Using TSP to Improve
Performance
Dan Burton
Software Engineering Institute
Carnegie Mellon University
Pittsburgh, PA 15213
Sponsored by the U.S. Department of Defense
© 2008 by Carnegie Mellon University
This material is approved for public release. Distribution is limited by the Software Engineering Institute to attendees.
November 2008
Using TSP to Improve Performance
Dan Burton has been with the Software Engineering
Institute (SEI) since 1991. He has been working on the
Personal Software Process (PSP) and Team Software
Process (TSP) initiatives, teaching and introducing the
PSP and TSP into organizations since 1996.
Before joining the SEI, he worked for Tartan, Inc., a small
Ada compiler developer, where he managed the
development of the first Ada compiler for a digital signal
processor (DSP).
He holds a Master of Science in electrical engineering from
the US Air Force Institute of Technology, and a Bachelor of
Science in electrical engineering from Carnegie Mellon
University.
© 2008 Carnegie Mellon University
2
Using TSP to Improve Performance
Topics
How does TSP work?
TSP Data
TSP and CMMI
TSP Results
© 2008 Carnegie Mellon University
3
Using TSP to Improve Performance
How does TSP work?
TSP achieves performance through
• Personal Software Process
Self-Directed
Teams
(PSP)
• Integrated measurement
framework
• Comprehensive quality
Comprehensive
Quality
Management
Coaching
TSP
performance
factors
management
• Self-directed teams
• Coaching
Integrated
Measurement
Framework
PSP
© 2008 Carnegie Mellon University
4
Using TSP to Improve Performance
Personal Software Process
The PSP is a process designed for individual use that applies to structured personal
tasks.
PSP builds the team member skills required for the TSP.
With PSP, developers learn to follow a defined process and how to measure, estimate,
plan, and track their work.
This leads to
•
better estimating, planning, and tracking
•
protection against over-commitment
•
a personal commitment to quality
•
personal involvement in process improvement
© 2008 Carnegie Mellon University
5
Using TSP to Improve Performance
Integrated Measurement Framework
TSP uses a powerful, flexible set of measures (most of which are
introduced in PSP) that provides the framework for
• instrumenting existing processes
• project management
• team process evaluation and improvement
• personal process evaluation and improvement
The TSP measurement framework is, as far as we know, unique.
© 2008 Carnegie Mellon University
6
Using TSP to Improve Performance
Comprehensive Quality Management
Quality management on TSP teams begins before the project starts, in
PSP training.
Individuals are responsible for the quality of their components.
The team is responsible for the quality of its deliverables.
Quality is always quantified using the integrated measurement
framework.
© 2008 Carnegie Mellon University
7
Using TSP to Improve Performance
Self-Directed Teams
TSP implements the concept of self-directed teams.
•
The team builds its own plans, negotiating trade-offs with management.
•
The team is committed, collectively and individually, to the plan because it is
their own.
•
Team members assume many well-defined responsibilities that a team
leader would otherwise normally handle.
The self-directed team does not replace or obsolete the team leader.
The team leader is still responsible
•
to management, for communicating progress and issues, and for results.
•
to the team, for communicating management issues, and for support.
•
for dealing with the unexpected.
© 2008 Carnegie Mellon University
8
Using TSP to Improve Performance
Coaching
World-class athletes and musicians (and the teams and orchestras that
they play with) all benefit from professional coaching.
The TSP team leader functions, in many respects, more like a coach
than a traditional manager.
However, the role of TSP Coach is usually assumed by someone other
than the team leader.
• TSP launches and relaunches
• TSP checkpoints
• fine points of exercising team roles
• help with applying PSP and TSP principles in unusual situations
To achieve full potential, a coach is needed.
© 2008 Carnegie Mellon University
9
Using TSP to Improve Performance
Principles
The following principles shaped the design of TSP
• Engineering is a team activity; self-directed teams do the best work.
• Operational processes provide the foundation for planning, tracking,
control, and improvement.
• Processes must be measured to be managed and improved.
• The cornerstone of a high-quality process is early defect removal
and the cornerstone of a satisfying product is early customer
involvement.
• Quality without numbers is just talk.
• Improving project performance will improve organizational
performance; improving individual performance will improve project
performance.
© 2008 Carnegie Mellon University
10
Using TSP to Improve Performance
Building High-Performance Teams
Team
Management
Team
Software
Process
Team
Building
Personal
Software
Process
Team
Member
Skills
Team communication
Team coordination
Project tracking
Risk analysis
Goal setting
Role assignment
Tailored team process
Detailed balanced plans
Process discipline
Performance measures
Estimating & planning skills
Quality management skills
TSP builds high-performance teams from the bottom-up
© 2008 Carnegie Mellon University
11
Using TSP to Improve Performance
The TSP Planning Process
TSP includes a structured
systematic process for
planning, the TSP Launch.
Business
and
technical
goals
Launch
Estimates, plans,
process, commitment
Re-launch
Each project cycle starts with a
TSP Launch or Re-launch to
plan the next cycle and the rest
of the project.
Development
Development
Development
phase
phase
phase
or cycle
or cycle
or cycle
Lessons, new
goals, new
requirements,
new risk, etc.
Phase or cycle
Postmortem
Each project cycle ends with a
postmortem that provides
qualitative and quantitative
feedback to guide the
remainder of the project.
Work products,
status, metrics,
results
Project
Postmortem
© 2008 Carnegie Mellon University
12
Using TSP to Improve Performance
The TSP Launch Process
1. Establish
Product and
Business
Goals
4. Build Overall
and
Next-Phase
Plans
2. Assign Roles
and Define
Team Goals
5. Develop
the Quality
Plan
3. Produce
Development
Strategy
6. Build Individual
and
Consolidated
Plans
7. Conduct
Risk
Assessment
9. Hold
Management
Review
8. Prepare
Management
Briefing and
Launch Report
Launch
Postmortem
The TSP launch process produces
necessary planning artifacts.
The most important outcome is a
committed team.
© 2008 Carnegie Mellon University
13
Using TSP to Improve Performance
The TSP Launch Products
Business needs
Management goals
Product
requirements
What?
How?
When?
Who?
How
well?
What
if?
Quality plan
Risk
evaluation
Team goals
Team strategy
Task hour plan
Team roles
Conceptual
design
Team process
Schedule plan
Task plans
Planned
products
Earned-value
plan
Detailed plans
Risk mitigation
plans
Alternate
plans
Size estimates
© 2008 Carnegie Mellon University
14
Using TSP to Improve Performance
Topics
How does TSP work?
TSP Data
TSP and CMMI
TSP Results
© 2008 Carnegie Mellon University
15
Using TSP to Improve Performance
The TSP Base Measures
You can’t manage what you can’t
measure.
To help teams know where they stand,
every TSP project gathers and uses
four base measures.
Status reports are generated from
these data.
Size
Effort
Defects
Schedule
Size, effort, and defects are measured
exactly the same way as in the PSP.
© 2008 Carnegie Mellon University
16
Using TSP to Improve Performance
What the TSP Base Measures Provide
Sample of Derived Measures
Derived Measures (continued)
Estimation accuracy (size and time)
Defect density
Prediction intervals (size and time)
Defect density by phase
Time in phase distribution
Defect removal rate by phase
Defect injection phase distribution
Defect removal leverage
Defect removal phase distribution
Review rates
Productivity
Process yield
%Reuse
Phase yield
%New Reusable
Failure cost of quality
Cost performance index
Appraisal cost of quality
Planned value
Appraisal/Failure COQ ratio
Earned value
Percent defect free
Predicted earned value
Defect removal profiles
Quality profile
Quality profile index
© 2008 Carnegie Mellon University
17
Using TSP to Improve Performance
The Overall Plan with Estimates and Planned Dates
© 2008 Carnegie Mellon University
18
Using TSP to Improve Performance
The SCHEDULE worksheet
© 2008 Carnegie Mellon University
19
Using TSP to Improve Performance
Quality Plan
The team completes the quality plan.
© 2008 Carnegie Mellon University
20
Using TSP to Improve Performance
Planned and Predicted End Dates
The Team worksheet provides the Planned and Predicted End Date
for each engineer. This is useful for review at weekly meetings to see
how balanced the overall workload is.
The Predicted End Dates are calculated based on the task hours
remaining, the planned schedule hours, and actual rate of earned
value
Note: two
engineers are
predicted to
finish late,
while the other
two are
predicted to be
early.
Predicted project completion
© 2008 Carnegie Mellon University
21
Using TSP to Improve Performance
WEEK -1
© 2008 Carnegie Mellon University
22
Using TSP to Improve Performance
WEEK -2
© 2008 Carnegie Mellon University
23
Using TSP to Improve Performance
Sample of TSP Analysis Charts -1
Cumulative Earned Value
Cumulative Planned and Actual Hours per Week
100.0
2500.0
90.0
2000.0
Earned Value
70.0
60.0
Cumulative Planned Value
50.0
Cumulative EV
40.0
Cumulative Predicted Earned Value
30.0
Cummulative Hours
80.0
20.0
1500.0
Cumulative Planned Hours
Cumulative Actual Hours
1000.0
500.0
10.0
Weeks
5/5/2003
4/21/2003
4/7/2003
3/24/2003
3/10/2003
2/24/2003
2/10/2003
1/27/2003
1/13/2003
12/30/2002
12/16/2002
12/2/2002
5/5/2003
4/7/2003
4/21/2003
3/24/2003
3/10/2003
2/24/2003
2/10/2003
1/27/2003
1/13/2003
12/30/2002
12/16/2002
0.0
12/2/2002
0.0
Weeks
Earned Value
Planned and Actual Hours per Week
14.0
200.0
180.0
12.0
160.0
140.0
8.0
Earned Value
6.0
Predicted Earned Value
120.0
Hours
Planned Value
Planned Hours
100.0
Actual Hours
80.0
60.0
4.0
40.0
2.0
20.0
Weeks
5/5/2003
4/21/2003
4/7/2003
3/24/2003
3/10/2003
2/24/2003
2/10/2003
1/27/2003
1/13/2003
12/30/2002
12/16/2002
12/2/2002
5/5/2003
4/21/2003
4/7/2003
3/24/2003
3/10/2003
2/24/2003
2/10/2003
1/27/2003
1/13/2003
12/30/2002
0.0
12/16/2002
0.0
12/2/2002
Earned Value
10.0
Weeks
© 2008 Carnegie Mellon University
24
Using TSP to Improve Performance
Sample of TSP Analysis Charts -2
Percent Defect Free
Defects Removed by Phase for Assembly SYSTEM
100.0%
500.0
450.0
Defects Removed by Phase
90.0%
Percent Defect Free
80.0%
70.0%
60.0%
50.0%
40.0%
400.0
350.0
300.0
Plan
250.0
Actual
200.0
150.0
100.0
50.0
30.0%
In System Test
In Acceptance Test
Te
st
Te
st
In
te
gr
od
e
C
Sy
st
em
Te
st
ni
t
at
io
n
tio
n
U
om
pi
le
C
In
sp
ec
ev
ie
w
od
e
R
C
od
e
C
In Product Life
Phase
Phase
Cumulative Defects Removed by Phase for Assembly SYSTEM
Defect Density by Phase for Assembly SYSTEM
1400
25.00
1200
800
Plan
600
Actual
Defects/KLOC
20.00
1000
15.00
Plan
Actual
10.00
400
5.00
200
Phase
In
te
gr
od
e
Te
st
Bu
i ld
an
d
Sy
st
em
Te
st
at
io
n
Te
st
ni
t
U
In
sp
ec
tio
n
om
pi
le
ev
ie
w
R
od
e
C
C
D
LD
R
LD
D
C
In
sp
ec
tio
n
0.00
ev
ie
w
0
Pl
an
R
e
ni
ng
Sy qui
re
st
e m me
n
Te
ts
R
EQ st
P
la
H
In
ig
sp n
hec
Le
In
ti o
te
g r vel
n
D
at
es
io
n
ig
T
n
H
LD es t
In Pla
sp
n
D
et
ai ec ti
le
on
d
D
e
D
Te LD sig
n
st
R
De ev
ie
v
w
D el o
pm
LD
e
In
s p nt
ec
tio
n
C
od Co
de
e
R
ev
ie
w
C
od Com
e
pi
In
Bu
sp le
i ld
ec
an
tio
d
n
U
In
te ni t T
gr
at e st
io
Sy n T
e
st
e m st
Te
st
Cumulative Defects Removed by Phase
an
d
In Build and
Integration Test
Bu
i ld
In Unit Test
H
R
0.0%
In Compile
LD
EQ
10.0%
In
sp
ec
ti o
n
In
sp
ec
tio
n
D
LD
R
ev
ie
D
w
LD
In
sp
ec
tio
n
0.0
20.0%
Phase
© 2008 Carnegie Mellon University
25
Using TSP to Improve Performance
Sample of TSP Analysis Charts -3
Process Yield for Assembly SYSTEM
300.00
100%
90%
250.00
80%
70%
200.00
60%
50%
40%
Yield
Plan
150.00
Actual
Plan
Actual
30%
20%
Te
st
cc
ep
ta
nc
e
A
Be
fo
re
%
ui
ld
%
B
Be
fo
re
C
%
od
e
Te
st
ys
te
m
S
Be
fo
re
%
an
d
Be
fo
re
U
C
Be
fo
re
tio
n
In
sp
ec
ev
ie
w
R
od
e
LD
D
C
In
sp
ec
tio
n
ev
ie
w
R
LD
D
In
sp
ec
tio
n
LD
H
R
EQ
In
sp
ec
ti o
n
%
0.00
In
te
gr
at
io
n
Te
st
pi
le
om
50.00
Te
st
10%
0%
100.00
ni
t
LOC/Hour
Inspection and Review Rates for Assembly SYSTEM
Phase
Phase
Phase Yields for Assembly SYSTEM
Quality Profile for Assembly SYSTEM
100%
Design/Code Time
1
90%
80%
0.8
70%
0.6
Plan
50%
Actual
40%
0.4
Design Review Time
Code Review Time
0.2
30%
20%
Plan
0
Actual
10%
Pl
a
eq nni
ng
ui
Sy
re
st
e m me
nt
s
Te
R
EQ st P
la
H
In
n
ig
s
pe
hLe
c
In
te
ve ti on
gr
at l De
io
si
n
Te gn
H
LD s t
Pl
In
an
sp
D
ec
et
ai
tio
le
n
d
D
es
D
ig
Te LD
n
R
st
De ev i
ve ew
l
D
LD opm
e
In
s p nt
ec
tio
n
C
C
od
od
e
e
R
ev
ie
w
C
C
od
om
e
pi
I
le
ns
Bu
pe
i ld
ct
an
io
n
U
d
ni
In
tT
te
gr
es
at
t
io
Sy n T
es
Ac ste
t
m
ce
Te
pt
an
st
ce
Te
st
0%
R
Yield
60%
Unit Test Ddefects/KLOC
Compile Defects/KLOC
Phase
© 2008 Carnegie Mellon University
26
Using TSP to Improve Performance
Sample of TSP Analysis Charts -4
Actual Time in Phase Percent for Assembly SYSTEM
Documentation
Requirements
REQ Inspection
REQ Inspection
Documentation
Requirements
Actual Defects Removed in Phase Percent for Assembly SYSTEM
Actual Defects Injected in Phase Percent for Assembly SYSTEM
Unit Test
Code
REQ Inspection
Requirements
Code Review
REQ Inspection
HLD Inspection
HLD Inspection
Requirements
DLD Review
High-Level Design
DLD Inspection
Detailed Design
Code Review
Code
Unit Test
High-Level Design
DLD Review
Detailed Design
DLD Inspection
© 2008 Carnegie Mellon University
27
Using TSP to Improve Performance
Topics
How does TSP work?
TSP Data
TSP and CMMI
TSP Results
© 2008 Carnegie Mellon University
28
Using TSP to Improve Performance
TSP and CMMI Are Complementary
CMMI is…
TSP is…
• a model of best practices
• an instance of best practices
• about “what” not “how-to”
• about “how-to” not “what”
• an improvement roadmap
• an improvement tool
• a capability benchmark
• a performance benchmark
© 2008 Carnegie Mellon University
29
Using TSP to Improve Performance
What Do CMMI and TSP Have in Common?
Five ideas from a broad array of fields provided the original foundation
for the CMMI.
• Planning, tracking, cost and schedule management
• Requirements definition and configuration control
• Process assessment
• Quality management and continuous improvement
• Evolutionary improvement
All of these same ideas found their way into the TSP except process
assessment.
© 2008 Carnegie Mellon University
30
Using TSP to Improve Performance
CMMI, TSP & PSP Relationship
CMMI - Builds
organizational
capability
TSP - Builds
quality products
on cost and
schedule
PSP - Builds
individual skill
and discipline
© 2008 Carnegie Mellon University
31
Using TSP to Improve Performance
When transitioning from a Project to a
Organizational focus
TSP can provide an organization with a common vocabulary to start
with in terms of how work should be performed
TSP will define many of the day-to-day processes, which allows teams
to focus on communication and the technical challenges before them
TSP provide a common data definitions and collection method to
address organizational data needs.
© 2008 Carnegie Mellon University
32
Using TSP to Improve Performance
Accelerated Process Improvement
TSP addresses or supports most of
the capabilities expected of a project
team through CMMI Level 5.
Level 5
Level 4
It provides a “starting point” for lowmaturity organizations.
CMMI Maturity
Level
Level 3
It provides a “next step” for highmaturity organizations.
Level 2
Directly Addressed
Supported
Three organizations have used TSP to
advance from ML1 to ML4 in only 2.5
years.
Partially Addressed
0%
50%
100%
Percentage of SPs
Not Addressed
Unrated
Source: Mapping TSP to CMMI, CMU/SEI-2004-TR-014
© 2008 Carnegie Mellon University
33
Using TSP to Improve Performance
CMMI and TSP Quality Performance
Average Defect Density of Delivered Software
8
7
6
5
4
3
2
1
0
Defects/KLOC
CMM
Level 1
CMM
Level 2
CMM
Level 3
CMM
Level 4
CMM
Level 5
TSP
7.5
6.24
4.73
2.28
1.05
0.06
Source: CMU/SEI-TR-2003-014
© 2008 Carnegie Mellon University
34
Using TSP to Improve Performance
Level 5 and Continuous Improvement
Project A
(TSP)
Size (KLOC)
Project B
82
151
31.8
43.0
$0.95K
$4.05K
Peer Review Exit Density
4.78
17.3
Delivered Defect Density
1.55
5.27
$78.K
$612K
3.7
14.6
Duration (months)
Normalized (per KLOC)
Integration / Acceptance Test Cost
Time to Accept (months)
Source: Northrop Grumman IT (a CMMI Level 5 organization)
http://www.dtic.mil/ndia/2003CMMI/kent.ppt
© 2008 Carnegie Mellon University
35
Using TSP to Improve Performance
TSP Advantages
Unlike most other engineering methods or processes, TSP
• leads to positive change in the behavior of individuals, teams,
and the organization through an embedded change management
strategy.
• improves performance, with quantifiable benefits, on first use.
• uses a project-by-project introduction strategy that pays for itself.
• is a disciplined and agile approach to engineering.
• has been applied to a broad range of application domains and is
scalable from very small to large projects and teams.
• is adapted to existing processes and methods rather than
replacing them.
© 2008 Carnegie Mellon University
36
Using TSP to Improve Performance
Topics
How does TSP work?
TSP Data
TSP and CMMI
TSP Results
© 2008 Carnegie Mellon University
37
Using TSP to Improve Performance
NAVAIR Benefits from TSP
Program
Size of
Program
Defect Density
AV JMPS
443 KSLOC
0.59
$2,177,169
P-3C
383 KSLOC
0.6
$1,478,243
Program
AVJMPS
H2.0
(Defects/KSLOC))
Schedule Variance
0.5% overrun
1.1% overrun
Cost Savings
from Reduced
Defects
Cost Variance
1.5% overrun
6.9% overrun
© 2008 Carnegie Mellon University
38
Using TSP to Improve Performance
Quality Benefits
TSP dramatically reduces the effort
and schedule for system test.
Most defects are removed during
reviews and inspections at a cost of 2
to 25 minutes per defect.
TSP System Test Performance Comparison w/Table
60%
50%
40%
30%
20%
10%
System test removal costs run from to
2 to 20 hours per defect.
0%
System Test % of Effort
System Test % of Schedule
TSP Min.
2%
8%
4%
TSP Avg.
4%
18%
17%
7%
25%
28%
40%
40%
50%
TSP Max.
These benefits continue after delivery.
Typical Projects
Failure COQ
• lower support costs
• satisfied customer
• better resource utilization
Source: CMU/SEI-TR-2003-014
© 2008 Carnegie Mellon University
39
Using TSP to Improve Performance
Reviews and Inspections Save Time
Xerox found that TSP quality management practices reduced the cost
of poor quality by finding and removing defects earlier when costs are
lower.
Defect Removal Time by Phase
1600
1405
1400
Minutes
1200
1000
800
600
400
200
5
22
2
25
32
Design
Review
Design
Inspect
Code
Review
Code
Inspect
Unit
Test
0
System
Test
Removal Phase
Source: Xerox
© 2008 Carnegie Mellon University
40
Using TSP to Improve Performance
Intuit Quality Improvement
TSP reduced defects found in system test by 60% over the previous
two releases of QuickBooks 2007 release.
Intuit has also recently reported a savings of $20M from a reduction in
customer support calls on QuickBooks 2007.
Source: Intuit
© 2008 Carnegie Mellon University
41
Using TSP to Improve Performance
Intuit Productivity Improvement
By putting a quality product into system test Intuit improved productivity
and reduced cost while delivering 33% more functionality than planned.
Source: Intuit
© 2008 Carnegie Mellon University
42
This is equivalent to a 57% increase
in productivity.
If you didn’t have such detailed
information, would you even know
that you had a problem? Or an
opportunity for such dramatic
improvement?
Task Hours
At Allied Signal average task hours
per developer per week were
improved from 9.6 hours to 15.1
hours through quiet time, process
documentation, more efficient
meetings, etc.
04
04/20
05/27/19
05/04/1998
05/11/1998
05/18/1998
06/25/1998
06/01/1998
06/08/1998
06/15/1998
06/22/1998
07/29/1998
07/06/1998
07/13/1998
07/20/1998
08/27/1998
08/03/1998
08/10/1998
08/17/1998
08/24/1998
09/31/1998
09/07/1998
09/14/1998
09/21/1998
10/28/1998
10/05/1998
10/12/1998
10/19/1998
11/26/1998
11/02/1998
11/09/1998
11/16/1998
11/23/1998
12/30/1998
12/07/1998
12/14/1998
12/21/1998
01/28/1998
01/04/1998
01/11/1998
01/18/1999
02/25/1999
02/01/1999
02/08/1999
02/15/1999
03/22/1999
03/01/1999
03/08/1999
03/15/1999
03/22/1999
04/29/1999
04/05/1999
04/12/1999
04/19/1999
05/26/1999
05/03/1999
05/10/1999
05/17/1999
05/24/1999
06/31/1999
06/07/1999
06/14/1999
06/21/1999
/28/1 99
/19999
99
Using TSP to Improve Performance
Improving Task Hours
Average Task Hours Per Week
18
Actual Task Hours per Week
16
15.1
14
12.6
13.3
12
10
9.6
8
6
4
Avg. Task Hours - Week
2
+57%
Avg. Task Hours - Phase
0
Source: Allied Signal
© 2008 Carnegie Mellon University
43
Using TSP to Improve Performance
Intuit Test Schedule Reduction
From data on over 40 TSP teams, Intuit has found that
• post code-complete effort is 8% instead of 33% of the project
• testing time is reduced from four months to one month
Non-TSP
TSP
Source: Intuit
Development
Development
Test
Test
© 2008 Carnegie Mellon University
44
Using TSP to Improve Performance
Microsoft Schedule Improvement
First-time TSP projects at Microsoft had a 10 times better mean
schedule error than non-TSP projects at Microsoft as reflected in the
following table.
Microsoft Schedule Results
Non-TSP Projects
TSP Projects
Released on Time
42%
66%
Average Days Late
25
6
Mean Schedule Error
10%
1%
Sample Size
80
15
Source: Microsoft
© 2008 Carnegie Mellon University
45
Using TSP to Improve Performance
Work-Life Balance
People are your most important resource.
Finding and retaining good people is critical to long-term success.
Intuit found that TSP improved work-life balance, a key factor in job
satisfaction.
Source: Intuit
© 2008 Carnegie Mellon University
46
Using TSP to Improve Performance
Intuit TSP Survey Results
Improved work-life balance with TSP is reflected in job satisfaction
surveys.
In my work group, we continually
improved our work processes
88
76
I feel encouraged to come up with new
and better ways of doing things
80
I like the kind of work I do
95
87
I have opportunities to improve my skills
82
60
TSP
90
82
I feel proud to work at Intuit
% Favorable
88
80
90
100
Non-TSP
“Engineers love it… Once they adopt it they can’t
imagine going back”
Source: Intuit
© 2008 Carnegie Mellon University
47
Using TSP to Improve Performance
Questions?
Dan Burton
dburton@sei.cmu.edu
412-268-1473
PSP/TSP website:
http://www.sei.cmu.edu/tsp
© 2008 Carnegie Mellon University
48
Using TSP to Improve Performance
Topics
What is PSP and TSP
Type of data you get from a TSP Tool
TSP and CMMI
TSP Results
The people side of change
TSP and the people side of change
© 2008 Carnegie Mellon University
49
Using TSP to Improve Performance
© 2008 Carnegie Mellon University
50
Using TSP to Improve Performance
The learning stages
Unconscious
What d’ya
mean “what
plan?”
What
plan?
Adapted from Hayes, 1993
Conscious
Look Ma
right on time!
Why am I
behind again?
Competent
Incompetent
Material obtained from Software Technology, Process & People (STPP)
© 2008 Carnegie Mellon University
51
Using TSP to Improve Performance
Normal Transition Scenarios
Self confidence
High
Gee, all those
forms!
Low
Hey, my AFR
reached 2.3!!!
I’ll watch TV instead
Time
Adapted from IMA, 1989
Material obtained from Software Technology, Process & People (STPP)
© 2008 Carnegie Mellon University
52
Using TSP to Improve Performance
Reaction to change
Intensity of
emotion
Over my dead body!
High
OK, can we talk about this?
Say what?
Hey, how about my job?
Oh, what the heck!
Low
Nah, It’ll just go away
Let me die
OK, I’ll have
another look
Time
Adapted from IMA, 1989
Material obtained from Software Technology, Process & People (STPP)
© 2008 Carnegie Mellon University
53
Using TSP to Improve Performance
Managing Change
• Create awareness of why the change is happening
• Build desire to support and participate in the change
• Provide the knowledge needed to change
• Demonstrate ability to implement new skills and behaviors
• Provide reinforcing environment to sustain the change
© 2008 Carnegie Mellon University
54
Using TSP to Improve Performance
What are the potential impacts of not managing change?
Managers may be unwilling to assign adequate resources to support the change.
Managers may create negative messages about the change to their peers and subordinates.
Employees may lose interest in their work resulting in productivity losses and negative
impacts with customers.
Employees may leave the organization
Unforeseen obstacles to the change seemingly appear from nowhere.
Funding is not made available to implement or sustain the change.
A sudden shift in priorities is observed with less emphasis on the current change.
© 2008 Carnegie Mellon University
55
Using TSP to Improve Performance
Topics
What is PSP and TSP
Type of data you get from a TSP Tool
TSP and CMMI
TSP Results
The people side of change
TSP and the people side of change
© 2008 Carnegie Mellon University
56
Using TSP to Improve Performance
Deploying TSP -1
Sprinkling a few TSP/PSP-trained engineers around an organization
will not produce noticeable results.
Installing TSP in an organization requires
• a team-based improvement focus
• careful planning
• senior management involvement and sponsorship
© 2008 Carnegie Mellon University
57
Using TSP to Improve Performance
Deploying TSP -2
•
•
•
TSP is introduced into an
organization on a project-by-project
or team-by-team basis.
TSP is more than a training program,
the new skills must be used in a team
setting soon after training ends.
Treat the deployment like a project.
– Set goals
– Assign responsibilities
TSP Introduction Steps
1.
Start by identifying external or internal resources
to lead the effort.
2.
If internal resources are selected, send them to
SEI training to become authorized instructors
and coaches.
3.
Train top management, then select two or three
initial projects or teams.
4.
Train the selected teams and their managers
then launch the teams.
5.
Monitor the projects and make adjustments as
needed.
6.
Expand the scope, selecting additional projects
or teams.
7.
Create or expand the pool of available SEIauthorized instructors and/or coaches.
8.
Repeat starting at step 4.
– Allocate resources
– Involve line management
– Plan and track
© 2008 Carnegie Mellon University
58
Using TSP to Improve Performance
Training Participants
Participant
Course
Notes
Executives and senior
management
TSP Executive Strategy Seminar
1 day + optional ½ day strategic planning session.
Middle and first-line
managers
Leading Development Teams
3 days
Software developers
PSP Fundamentals
5 days
PSP Advanced
5 days (optional)
PSP I
5 days
PSP II
5 days
(Alternative to PSP Fundamentals and
Advanced)
Other team members
Introduction to Personal Process
2.5 days
TSP Team Member Training
2.5 days
(Alternative to Introduction to Personal
Process)
Instructors
PSP Instructor Training
5 days
Pre-requisite training: PSP Fundamentals and PSP
Advanced or PSP I and PSP II
Coaches
TSP Coach Training
5 days
Pre-requisite training: PSP Fundamentals and PSP
Advanced or PSP I and PSP II
© 2008 Carnegie Mellon University
59
Using TSP to Improve Performance
Scale-Up Approach
The broader TSP introduction should repeat the pilot project steps
across the organization.
• Define each location’s goals and responsibilities.
• Work project by project.
• Build an experience base.
• Train managers and executives first.
• Train developers.
• Launch projects with TSP.
Treat each new introduction as a test.
• Use TSP methods.
• Gather data.
• Evaluate results.
• Adjust plans and methods as needed.
© 2008 Carnegie Mellon University
60
Using TSP to Improve Performance
The Scale-Up Team
Form a scale-up leadership team.
• TSP trained
• All required skills, disciplines, and constituencies
• A core of full-time members
Conduct a TSP launch of the scale-up effort.
• All team members present
• Management participate in opening and closing meetings
• Monitor and review like an engineering project
© 2008 Carnegie Mellon University
61
Using TSP to Improve Performance
Sustaining the TSP - 1
To sustain long-term TSP transition, incorporate the TSP transition
goals into the organization’s business system.
• Reward and recognition programs
• Salary and promotion reviews
• Career planning
• Bonus criteria
• High-potential management programs
When TSP success is recognized as a ticket to personal advancement,
sustained improvement will be assured.
© 2008 Carnegie Mellon University
62
Using TSP to Improve Performance
Sustaining the TSP - 2
As with any disciplined activity, the TSP needs continuing
reinforcement from management.
Establish regular quarterly management reviews.
• Review project performance.
• Examine key process measures.
• Establish and review benchmark comparisons.
Identify, recognize, and reward superior individual, team, and
management work.
© 2008 Carnegie Mellon University
63
Using TSP to Improve Performance
Maintain Continuing Oversight
While goals, responsibilities, and resources are essential, they are not
enough to sustain a major behavior change.
If senior management appears to lose interest in the TSP transition
effort
• progress will be slow
• the effort will not succeed
If you take all of these steps and show continuing interest, TSP
transition will be rapid and effective.
© 2008 Carnegie Mellon University
64