Upload
tushpak
View
214
Download
0
Embed Size (px)
Citation preview
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
1/105
Galorath / DCG New York Event October 2011
Dan Galorath
Copyright 2011 Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
2/105
Ite
m
Who What First
AnswerBG Introductions, logistics, etcBG Format o f meet ingDG Opening remarks, personal perspective on
software estimation, give a few examples of Galorath clients where implementing theestimation tool has paid off (15 minutes max, fewslides). Mention Ford, Euroclear
MH Opening remarks, personal perspective onsoftware estimation, give a few examples of DCGclients where implementing the centralizedestimation has paid off (15 minutes max, fewslides). Mention Ford, Euroclear
BG Questions sent in before the meeting (if any) DG + MHrespond as
pre-assignedby BG
BG Any Questions from the audience? DG + MHrespond as
requested byBG
BG If other questions exhausted and time remains,start working through these questions (DG & MHcan have prepared slides to help with question if needed):
BG About 20 minutes before end: How is it you
typically engage with a new customer and whatdoes that look like?
Important of change managementImportance of process review and changeQuick Start AssessmentCall to Action: stay and sign up
MH then DG
BG Why is software estimation so difficult and so ofteninaccurate?
MH
BG How can we make software estimation easier andmore accurate?
DG
BG What stops people from implementing softwareestimation best practices?
MH
BG What are some ways to get around the challenge s? DG
BG We get our subject-matter experts to do ourestimates surely they know best what it willtake?
MH
BG We are moving towards agile development so wewont need software estimating any more, will w e?
DG
BG Does Experience really matter in Estimation whenyou go Statistical?
MH
BG I have heard that a point estimate is a bad thing?Is that true and if so why?
DG
BG What are the best inputs to use for softwaredevelopment estimation? Function points?Something else?
MH thenhandoff toDG to talk
about what
SEER canhandle
BG We have our own estimation parameters built intoan excel spreadsheet that our SMEs use. Is thatOK?
DG
2011 Copyright Galorath Incorporated 2
Opening Remarks:
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
3/105
Full program Opening Remarks: Item Who What First Answer 1. BG Introductions, logistics, etc 2. BG Format of meeting 3. DG Opening remarks, personal perspective on software estimation, give a few examples of Galorath clients where implementing the
estimation tool has paid off (15 minutes max, few slides). Mention Ford, Euroclear
4. MH Opening remarks, personal perspective on software estimation, give a few examples of DCG clients where implementing the centralizedestimation has paid off (15 minutes max, few slides). Mention Ford, Euroclear
5. BG Questions sent in before the meeting (if any) DG + MH respond as pre-assigned by BG 6. BG Any Questions from the audience? DG + MH respond as requested by BG
7. BG If other questions exhausted and time remains, start working through these questions (DG & MH can have prepared slides to help withquestion if needed): a. BG About 20 minutes before end: How is it you typically engage with a new customer and what does that look like? Important of change management Importance of process review and change Quick Start Assessment Call to Action: stay and sign up MH then DG b. BG Why is software estimation so difficult and so often inaccurate? MH c. BG How can we make software estimation easier and more accurate? DG d. BG What stops people from implementing software estimation best practices? MH e. BG What are some ways to get around the challenges? DG
f. BG We get our subject-matter experts to do our estimates surely they know best what it will take? MH g. BG We are moving towards agile development so we wont need software estimating any more, will we? DG h. BG Does Experience really matter in Estimation when you go Statistical? MH i. BG I have heard that a point estimate is a bad thing? Is that true and if so why? DG j. BG What are the best inputs to use for software development estimation? Function points? Something else? MH then handoff
to DG to talk about what SEER can handle
k. BG We have our own estimation parameters built into an excel spreadsheet that our SMEs use. Is that OK? DG l. BG What is the ROI for buying a software estimation tool? DG m. BG What differentiates the different software estimating tools on the market? MH n. BG Is it better to have a centralized estimating team or to give everyone access to the tools? MH o. p. q. r. s.
2011 Copyright Galorath Incorporated 3
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
4/105
3. DG Opening remarks
How can we make software estimation easier andmore accurate?
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
5/105
Some of the most important businessdecisions about a project are made at the
time of minimum knowledge and maximum uncertainty .
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
6/105
Estimation Is Critical
Estimating is critical for all kinds of systems Yet many treat is as a second rate process
Everyone estimates. Just most get it wrong anddont have a process
Having a repeatable estimation process is critical toboth estimating AND to successful projects
Estimation and measurement go hand in hand
2009 Copyright Galorath Incorporated 6
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
7/105
Introduction: Software ProjectsMust Return Business Value
Economics is primarily a science of choice, andsoftware economics should provide methods foranalyzing the choices software projects must make.
Leon Levy Software projects should be based on choices that
provide the maximum business value to theorganization Eli Goldratt
Some say business value is not our problem While others generally need to perform benefit analysis
We need to start building systems that optimize the
business Make IT part of the solution
2009 Copyright Galorath Incorporated 7
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
8/105
An Estimate Defined
An e s t i m a t e is the most knowledgeable statement youcan make a t a p a r t i c u l a r p o i n t i n t i m e regarding:
Effort / Cost
Schedule Staffing
Risk
Reliability
Estimates more precise with progress A WELL FORMED ESTIMATE IS A
DISTRIBUTION
2011 Copyright Galorath Incorporated 8
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
9/105
What can a parametric model tellyou?
9 9
What islikely tohappen
Feel lucky?
Firm FixedPrice?
Understand the risk before you commit!
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
10/105
The Gap Between IT & Management Needs (Source: Fraunhofer)
2011 Copyright Galorath Incorporated 10
Financial
Customer Business Process
Innovation / Growth
Vision & Strategy
Goals
Data CollectionMeasurementMetrics
AnswersQuestionsGoal Attainment
Software Development & Maintenance
6
GQM
PSM
CoBIT
Business/OrganizationLevel
ITGovernance
ITGovernance
Customer
Focus
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
11/105
Who Uses SEER?
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
12/105
How?We created a global I.T. estimation service
We combined software function point & estimation services in one place.We combined hardware and software estimates into one processWe created a global community of estimators (Europe, AP, North America)We create estimates based on a common toolset calledSEER .
28-Oct-11
Slide 12
The complexities of rolling out an IT Estimation ServiceCopyright 2010 Ford Motor Company. All rights reserved.
Software Estimation
Hardware Estimation
Development
Cost(SEER-SEM)
InfrastructureCost
(SEER-IT)
Overall projectCost
Request for Estimate
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
13/105
When? Types of Estimate.
28-Oct-11
Slide 13
The complexities of rolling out an IT Estimation ServiceCopyright 2010 Ford Motor Company. All rights reserved.
From the Project Apollo:Portfolio Management
Release2 Module 2.3IT TCODefine Target
IT Vision andRoadmap
Set budgetallocationguidelines
Monitor App licati on
Portfolio ID &
Assess
RebalancePortfolio to
Create aMulti-Year ITPlan
PrioritizeCandidateand PlannedPrograms to
Optimize
Value
RefineBusiness
Case / ID&
Assess
Monitoringand Tracking Analysis Analysis
Update Application Ass et Bas e DesignDesignBuildBuildImplementImplement
KOKO
Identify Ideas andSelect Opportunities
ProgramPrioritization
AndSelection
Cycle Planning
Program/Project Execution
Define Business StrategyDefine IT Strategy
E5 E4
E1
Planning Estimates (likely requested by Consultancy group or Planners)E1 Initial estimate for budgeting. A finger in the air view of project costs.E2 Program/project funding estimate conducted. A refined estimate, used to reserve funding for a project.
E3 Program estimate used for Program approval, developed after detailed work has been completed to develop firm ben-efits and business case ( Long L ead Programs)
Project Estimates (likely requested by PM or lead BA)E4 Project deliverables estimate completed at the end of Analysis/Elaboration, when requirements are firmE5 Estimate/Function Point count based on completed deliverables. Feeds back into knowledge base.
Estimation
Points
Opportunityis created
and logged
Develop InitialBusiness Case for
Candidates
Projectis createdE3
E2
E1E2E3
E4E5
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
14/105
Why bother estimating? Who wins?Consultancy Group :
Consistent, reliable estimates will help you allocate budget efficiently.
Cycle Planning: what can be delivered within your budget constraintsSingle process (which this is already in place in the U.S.)
General Management:It is a key tool for IT Demand Management managing supply vs demandIt will enable us to analyse make vs. buy decisions.It drives Function Point counting, which in turn enables us to measure:
Quality (project defect rates), development costs per FP, support costs etc
All:It will reduce the reliance on SMEs for estimating project costs
It reduces complexity with no need to engage hw/sw/Architects separatelyIt is a no-cost service estimation team costs are not charged to your project.We all benefit from the experience of years of software development experience -which should improve your success rates on complex projects.It documents assumptions and allows re-use of estimates when project scopechanges, when a project gets deferred etc.
28-Oct-11
Slide 14
The complexities of rolling out an IT Estimation ServiceCopyright 2010 Ford Motor Company. All rights reserved.
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
15/105
15
Fo r int e rna l use
6. Step3: Budget estimation & SEER => assessment of ITproductivity
As of J anuary 2010:
ITPAO started to c onduct Function Pointcounting a t 3 different stages during the projec tlife cycle
At BC : independent budget assessmentAt end DSG: monitor evolution of
functionalitiesAt project end: historical DB build-up
SEER-SEM usage is extended to those steps:At end DSG: with the intention to measure the
impact of functional changes on budget (incase of off-shored p rojec ts)
At project end: with the intention to furthertarget budget estimation of future projectsba sed on some key characteristics (calibration)Function Point counting results shared with PLs
FPA/SEER rolled-out through project life cycle withthe objective to assess development productivity:
FPA/SEER usage along the life cycle of the
project will allow to identify patterns per projecttypes thus allowing for a more detailed monitoringof the projec ts.
Post-launch, a comprehensive FPA is performedto doc ument the main characteristics of theproject, build-up a historical database and mostimportantly, assess the factors of changeimpacting the p rojec t; input to improve ourdevelopment.
Overall, data collected and analysed will allowto better anticipate issues per project type andtherefore contain/ mitigate risks i.e. better controlproject budget spend.
Post-ProjectDeployTestBuildDesignInitiate (BC)Initiate (Pre-BC)Feasibilitystudy
Main objective : budgetestimation independent
assessment
Main obj ective :productivity analysis &estimation assessment
Main objecti ve : scopebaselining i n terms of FPs
Function point calculationbased on DetailedFunctional Design
Function point calculationbased on High Level
Functional Design
Function pointcalculation based on
final Detailed FunctionalDesign
2 years ago 1 year ago Today Future
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
16/105
16
Fo r int e rna l use
7. Step4: Budget estimation & SEER => support tooutsourcing contracts discussions
As of Q3 2010:
FPA / SEER will also be used to validate the budget based on FP assessmentprovided by the outsourcers in the c ontext of our outsourcing strategy: at the initiation of the projec t at the end of the design phase.
It provides an independent benchmark and adds transparency to thenegotiation process.
Post-ProjectDeployTestBuildDesignInitiate (BC)Initiate (Pre-BC)Feasibilitystudy
Main objective : budgetestimation independent
assessment
Main objective :productivity analysis &estimation assessment
Main objective : scopebaselining i n terms of FPs
Function point calculationbased on DetailedFunctional Design
Function point calculationbased on High Level
Functional Design
Function pointcalculation based on
final Detailed FunctionalDesign
Independent benchmarkused to support
the negotiation process
2 years ago 1 year ago Today Future
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
17/105
Beyond Estimation
Metrics Analysis & Benchmarking Integration to Project Planning& Portfolio Systems
Project Monitoring & Control
AND MO REEnterprise Shared Database
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
18/105
e. What are some ways to get aroundthe challenges? DG
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
19/105
19
10 Step Software Estimation Process:Consistent Processes = Reliable Estimates
1. EstablishEstimate Scope
2. Es tab lis h Tec hn ic alBaseline, GroundRules, Assump tions
3. Collect Data
4. Estimate and Validate
Software Size
5. Prepare
BaselineEstimates
7. Quantify Risks andRisk Analysis
6. Review, Verifyand ValidateEstimate
8. Generate aProject Plan
9. Document Estimateand LessonsLearned
10. Track ProjectThroughout
Development
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
20/105
High Level Overview
Determine & Quantify
Benefits overtime
Determinetotal
ownershipcosts against
schedule
Determine if project
worthwhile onits own or as
part of aportfolio
2009 Copyright Galorath Incorporated 20
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
21/105
g. We are moving towards agiledevelopment so we wont need softwareestimating any more, will we? DG
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
22/105
An Agile Approach to Planning
From: Agile Estimating and Planning by Mike Cohn1. Rapidly and reliably generating useful new capabilities
& knowledge
2. Flow of new capabilities & knowledge guides the work3. Plan for learning more than what end product is
4. Traditional projects emulate a 10K race You know where the finish line is
Try to get there as fast as possible
5. Agile projects are like a timed race see how far you can run in sixty minutes!
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
23/105
Why Agile Needs Estimation & PlanningAs Much or More Than Traditional
The 90% complete syndrome on Agile With just a little more time we can do so much more
When do you stop?
When do you kill an agile project that is failing withouteffort / schedule commitments
Cant do business case analysis if we dont know howlong & how much
Are you sure the project is agile & not justunplanned?
Lowest level feature planning by itself can makeprojects take longer
Velocity can be useful but may not forecast actualproject completion
2011 Copyright Galorath Incorporated 23
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
24/105
Software & IT Systems Are AboutBusiness Value
Stakeholder Needs
Agile
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
25/105
I have heard that a point estimate is a badthing? Is that true and if so why? DG
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
26/105
Human Nature Optimism in
EstimatesH B R A r t i c l e e x p l a i n s t h i s P h e n o m e n o n :
Problem: Humans seem hardwired to be optimists We routinely exaggerate benefits and discount costs
Delusions of Success: How Optimism UnderminesExecutives' Decisions (Source: HBR Articles | DanLovallo , Daniel Kahneman | Jul 01, 2003)
26
Solution: Temper with outside view
Viable estimation is temperingPast Measurement Results t raditional forecasting w/ statisticalParametrics Can help
Dont remove optimism, but balance optimism & realism
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
27/105
Contractors At Least Level 3 Would Be
Acceptable
Level0
Informal or noestimating
Manual effortestimatingwithout a
process
Manual effortestimatingwithout a
process
Level1
Direct TaskEstimationDirect TaskEstimation SpreadsheetsSpreadsheets
Ad HocProcess
Level2
FormalSizing(e.g.
functionpoints)
DirectTask
Estimation
DirectTask
Estimation
Simple model(Size *
Productivity)or informal
SEER Use
Simple model(Size *
Productivity)or informal
SEER Use
Somemeasureme
nt & analysis
InformalProcess
Level3
FormalSizing
RobustParametricestimation
(SEER)
RobustParametricestimation
(SEER)
Estimate vs.actual capture
FormalizedMultipleEstimateProcess
Rigorousmeasurement
& analysis
Parametricplanning &
Control
RiskManagement
Repeatableprocess
Level4
Formal sizing Repeatableprocess
Robustparametricestimating
(SEER)
Robustparametricestimating
(SEER)
Rigorousmeasurement
& analysis
Parametricestimation
with tracking& control
RiskManagement
Processimprovement
via lessonslearned
Level5
Formal sizing Repeatable
process
Robustparametric
estimating(SEER)
Robustparametric
estimating(SEER)
Rigorousmeasurement
& analysis
Parametricestimation
with tracking& control
Risk
Management
Continuousprocess
improvement
Why should we care? Maturity is related to estimate viabilityBetter estimation process more likely to be successful in
execution
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
28/105
j. BG What are the best inputs to
use for software development estimation?Function points? Something else? MHthen handoff to DG to talk about whatSEER can handle
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
29/105
SEER Handles Any Sizing FromStandard To Organization Specific
2011 Copyright Galorath Incorporated 29
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
30/105
SEER Provides Numerous Size Measures ANDAllows Organization Specific
2011 Copyright Galorath Incorporated 30
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
31/105
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
32/105
COTS Estimation (ExistingComponents)
32
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
33/105
Galoraths - Function Based Sizing
33
i f C d C A f
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
34/105
Lines of Code Can Be Accurate If Definitions Are Consistent
34
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
35/105
K. We have our own estimationparameters built into an excel spreadsheetthat our SMEs use. Is that OK? DG
2011 Copyright Galorath Incorporated
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
36/105
Basic Estimation Tribes (Adapted fromDCG/Galorath Webinar)
1. Napkins Ad-hoc, hero driven estimators past successes arelegendary napkin scribbling taken as gospel
2. Guts Feeling /experience oriented project managersyears of experience, good and badtrusted regardless of actual results
(tenure benefits of being a survivor)3. Spreadsheets Former Napkins and Guts translating tribal
knowledge onto spreadsheets Bestows mathematical accuracy and empirical integrity on home-
grown estimation algorithms
Tribal estimation knowledge can and does work However, it comes with high risk and cost
Rarely repeatable
Consistency is sporadic
Heroic energy is kept in reserve used to mitigate risk
Knowledge almost never institutionalized outside of personalknowledge and desktop PC files
Knowledge lost when heroes retire or leave
2009 Copyright Galorath Incorporated 36
B l i g R & S h d l I
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
37/105
Balancing Resources & Schedule IsA Science
For a given Size, Complexity and TechnologyMinimum Time
To Complete(Effort Increases
to Reduce Schedule )
Work ExpandsTo Fill Time
(Effort Increases
due to lack of pressure )
Minimum Time
Optimal Effort(Lower Effortfor LongerSchedule)
Calendar Time
E f f o r t
M o n
t h s
Effort Increasedue to Longer Schedule
S ft E ti ti B i M d l &
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
38/105
Software Estimation Basic Model & Associated Metrics
38
Start Finish
E f f o r t K
S i z e S t R e u s e D I T
S i z e S e ( w o r k u n i t s )
S i z e ( Ef f e c t i v e Se
& T o t a l St )
D e f e c t s C o u n t ( Q i Q r )
Ca l e n d a r T i m e
O n - g o i n g I t e r a t i o n s o f Effort ( A C W P o r Sp e n t )
Progress ( B CW P o r Ea r n e d Va l u e ) Defects ( Q i Q r )
Growth ( S g )
SoftwareDevelopment
Process
S t a f f a &
C o n s t r a i n t s
Ef f e c t i v e T e ch n o l o g y Ct e
People Process
Technology
S t a k e h o l d e r R e q u i r e m e n t
s D e l i v e r e d So f t w a r e
Effective complexity
D
Maintenance/
Block ChangeDevelopment
Process
D e v e lo p m e n t L e g a c y ,
M a i n t e n a n c e S p e c i f i c s & C o n s t r a i n t s
a n d / o r B l o c k Ch a n g e s
A s R e d e v e l o p m e n t
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
39/105
Poor Estimates Effect on Projects
Inaccurate estimates can reduce project success: Poor implementations
Critical processes dont scale
Emergency staffing
Cost overruns caused by underestimating project needs
Scope creep Forever changing project goals
Frustration Customer dissatisfaction
Cost overruns and missed schedules
Project Failures
Important project business decisions made early withminimum knowledge & maximum uncertainty
39
Why should we care: Poor estimates and plans area root cause of program failure
Impact of Individual Personnel and
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
40/105
Impact of Individual Personnel andTeam
2011 Copyright Galorath Incorporated 40
122%100%
75%55%
0%
20%
40%
60%80%
100%
120%
140%
Fully Engaged Engaged Somewhat Engaged Disengaged
P r o
d u c
t i v i t
y
L e v e
l
Individual Productivity By Level of Engagement
Personnel Differences Make Big Project
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
41/105
Personnel Differences Make Big ProjectDifferences (SEER Uses Team Effectiveness)
Level Characteristics Impact Actions Possible
FullyEngaged
Passionate about work & organization Will do whatever it takes to deliver results Feels like a true owner Delivers consistent, high quality results Finds innovative solutions to the toughest
problems Seen as role model & leader
122% Manager Action Cankeep about 100%fully engaged
Engaged Focused on their deliverables, project, andindividual responsibilities
Delivers good, solid performance Always does their fair share of work Works well with others and will help out
when asked
100% 1 IN 5 CANINCREASE TO FULLYENGAGED
SomewhatEngaged
Selective about where they put theirenergy
Spends a lot of time doing things that arenot helping customers or the organization
Delivers when they have to, or when youare watching
Does what it takes to get by
75% 1 IN 2 CANINCREASE TOENGAGED
Disengaged Only works when they have to, they reallywant to be doing something else
Results are NOT meeting the standard Has a We They perspective Negative about the organization, as well as
in their interactions with co-workers and
customers Master at distracting others
55% 1 IN 10 CANBECOME ENGAGED
2011 Copyright Galorath Incorporated 41
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
42/105
L. What is the ROI for buying a softwareestimation tool? DG
2011 Copyright Galorath Incorporated
Estimate Total Ownership Costs;
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
43/105
Estimate Total Ownership Costs;60+% Can Be Infrastructure & Services
Software Development
Software Maintenance
IT Infrastructure IT Services
2009 Copyright Galorath Incorporated 43
Impact of Individual Personnel and
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
44/105
Impact of Individual Personnel andTeam
2011 Copyright Galorath Incorporated 44
122%100%
75%55%
0%
20%
40%
60%80%
100%
120%
140%
Fully Engaged Engaged Somewhat Engaged Disengaged
P r o
d u c
t i v i t y
L e v e
l
Individual Productivity By Level of Engagement
Personnel Differences Make Big Project
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
45/105
Personnel Differences Make Big ProjectDifferences (SEER Uses Team Effectiveness)
Level Characteristics Impact Actions Possible
FullyEngaged
Passionate about work & organization Will do whatever it takes to deliver results Feels like a true owner Delivers consistent, high quality results Finds innovative solutions to the toughest
problems Seen as role model & leader
122% Manager Action Cankeep about 100%fully engaged
Engaged Focused on their deliverables, project, andindividual responsibilities
Delivers good, solid performance Always does their fair share of work Works well with others and will help out
when asked
100% 1 IN 5 CANINCREASE TO FULLYENGAGED
SomewhatEngaged
Selective about where they put theirenergy
Spends a lot of time doing things that arenot helping customers or the organization
Delivers when they have to, or when youare watching
Does what it takes to get by
75% 1 IN 2 CANINCREASE TOENGAGED
Disengaged Only works when they have to, they reallywant to be doing something else
Results are NOT meeting the standard Has a We They perspective Negative about the organization, as well as
in their interactions with co-workers and
customers Master at distracting others
55% 1 IN 10 CANBECOME ENGAGED
2011 Copyright Galorath Incorporated 45
anua st mates: uman easons orE (S f A l i & M
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
46/105
Error (Software Analysis & MeasurementCan Help)
Desire for credibility motivatesoverestimate behavior (80%probability?) So must spend all the time to be reliable
Better approach force 50% probability &
have buffer for overruns Technical pride causes underestimates Buy-in causes underestimates
46
An ROI Analysis of A New System:
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
47/105
y yShould We Fund This?
2011 Copyright Galorath Incorporated 47
Can we do better? Will stakeholders tolerate a
loss for 3 years? What is the risk?
Example of Analysis of Estimation
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
48/105
p yProcess & Tools Analysis
2011 Copyright Galorath Incorporated 48
Human Nature Optimism In
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
49/105
Human Nature Optimism In
SoftwareH B R A r t i c l e e x p l a i n s t h i s P h e n o m e n o n :
Problem: Humans seem hardwired to be optimists We routinely exaggerate benefits and discount costs
Delusions of Success: How Optimism UnderminesExecutives' Decisions (Source: HBR Articles | DanLovallo , Daniel Kahneman | Jul 01, 2003)
49
Solution: Temper with outside view
Software Analysis & Measurement can temper
Viable estimation can temper
Dont remove optimism, but balance optimism & realism
Software Projects Must Return
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
50/105
Business Value Economics is primarily a science of choice
software economics should provide methods for analyzingthe choices software projects must make. Leon Levy
Base choices on those providing the maximum businessvalue to the organization Eli Goldratt
Measurement and its uses such as estimating and defectanalysis help this science of choice
Some say business value is not our problem
While others generally need to perform benefit analysis We need to build systems that optimize the business
Make IT part of the solutionIf IT & measurement dont generate sufficient profitmoney will go elsewhere
Looking At Maintenance and
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
51/105
Technical Debt
2011 Copyright Galorath Incorporated 51
Conclusions
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
52/105
Conclusions
Viable estimation process & tools yield significant ROI Rigor in the process helps make the best decisions Viable estimation reduces the surprises management
so dislikes Standardized measures and tools provide:
Ability to benchmark
Should cost analysis both internally and with vendors Ability to quantify internal & external improvements
2011 Copyright Galorath Incorporated 52
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
53/105
DCG - Galorath Estimation Workshop
October 27, 2011
Why Do We Care About
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
54/105
Copyright 2010. David Consulting Group
. . . because your customers pay the bills.
. . . because selling a current customer somethingnew costs a lot less then getting a newcustomer.
. . . because in the long run ALL customers canchange who are servicing their needs.
. . . because happy customers promote yourorganization.
External Customer Satisfaction?
Why Do We Care About
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
55/105
Copyright 2010. David Consulting Group
. . . because your customers pay the bills whetherdirectly or through intra-department
relationships.. . . because in the long run internal customers can
change who are servicing their needs.
. . . because we all are supposed to be working forthe same organization.
. . . because happy customers lead to lessinfighting and quieter relationships thereforehigher value.
Internal Customer Satisfaction?
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
56/105
Copyright 2010. David Consulting Group
Do internal and external customers have
the same expectations for: Dates? Cost? Profit Margins? Resource Utilization? Quality?
Differing Expectations?
Do we need to predict how wewill perform on these topics?
How Are Customer Satisfaction and
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
57/105
Copyright 2010. David Consulting Group
Ask yourself if success is a result of planning orluck?
Estimation Related
If success is going to be more than luck DCGsuggests the following steps:
1. Estimate performance
2. Plan to deliver3. Monitor and measure performance4. Perform a retrospective when done
Customer Satisfaction is a suitcase that when unpackeddescribes many of the components that must be estimated,planned, monitored and measured.
B d i E i i Pl i
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
58/105
Copyright 2010. David Consulting Group
Budgeting v Estimation v Planning
Budgeting
Defines how much we have to spend and influencesscope
Tends to ignore the cone of uncertainty
Estimation
Rough or approximate size extent or nature
Focused by the cone of uncertainty
Planning
Definition of tasks and allocation of resources
Focused on the narrow part of the cone of uncertainty
Pl i O i d Ti i
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
59/105
Copyright 2010. David Consulting Group
Planning Onion and Timing
Budgeting
Estimating
Planning
Mike Cohns Planning Onion
Analogy
Parametric - Light
IntegrationPlanningStandup Meetings
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
60/105
When Is An Estimate A Commitment
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
61/105
Copyright 2010. David Consulting Group
When Is An Estimate A Commitment
An estimate stated as a single number is a
COMMITMENT if there is no understanding thatan estimate is a prediction therefore will vary.
Estimates are:
Price Commitment
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
62/105
Example Implementations
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
63/105
Copyright 2010. David Consulting Group
Ford, Detroi t Used FP Counters from DCG since 2005 for project metric
normalization Acquired SEER for Software Realized that FP counts are an excellent input to SEER Converted DCG consultants to central team for project
sizing/estimating
Euroclear Use Galorath to do estimates
Galorath use DCG to do FP counts
Example Implementations
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
64/105
Copyright 2010. David Consulting Group
DIFFICULT & INACCURATE?Estimating
So Why Doesnt This Stuff Always Work?
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
65/105
Copyright 2010. David Consulting Group
So Why Doesn t This Stuff Always Work?
Consistency of
Method
Self Knowledge
Uncertainty
Effective Estimating
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
66/105
Copyright 2010. David Consulting Group
Effective Estimating
The Software Engineering Institute (SEI)requirements for good estimating: Corporate historical database Structured processes for estimating product
size and reuse Mechanisms for extrapolating benchmark
characteristics of past projects Audit trails Integrity in dealing with dictated costs and
schedules
Data collection and feedback processesfoster correct data interpretation
What Defines Good
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
67/105
Copyright 2010. David Consulting Group
What Defines Good
Hitting the midpoint of the original range? Making your stakeholders happy? Assuring a correct profit margin?
A good estimate is an estimate that provides aclear enough view of the project reality to allowthe project leadership to make good decisionsabout how to control the project so that it hits itstarget. McConnell
Impact of Under and Over Estimating
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
68/105
Copyright 2010. David Consulting Group
Impact of Under and Over Estimating
Under and Over Estimation
100%100%
ParkinsonsLaw
PlanningErrors,Compression,
Defects,IncreasedRisk
NegativeImpact
Make It Smaller
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
69/105
Copyright 2010. David Consulting Group
Make It Smaller
Pressure comes from many angles. Internalpressure comes from natural optimism and aneed to please.External comes from business pressures.
EstimateWith most pressure beingfocused on reducing theit is not surprising that mostestimates are low.
Why Estimate Better
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
70/105
Copyright 2010. David Consulting Group
W y
Improved ability to provide status Higher quality (better behavior) Better coordination between teams Better long term budgeting
Early warning risk information Indicates we have mismatches between targets,estimate and commitments
Better than guessing
Accuracy v Precision
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
71/105
Copyright 2010. David Consulting Group
y
Accuracy describes the nearness of a measurement to thestandard or true value.
Precision describes how exact the number is.
Pi = 3 accurate to one digit but not precise Pi = 3.37882 precise to 6 digits but not accurate Pi = 3.14159 accurate and precise to 6 digits.
Wikipedia
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
72/105
Cone of Uncertainty / Estimation Funnel
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
73/105
Copyright 2010. David Consulting Group
y
+Estimate too
high 1.5x
2x
1,25x
.8x.67x
.5x
*Probability
.25x
4x
P1 P2 P3 P4 P5
Actual-
Estimate toolow
EstimationFunnel
Strategy
Knowledge transforms your ability to accurately
predict and control projects .
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
74/105
Copyright 2010. David Consulting Group
WHAT STOPS PEOPLE?Estimating
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
75/105
Estimation Flavors
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
76/105
Copyright 2010. David Consulting Group
Parametric
Decomposition/Recomposition J udgment
Hybrid
Methods
McConnells Count, Compute, Judge Advice
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
77/105
Copyright 2010. David Consulting Group
Count if possible Must be available earlier rather than later
Must be correlated to the scope of what you are estimating Compute when done counting
Convert what you have counted into an estimate using history . .. form somewhere
History has to relate to what you count
J udge as a last resort Only when you dont have data (pressure will be downward away
from reality)
Judgment
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
78/105
Copyright 2010. David Consulting Group
Estimation Methods based on expert opinions afterconsideration or deliberation Structured based defined methods (Delphi) Unstructured
Viability is very dependent on the expertise of those
involved. It is usually recommended that estimates aredeveloped by the people doing the work Effectiveness and accuracy are enhanced when multiple
estimates are developed and synthesized Common estimate weighting plans include 1/3/2 or 1/4/1
(H, M, L)
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
79/105
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
80/105
Copyright 2010. David Consulting Group
DOES EXPERIENCE REALLYMATTER?
Estimating
Historical Data
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
81/105
Copyright 2010. David Consulting Group
Yours, Ours, Mine? Industry, Organization, Project/Program
Historical data is important because:Past predicts the future!
Use the data to calibrate and refine based onperformance magnitude of relative error
Estimation Flavors
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
82/105
Copyright 2010. David Consulting Group
Parametric
Decomposition/Recomposition J udgment
Hybrid
Methods
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
83/105
Copyright 2010. David Consulting Group
BEST INPUTS?Estimating
A Basic Estimating Model
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
84/105
Copyright 2010. David Consulting Group
PROJECTSIZE XX
RISKFACTORSPROJECTCOMPLEXITY
DEFINITION CAPABILITY ESTIMATE
S c h
e d ul e
Effort CostsREQUIREMENT
Estimating Influences
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
85/105
Copyright 2010. David Consulting Group
PROJECTSIZE XX
RISKFACTORSPROJECTCOMPLEXITY
DEFINITION CAPABILITY ESTIMATE
S c h
e d ul e
Effort CostsREQUIREMENT
Basics: Inputs and Outputs
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
86/105
Copyright 2010. David Consulting Group
Inputs Sized Requirements Skills People Equipment
Estimate Outputs Effort Cost
All data should be leveraged when planning
Planning Outputs Schedules Resource Allocation
Basic Estimation Model
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
87/105
Copyright 2010. David Consulting Group
Functional Size Function Points Use Case Points Story Points Tee Shirt Sizing
Behavioral
Attributes Process Disciplines Sourcing Model Skills
Technical
Complexity Security Languages Architecture
Productivity and DurationSignatures
What How With What
Why Is Size Important
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
88/105
Copyright 2010. David Consulting Group
Team Size
RiskManagementImportance
Size: Internal v External Definitions
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
89/105
Copyright 2010. David Consulting Group
Organizational Specific
DefinitionsIndustry Defined
Modules,
Use Cases,Test Cases
Story
Points
Lines of
Code
Use Case
Points
Cosmic,
NESMA FPIFPUG Function PointsMark II
Size: Accuracy
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
90/105
Copyright 2010. David Consulting Group
Less Accurate More Accurate
Hours,Days
StoryPoints
Lines of Code
Use CasePoints
COSMICNESMA FP
IFPUG Function PointsMark II
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
91/105
Complexity
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
92/105
Copyright 2010. David Consulting Group
Language Architecture Technologies Logical and Mathematical Algorithms
Data Relationships Code Structure Performance Memory
Economies and Diseconomies of Scale
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
93/105
Copyright 2010. David Consulting Group
Diseconomies Team Size Project Size Process Maturity Architecture Team Cohesion
Economies Product Complexity Personal Capabilities
Diseconomies become morepronounced as projects get bigger.
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
94/105
Deciding Estimates based on riskMonte Carlo Analysis
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
95/105
Copyright 2010. David Consulting Group
Monte Carlo methods are a class of computational algorithms thatrely on repeated random sampling to compute their results.
Monte Carlo methods are often used in simulating physical and
mathematical systems. Monte Carlo simulation methods are especially useful in studyingsystems with a large number of coupled degrees of freedom
Monte Carlo methods are useful for modeling phenomena withsignificant uncertainty in inputs, such as the calculation of risk inbusiness.
It is a widely successful method in risk analysis when compared withalternative methods or human intuition.
Deciding Estimates based on riskMonte Carlo Method(s)
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
96/105
Copyright 2010. David Consulting Group
There is no single Monte Carlo method; instead, the term describesa large and widely-used class of approaches. However, theseapproaches tend to follow a particular pattern:
Define a domain of possible inputs. Generate inputs randomly from the domain using a certain specified probability
distribution. Perform a deterministic computation using the inputs. Aggregate the results of the individual computations into the final result.
For our purposes, we can consider two methods for Define adomain of possible inputs for our software estimates: User selects a range of possible values for an input parameter (explicitly or by
comparison) and specifies a likely distribution within that range (may default touniform) e.g. size may be uniformly distributed between 90FP and 110FP
(based on a count of 100FP) User specifies or system defaults to a distribution around a single selected value
(e.g. Size is normally distributed around 100FP)
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
97/105
Deciding Estimates based on riskMonte Carlo Method - Example
I l M C l h d ld d l
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
98/105
Copyright 2010. David Consulting Group
In a real Monte Carlo method, a program would use many random valuesbut for our purposes, lets just take a few values within a uniformlydistributed range that we are going to select.
Imagine that we have some prior experience estimating apple picking andwe have a sense that there will be some variation in our input values e.g.
Estimate Min Max
Number of Apple Trees 20 20 20
Number of Apples per Tree 100 20 150 Time to pick one basket 10 5 12
Number of Apples per basket 40 38 43
If we imagine that, with a uniform probability distribution for ease of manualcalculation, any of thesee values is equally likely, then there are 3 4 (= 81)possible values for total picking time. We can plot these.
Monte Carlo Method ExampleBasic Plot of all 81 points - doesnt help much
1000.00
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
99/105
Copyright 2010. David Consulting Group
0.00
100.00
200.00
300.00
400.00
500.00
600.00
700.00
800.00
900.00
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81
Monte Carlo Method ExampleStill happy with 500 minutes?
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
100/105
Copyright 2010. David Consulting Group
0
2
4
6
8
10
12
14
16
Time 0-99 Time 100-199Time 200-299Time 300-399Time 400-499Time 500-599Time 600-699Time 700-799Time 800-899Time 900-999
Probability Distribution
Monte Carlo Method ExampleHow about now? Is 65% confidence enough?
Cumulative Probability
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
101/105
Copyright 2010. David Consulting Group
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Time 0-99 Time 100-199
Time 200-299
Time 300-399
Time 400-499
Time 500-599
Time 600-699
Time 700-799
Time 800-899
Time 900-999
Cumulative Probability
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
102/105
Copyright 2010. David Consulting Group
AGILE?Integrating Estimation Techniques: Hybrid Agile Estimation
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
103/105
Translating Size Into An Estimate
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
104/105
Copyright 2010. David Consulting Group
Input: SizedBacklog Disaggregationand Resize
Planning Poker
(factor inattributes andtechnical )
And /or
EstimationWorksheet
Sprint Planning(Productivity or
Commitment
Driven)
Quick Overview Of Sprint Planning Session
ProductOwnerPrioritizes
7/28/2019 New York Galorath DCG Event Oct 2011 Combined Slides
105/105
Copyright 2010. David Consulting Group
Product Owner Prioritizes
Teams Sequence
Apply Estimates
Team Commits