Cost of Delay is a lightweight approach to feature and product prioritization that asks a simple question: how much does it cost you not to have something? Reinertsen has said that Cost of Delay is the most important thing to quantify when producing a product. Great, but how do you start? How do you assign a dollar amount to something you have not built yet? How do we make sure that our teams focus on building the most important thing right now? This talk will give you the tools you need to understand Cost of Delay, as well as a set of techniques, from simple proxies to more sophisticated real-dollar analyses to help you understand the impact of delays on your organization.
20. 20
“Highest-Paid Person’s Opinion”
Pros: Easy. All the burden of thinking is
deferred to the hippo.
Cons: The hippo rarely has enough
information to make good decisions.
Also, as in nature, when hippos
compete, it gets messy. Value is not
maximized at the same rate as ego
gratification.
HIPPO PRIORITIZATION
21. 21
• Must Have
• Could Have
• Should Have
• Won’t
Pros: Brings people together to have a
conversation about what’s important.
Cons: Everything quickly turns into
“must have.” If everything is a priority,
nothing is.
MOSCOW PRIORITIZATION
22. 22
The organization allocates a proportion
of development capacity to different
departments or stakeholders.
Pros: It feels fair. Everyone feels taken
care of.
Cons: The is no larger strategic vision
that helps maximize value for the entire
organization. Local optimization at the
expense of overall value creation.
EQUITY
23. 23
Before committing to work, we analyze:
• Benefit/cost ratio
• Payback analysis
• Net present value discounts
Pros: More objective.
Cons: A ton of wasteful analysis is done
before you get approval. Leads to log-
jam up front, and unsustainably slow
value delivery on the back-end. Not
BENEFIT/COST RATIO HIGH JUMP
24. 24
In order to make better economic decisions, we need a
framework for decisions that:
• Gives us a solid economic basis for our decision making
• Enables a focus on value and urgency
• Doesn’t waste a lot of time on unnecessary analysis
WHAT WE NEED
33. BLACK SWAN FARMING
TYPES OF VALUE Increasing sales to new or existing
customers. Delighting or disrupting to
increase market share and size.
INCREASE REVENUE
Improvements and incremental innovation
to sustain current market share and
revenue.
PROTECT REVENUE
Improvements to sustain current cost base.
Costs we are not currently incurring but
may in the future.
AVOID COSTS
Costs that we are currently incurring that we can
reduce. More efficient, improved margin or
contribution.
REDUCE COSTS
34. 34
Cost of Delay Profile #1
ShortLifecycle,PeakAffectedbyDelay
When there is a significant first-mover advantage. The cost to late
delivery is significant.
$BenefitsReceived
Time
Short Lifecycle - Peak Affected by Delay
Planned Entry
Late Entry
35. 35
Cost of Delay Profile #2
LongLifecycle,PeakUnaffectedbyDelay
Often seen when there is a manual process being automated.
$BenefitsRealized
Time
Long Lifecycle - Peak Unaffected by Delay
Planned Entry
Late Entry
Delay
Cost
36. 36
Cost of Delay Profile #3
LongLifecycle,PeakAffectedbyDelay
$BenefitsRealized
Time
Long Lifecycle - Cost Affected by Delay
Planned Entry
Late Entry
38. 38
How to Get Started With Cost of Delay
Pros:
• Simple
• Easy to Reason about
• Good way to swag
what’s worth doing
further analysis on
Cons:
• Imprecise
• Still might be liable
to HIPPO warfare
39. 39
Step Two: Business Value Proxies
AKA,WHATSAFEDOES
User Business
Value
Time
Criticality
Risk-Reduction
and/or
Opportunity
Enablement
Cost of Delay
Where every factor is on a modified Fibonacci
scale of 1, 2, 3, 5, 8, 13, or 20.
In a list, every column must have at least one
value that is 1.
40. 40
Step Two: Business Value Proxies
AKA,WHATSAFEDOES
• Simple
• Easy to work with/reason
about
• Trying to pin down precise
outputs on imprecise inputs
is often a waste of time
• Having all three components
at the same level can lead to
contradictions.
• If you are just giving rough
proxy values, there’s still
ample room for a HIPPO
fight. (You need to go
beyond play money).
PROS CONS
42. 42Step Three: Express CoD in Real Dollars
BECAUSEACTUALMONEYISAREALMOTIVATOR
• Both simple and expressive
• Nothing captures urgency
quite as compellingly as the
almighty dollar
• Puts real teeth into alignment
• Even HIPPOs have to pay
attention
• Still depends on trust (the
extravagant liar problem)
• No guarantee of accuracy
PROS CONS
47. SJF (SHORTEST JOB FIRST) SCHEDULING
47
When delay costs are homogeneous,
do the shortest job first.
Project Duration Cost of Delay
A 1 month $3m/mo
B 3 months $3m/mo
C 10 months $3m/mo
CostofDelay
Duration
Shortest Job first
CostofDelay Duration
Longest Job first
B
C
A
Delay
Cost
B
C
C
C B
ADelay
Cost
48. HDCF (HIGH DELAY COST FIRST) SCHEDULING
48
When job durations are
homogeneous, do the high
cost-of-delay job first.
CostofDelay
Duration
Highest COD first
Project Duration Cost of Delay
A 3 months $10m/mo
B 3 months $3m/mo
C 3 months $1m/mo
A
B B
BDelay Cost
C
CostofDelay Duration
Lowest COD first
A
B
B
B
Delay Cost
C
49. WSJF (WEIGHTED SHORTEST JOB FIRST) SCHEDULING
49
Project Duration Cost of Delay COD/Duration
A 1 month $10m/mo 10
B 3 months $3m/mo 1
C 10 months $1m/mo 0.1
When job durations and delay
costs are not homogeneous,
use WSJF.
CostofDelay
Duration
Highest WSJ first
B
C
CostofDelay Duration
Lowest WSJ first
B
B
BDelay Cost
C
Delay Cost
A
B A
50. WSJF (WEIGHTED SHORTEST JOB FIRST) SCHEDULING
50
Project Duration Cost of Delay COD/Duration
A 4 months $10m/mo 2.5
B 3 months $6m/mo 2.0
C 10 months $14m/mo 1.4
When job durations and delay
costs are not homogeneous,
use WSJF.
What is the best sequence?
51. DEPENDENCIES/BIG-BANG DEPLOYMENT OBVIATE WSJF
51
Project Duration Cost of Delay COD/Duration
A 1 10 10
B 3 3 1
C 10 1 0.1
If you dependencies require you
to batch release, cost of delay is
a moot point.
CostofDelay
Duration
Highest WSJ first
B
C
CostofDelay Duration
Lowest WSJ first
B
CA
B
A
Delay CostDelay Cost
54. FIFO (FIRST IN, FIRST OUT) SCHEDULING
54
Project Duration Cost of Delay
A ? ?
B ? ?
C ? ?
When job durations and delay
costs are unknown, use FIFO.
CostofDelay
Duration
First in, First Out
C
B
A
Delay
Cost
55. 55
Decision Table
homogeneous or Unknown Heterogeneous
homogeneous or Unknown First in, First out Highest Cost of Delay First
Heterogeneous Shortest Job First
Weighted Shortest Job
First
Duration
Cost of Delay
Use this table when
choosing a sequencing
strategy for a given set
of options, depending
on whether their cost of
delay and duration are
known or similar.
61. CLASSES OF SERVICE AND POLICIES
61
When we have anything in
expedite, pull that first.
If a fixed-date item is due
within 30 days, pull it.
Pull standard-urgency items
until we reach our WIP limit.
Always be working on one
intangible item, if we have any.
Expedite:
Finished within
2 days 85% of
the time
Fixed-Date:
95% due-date
performance
Standard:
70th percentile:
15 days
Intangible:
10-20%
allocation