it tookStory Point Totals
Level of Automation Feature Usage
5. 4 Types of Metrics
1. Fitness Criteria
2. Improvement Drivers
3. General Health
Indicators
4. Vanity Metrics
Metrics aren’t all the same.
Let’s consider a way to classify them.
6. .
Has thresholds
Fit – positively
affects choice
Unfit – Negatively
affects choice
Overserving – No
longer affects choice
Fitness Criteria – The only thing that
directly affect customer choices
e.g. Fast Enough
Sufficient Quality
Sufficient Selection
Taste
Weight, etc…
7. e.g.
Dollar per Win Ratio
Level of Automation
Profit per customer visit
Test Coverage
Impact to fitness
criteria needs to be
understood.
May have unintended
consequences and
diminishing returns.
Improvement Drivers- Have a target
to be achieved.
8. e.g.
Cholesterol levels
Static Code Analysis
Employee Engagement
Scores
Staff Turnover
Test Coverage
Can be within a defined
healthy range without
intervention
“let’s keep an eye on that”
General Health Indicators – A range of healthy variability.
Is the business healthy enough to pursue its strategy?
9. Vanity Metrics – Make you feel good, doesn’t affect
customer choice.
Mistaken for one of the
other 3 metrics.
In many cases driven
from company culture.
e.g.
Consecutive Games Won
Number of Agile Teams
On-Time Departure
10. 4 Types of Metrics
1. Fitness Criteria – Affect
Selection
2. Improvement Drivers – Have
Targets
3. General Health Indicators –
Have Safe Ranges
4. Vanity Metrics – Make you feel
good
fitterforpurpose.com
11. Reexamining the dashboard
Sprint Velocity
by team
Happiness IndexTest Coverage
Number of teams
Doing Scrum
How long it tookStory Point Totals
Level of Automation Feature Usage
?
Fitness Criteria
Improvement
Drivers
Health
Indicators
Vanity Metrics
12. Focusing on how long the
customer waits *
* Lead Time
The Organization
Value
Customer
Customer
“I promise”
“Here it is”
13. The Cumulative Flow Diagram (CFD)
Compares the rate that work arrives to the rate it is being
completed.
7
6
5
4
3
2
1
1 2 3 4 5 6 7 8 9
Average time customer waits
Average work in
the process
Time (days)
WorkItems
14. Lead Time Distribution Histogram
0
5
10
15
20
25
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58
Frequency
Duration (days)
AVG 5.7
90th 14.4
Average - not particularly useful
Confidence Percentile – Useful in providing
confidence intervals for prediction & planning.
“90% probability we can delivery within 15 days”
Length of tail. Gives us information
about our delivery variability. i.e. size
of the risk
15. Service Control Chart Point in time where
change was
introduced.
New upper
boundary
of variation.
Control Charts
We introduce a lot of changes to try to improve.
How do we know they are working?
16. “How Agile are We?”
Weekly
Daily
Hourly
Monthly
Quarterly
Yearly
Replenishment Frequency
Weekly
Daily
Hourly
Monthly
Quarterly
Yearly
Release Frequency
How Fast
How
Predictable
Frequency FrequencySpeed Predictability
Market
Strategy
www.leanatoz.com
17. Why do we need to know about flow?
Will adding more
horsepower to these
cars solve the problem?
18. A look at Flow Efficiency
Wait Wait Wait WaitWork Work Work Work
Flow Efficiency =
work
work + wait
x 100%
< 2% common < 15% frequent < 40% good + rare
Industry Range
20. Have you ever seen the weather
predicted this way?
Week Complete!
Mon. Tue. Wed. Thu. Fri. Sat. Sun.
21. We’re more used to seeing this
Percentage
likelihood Range of
possibilities
~ scientific
for “ish”
Range
22. Contrasting the two approaches
Week Complete!
Mon. Tue. Wed. Thu. Fri. Sat. Sun.
Deterministic Probabilistic
More Useful
But… how useful is a
weather report for
80 days from now?
23. A number of tools allow you to
forecast probabilistically
github.com/FocusedObjective
25. In order to predict, we need to
understand variation
Mediocristan
26. Mediocristan
What is the variation of height in this room?
4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.8 8 8.2 8.4
“Bell Curve” / “Gaussian Distribution”
MEDIAN = 5.8 ftAVG=5.7 ft
27. Mediocristan
What is the variation of height in this room?
4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.8 8 8.2 8.4
“Bell Curve” / “Gaussian Distribution”
Q: What happens if
Andre the Giant
walked into this
room?
4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.8 8 8.2 8.4
André René Roussimoff
“Andre the Giant”
7 ft 4 in
MEDIAN = 5.8 ftAVG=5.7 ft
A: Very little
28. Extremistan
What is the variation of wealth in this room?
Exponential / Pareto Distribution
MEDIAN
= $701K
AVG
= $844K
Emergence of a tail
29. Extremistan
What is the variation of wealth in this room?
Exponential / Pareto Distribution
Q: What
happens if Bill
Gates walked
into this room?
MEDIAN
= $701K
AVG
= $844K
Emergence of a tail
A: On average
everyone in this
room is a
billionaire!
MEDIAN = $701K AVG = $1B
Bill Gates
Net Worth $94.6B
31. How many features
will be required?
How long will each
feature take to do?
How many delays? Impact of each delay?
Throughput
Variability is
embedded from
multiple sources
Work in Process
Where is all this variation coming
from in Knowledge Work?
How can we
possibly answer
“when will I get it?”
33. How many features
will be required?
How long will each
feature take to do?
How many delays? Impact of each delay?
Throughput
Demand Shaping
Batch Size
Feature Split Rates
Policies
Classes of Service
Skill Liquidity
Feature Split Rates
Work in Process
Batch Size
Service Flow
WIP
Limit WIP
Cut the tails.
And you will be
able to predict!
34. Work in
Process
Lead Time
Classes of
Service
Lead Time
System Capacity
Work Types
The Kanban
Method
gives you
control over
variability
35. Expedite
“Security Breach”
Fixed Date
“Stadium on time for Olympics”
Standard
“Regular delivery from Amazon”
Cost of Delay Profiles
Time
Impact
of
Delay
Time
Impact
of
Delay
Impact
of
Delay
Time
Don’t treat all work the same
36. Managed Tails. Allow for predictions.
Deferred Commitment based on “When do you need it by?”
Need it byToday
New capability.
Manages the tail.
Allows prediction.
Optimal
Start Date
Old capability. Tail
not managed. Hard
to predict. Tendency
to come in late.
Allows you to defer
committing to
start.
37. “When do you need it by” in action!
Ok, so when
will it be
done?
When do
you need it
by?
We need it for
the tradeshow
in 8 weeks
Our capability for this
type of work is 3 weeks.
As long as we commit to
start the work at little
over 3 weeks before
the show, it will get
done.
We should be able to
commit to the work in in a
few weeks. But I’m
confident it will be ready in
time for the tradeshow.
That’s fantastic!
38. Thank you Find out more
Q&A
Resources:
o KMP I & II – Lean Kanban Institute
o ESP – Lean Kanban Institute
o Fit for Purpose – Zheglov & Anderson
o Antifragile & Skin in the Game – Nassim Taleb
o The Flaw of Averages – Sam Savage
o Focused Objective Tools - Troy Magennis
o Swift Kanban – Digité Software
KanbanTO Meetup
Schedule a Consult
@martinaziz
Martin Aziz
Business Change Consultant
martinaziz@gmail.com