DatePlace
Presenter
© Copyright Total Metrics Pty Ltd
What makes a What makes a successfulsuccessfulMeasurement Program?Measurement Program?
- a case study
September 2008ISMA - WashingtonPam Morris
Total Metricswww.totalmetrics.com
© Copyright Total Metrics Pty Ltd
Presenter - Pam Morris
CEO - TOTAL METRICS o Certified IFPUG (IFPUG CPC, CFPS, CSMS - 3)
Consulting, Training Tools and Standardso Metrics Implementation, Function Point Analysiso SCOPE Project Sizing Software™
Committee Member of:o Australian Software Metrics Association (ASMA) – Executive (1991 - )o ISBSG – Vice President Executive (2000 - )o International Function Point User Group (IFPUG) (1993 - 2000)o COSMIC-FFP Core Committee – (1997 - )o International Standards Organisation (ISO) - WG12 (1993 – 2007)o Standards Australia - IT15 (1993 - )
© Copyright Total Metrics Pty Ltd
Has anything changed?
Risk of Failure of Measurement Programs
050
100150200250300350400450
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991
Year of Implementation
Number of
MeasurementPrograms
(USA)
Failed Programs Successful Programs
• “80% of all measurement programs fail “Source : Howard Ruben Associates 1994
Overview of Topics
• Background • Measurement Process• Lessons Learned• Critical Success Factors
Background
• Australian Government Department• Large Legacy Application - ~14,000 fps• Mid-range – Cool:Gen, Java• 60 developers• Initial Objective : Verify improvements gained
by Re-factoring activity
© Copyright Total Metrics Pty Ltd
Measurement Process– ISO/IEC 15939:2007
1. Establish and Sustain Measurement and Management Commitment• Management had clear stated objectives • 4 Year commitment• Buy in from CIO to Project Team Leaders
STEP 1
© Copyright Total Metrics Pty Ltd
2. Plan the Measurement Process• Workshops to agree:
– KRA, KPIs– Report Templates – Data Collection Templates– Tools
Metrics Consultant Resource:• 1 consultant• Effort = 22 days• Duration = 1 Calendar Month
Client Resource:• 4 Management• Effort = 1 ½ day workshops +
Review 2 Drafts • Duration = 1 Calendar Month
STEP 2
© Copyright Total Metrics Pty Ltd
Reporting Structures
√√Application CumulativeCostARLS Maintenance Intensity13
√√Project / 6 month snapshotCost / QualityARLS Rework Analysis - Detail12
√√Release / CumulativeCost / QualityARLS Rework Analysis – Summary11
√√Release / 6 month snapshotCost / QualityARLS Time Spent in Testing10
√√Project / 6 month snapshotQuality / CostARLS Development Stage Analysis9
√√Release / 6 month snapshotQualityARLS Analysis of Defects – by Source of Origin
8
√√Release / 6 month snapshotQualityARLS Analysis of Defects – by Severity
7
√√Project / 6 month snapshotQualityARLS Project Quality and Testing Effectiveness
6
√√Project / 6 month snapshotCost & QualityARLS Project Productivity and Quality5
Supplementary Reports
√√Application / CumulativeCost (Investment)ARLS Baseline Growth4
√√√Release / CumulativeQualityARLS Release Quality and Testing Effectiveness
3
√√Release / CumulativeCostARLS Productivity and Release Size2
√√Release / CumulativeCost & QualityARLS Productivity and Quality1
Main Reports
QC
M
ngment
TeamLeadersProjectB
oard
IT SteeringC
omm
ittee
NameNo
Target Audience
Report LevelKey Result Area
Report
Report Templates
• Each Report had agreed:– Purpose– Target Audience– Frequency / Level– Rules for Calculation– Description :
• How to read the report• What it was demonstrating• The types of decisions it would support
© Copyright Total Metrics Pty Ltd
Report TemplatesARLS Release Test Effectiveness
4 2 %
2 0 %2 4 %
9 %13 %
18 % 17%
1%
6 0 %
76 %
6 2 %
9 7%
8 6 % 8 6 % 8 7%8 2 %
9 0 %
3 8 %
6 4 %
8 2 %79 %79 %
70 %6 7%
77%
6 5%
3530
4
13
3
3429
75
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
External Benchmark (Total M
etricsIndustry Figures)
FEB 2005 (Size =345 FPs)
MAY 2005 (Size =951 FPs)
AUG
2005 (Size =1461 FPs)
NO
V 2005 (Size =1175 FPs)
FEB 2006 (Size =695 FPs)
MAY 2006 (Size =562 FPs)
AUG
2006 (Size =993 FPs)
NO
V 2006 (Size =741 FPs)
0
10
20
30
40
50
60
70
80
90
100D
efects(Weighted)/1000FPs in M
1 in Production
Effectiveness of Product Stability Testing (PST)Effectiveness of System TestingEffectiveness of Integration TestingRelease Defect Density for First Month in Production (Weighted and normalised for Release size)
Note: Testing Effectiveness compares the number of defects found at a particular stage of testing against how many defects w ere actually le
• Eg Description• This report shows the overall quality of the ARLS development process
since the degree to which defects are released into production are a good indication of the maturity of software development.
Data Collection Templates• 5 Base Measures and Tools Agreed:
– Functional Size (fps)• IFPUG 4.2• SCOPE Project Sizing Software™
– Effort (hours)• ISBSG Definitions Level 2 • NIKU™
– Defects (number)• Origin, severity • ISBSG Customised • Test Track Pro ™
– Duration (Calendar Days )• ISBSG Definitions• NIKU™ Rules for Calculation
– Full-time Equivalents (people)• ISBSG Definitions
© Copyright Total Metrics Pty Ltd
STEP 3
• Establish Baseline
Metrics Consultant Resource:• 1 consultant• Effort = 5 days• Duration = 1 Calendar week
Client Resource:• Project Teams • Effort = ? • Duration = 3 Calendar Months
Metrics Consultant Resource:• 1 consultant• Effort = 33 days• Duration = 2 Calendar Months
Client Resource:• 8 application experts• Effort = ~1/2 day each• Duration = 2 Calendar Months
• Ongoing Measurement– ~ 6 projects every 3 month Release (846fps)
3. Perform the Measurement Process
© Copyright Total Metrics Pty Ltd
3. Perform Measurement Process
• Analysis of the Results – 49 KPIs
Metrics Consultant Resource:• 1 consultant• Effort = 10 - 15 days• Duration = 3 Calendar weeks
Client Resource:• Management Reviews• Effort = 1 days • Duration = 1 Calendar day
Metrics Consultant Resource:• 1 consultant• Effort = 5 days• Duration = 1 Calendar week
• Reporting the Results– Benchmark Report (6 monthly) – 100 pages
Client Resource:• 1 Metrics Analyst• Effort = 10 days • Duration = 1 Calendar Month
© Copyright Total Metrics Pty Ltd
4. Feedback into Technical and Management Processes Release Level - Productivity and Quality Trends
17.616.918.0
11.6
15.2
12.6
21.922.9
29.0
74.7
3.4
34.0
12.9
3.6
30.235.1
0
5
10
15
20
25
FEB 2005 (Size =321FPs)
MAY 2005 (Size =951
FPs)
AUG
2005 (Size=1461 FPs)
NO
V 2005 (Size=1175 FPs)
FEB 2006 (Size =695FPs)
MAY 2006 (Size =562
FPs)
AUG
2006 (Size =993FPs)
NO
V 2006 (Size =741FPs)
Prod
uctiv
ity (H
Rs/
FP)
0
10
20
30
40
50
60
70
80
90
100 Defects (W
eighted) / 1000FPs in 1st Month in Production
PDR for Release in HRs/FP
Release Defect Density for First Month in Production (Weighted and normalised for Release size)
Trend for Defect Density across Releases
Trend for PDR across Releases
PDR Trend
Defect Density Trend
Defect Density Trend
STEP 4
• Product Quality
– Observations• Most defects originated in Build phase• Testing was introducing defects• Testing efficiency was below industry
standard• Time spent early life cycle was below
industry standard• Large variability between projects
4. Feedback into Technical and Management Processes
• Product Quality
– Improvements Introduced• Peer Reviews• Formal Unit Test process• Focus on System Testing• Formal Requirements Management and
Design Process
4. Feedback into Technical and Management Processes
4. Feedback into Technical and Management Processes
• Productivity– Observations
• Less productive than Industry• Small projects (<100fps):
– have lower productivity– Small projects behave unpredictably.
• Larger Projects (>250 fps) took longer users optimum 12 months
• FP size gave accurate early life cycle estimates
• Large variability between projects
Less Productive than Industry Median
81892845776842Number in sample
80.740.556.1Maximum Value
30.012.412.523.418.116.821.1Bottom 25% of
Productivity
14.46.79.110.112.112.2Median rate
6.53.76.87.5Top 25% of
productivity
1.80.92.7Minimum value
Case Tools4 GL ProjectsCool:GEN Nov-06Aug-06May-06Feb-06Nov-05Aug-05May-05Feb-05
Industry Values ( R10 - 2007 )Project Median PDR Comparison to Industry by Release
Position
© Copyright Total Metrics Pty Ltd
Small Project are more unpredictable
PDR and Project Size
Industry Median COOL GEN Project Size, 209, PDR = 9.1
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
13.0
14.0
15.0
16.0
17.0
18.0
19.0
20.0
0 100 200 300 400 500 600
Project Size (FPs)
PDR
As projects get larger i.e.. >180 PDR Spread gets less i.e.. PDR is more predictable and more
© Copyright Total Metrics Pty Ltd
FP Size has good correlation with effort
Effort Vs Functional Size (All Projects)
y = 8.6607x + 555.82R2 = 0.6668
0.0
1000.0
2000.0
3000.0
4000.0
5000.0
6000.0
7000.0
0 100 200 300 400 500 600 700Project Size (Function Points)
Effo
rt H
ours
Total Work Effort For ProjectLinear (Total Work Effort For Project)
© Copyright Total Metrics Pty Ltd
Estimated FP Size Produced Accurate Effort Estimates
DRIVES Release Effort Estimates Vs Actual Release (Project) Effort in Hours
FEB 20065,969
NOV 200514,702
MAY 20068,426
AUG 200611,862
NOV 200610,366
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
15000
16000
17000
18000
19000
20000
0 4
Rel
ease
Eff
ort H
ours
(D
evel
opm
ent)
Actual Effort (excludes Release Management) (+10% error bars)
Predicted Effort hours using DRIVES (Release Level) median PDR
2nd Benchmark 3rd Benchmark 4th Benchmark
© Copyright Total Metrics Pty Ltd
Only 33% of Projects delivering New functionality to the Business and Net Growth is decreasing
Release Functional Impact Analysis - Summary
880269
4
125
109347
402
687
341
426
218255
182
770
808525
477
307
020
0
4
26
0
Net Growth, 215
Net Growth, 109
Net Growth, 453
Net Growth, 334
Net Growth, 343
Net Growth, 835
Net Growth, 976
Net Growth, 400
0
200
400
600
800
1000
1200
1400
FEB 2005 MAY 2005 AUG 2005 NOV 2005 FEB 2006 MAY 2006 AUG 2006 NOV 2006
Tota
l Num
ber o
f Fun
ctio
n Po
ints
0
200
400
600
800
1,000
1,200
1,400
Net
Gro
wth
for R
elea
se (f
ps)
Fps Deleted
Fps Changed
Fps Added
Net Growth
Trend Line for Net Growth
Nett Grow th For Release = (Added Fps)+ (Changed After size - Changed Before Size) - (Deleted)
© Copyright Total Metrics Pty Ltd
STEP 5
• Metrics Review Workshop – 2 hours
Metrics Consultant Resource:• 1 consultant• Effort = 3 days• Duration = 1 Calendar week
Client Resource:• Metrics Analyst + Training• Effort = 5 days • Duration = 1 Calendar month
Metrics Consultant Resource:• 1 consultant• Effort = 2 hours• Duration = 1 Calendar day
• Implementing Changes– Data Collection and Recording
Client Resource:• 5 Management team• Effort = 1 day• Duration = 1 Calendar day
5. Evaluate Measurement
© Copyright Total Metrics Pty Ltd
Changes Introduced• Defects
– All defects now captured – early life cycle– Unit Testing defects now captured
accurately– Defects now allocated correctly to phase
☺4th Benchmark more defects being reported
• Effort– QC Effort now allocated to the project not
the Release overhead.☺ 4th Benchmark higher Project PDR being
reported
© Copyright Total Metrics Pty Ltd
More Defects being Reported
DRIVES - Total number of Defects Found in Release(Weighted normalised defects per 1000 function points)
591
791
626
824
433
771
304
180
0
100
200
300
400
500
600
700
800
900
1000
Def
ects
(Wei
ghte
d N
orm
alis
ed )
/100
0 FP
s
© Copyright Total Metrics Pty Ltd
Comparison ALL PROJECT PDR to Industry Medians
7.5
23.4
20.4
11.6
30.3
13.5
31.4
10.1
18.1
12.112.2
16.8
21.1
21.1
16.615.2
14.9
5
10
15
20
25
30
Benchmarking Release
Prod
uctiv
ity (H
Rs/
FP)
all P
roje
cts
in a
Rel
ease
Release Project MedianRelease Project AverageMedian for all Projects All Releases(ISBSG : Coolgen n=28) = 9.1 hrs/fp(ISBSG : 4GL n=89) = 6.7 hrs/fp(ISBSG : Case Tools n=81) = 14.4 hrs/fp
ARLS MEDIAN
Lower Project Productivity (higher PDR)Reported
© Copyright Total Metrics Pty Ltd
Critical Success Factors
• Formal Process– Started at beginning– Clear Stated objectives– Vision – long term commitment– Adequate Budget and Resources– Used skilled Metrics personnel– Invested in a baseline functional size
• Management– Realistic expectations– All levels interested, results are shared– Acts on the results– Open to change– Sees bad news as an opportunity– Measurement is viewed as important
News Flash - May 2008
• Two other IT Areas want what they have got!
• 20 other Applications want to be involved and get what the ARLS team are getting!
True Measure of Success!
Total Metrics Pty Ltd667 Burke RoadCamberwellVictoria 3124 Australia
Phone: +61 3 9882 7611Fax: +61 3 9882 7633
Total Metrics Pty Ltd667 Burke RoadCamberwellVictoria 3124 Australia
Phone: +61 3 9882 7611Fax: +61 3 9882 7633
Contact Details
Presentation on Total Metrics WWW site:WWW.totalmetrics.com