Software Technology Transfer

1
Practical aspects of data collection in industry

• Why is measurement so difficult to
  institutionalise?
• FiSMA cases and lessons learned

2
Measurement?
Software and system measurement is closely
related to, and essential part of Measurement
programs, Data collection and Benchmarking.
There may be measurement without any other
connections or any public indications like those
three above, but such is very difficult to
survey.
3
Is measurement important?

• Both SPICE and CMMI standards demand evidence
  from measurement at higher maturity and
  capability levels.
• Common sense tells us that we need measurement
  and data collection if we want to set realistic
  business goals and quantify improvement.

4
Finland and software measurement

• Finland is (in many ways) one of the top
  countries in the Measurement World.
• Activity in software project data collection is
  also an indicator of maturity level.

Where does the ISBSG data come? (2005)
per capita Finland and Australia are the most
efficient so far
5
ISBSG data collection histogram

• Getting better year by year
• 2350 projects by the end of 2003
• 4000 projects by today (2006)
• Data collection looks popular and important!

6
Experience data collection histogram

• 750 projects
• The best organisations at CMMI/SPICE level 4,
  several on level 3.
• SPI and project estimation training and
  consulting services are well established,
  benchmarking services available.

7
Success of measurement programs

• Of 610 measurement programs in place in 1998,
  only 140 survived the two-year mark as reported
  in data collected by Howard Rubin. Moreover, Dr.
  Rubin's data points since 1988 show a consistent
  78 or higher failure rate for measurement
  programs.

Dekkers and McQuaid, IEEE IT Pro, 2002
8
Introduction to FiSMA measurement methods and
data collection tools

• Experience Pro 3.1
• Functional Size Measurement and Project Delivery
  Rate (FSM PDR, FP h/FP)
• 21 productivity factors for development and 22
  productivity factors for maintenance (ND21
  MT22)
• Context variables (5-30) and other metrics
  details
• Data quality criteria 2005 (max 100 points)

9
FiSMA measurement principle

• Same metrics, same methods and same tools
  throughout the development life-cycle and
  maintenance.

10
Sample of FiSMA cases data collection
1990-2005
11
Company-99 story (template)

• Market position and competition
• Ownership and visible goals
• Measurement training, metrics specialist
  recruitment and RD activity
• Method and tool adoption approach
• Recognition of success, rewarding and career
  opportunities

12
Why do so many measurement programs fail?
Critical success factors of SPI
Source P.Willman
13
CSFs of FiSMA sample companies
14
SPI trinity
Improving measurement and estimation processes
Functional size measurement Effort measurement
rules Classification rules Situation analysis
method etc...
Experience Pro MS Office tools KnowledgePlan ISBSG
tools etc...
Measurement process owner? Top management
connection? Memberships in SMA-network? Support
and training? etc...
15
People skills

• Situational leadership
• Nonverbal language
• NLP
• Cultural aspects
• Management by walking around
• One minute management
• Statuses
• etc ...

16
Software Process Improvement Map
17
Finland is known as a technologically advanced
and exceptionally competitive country, where the
software and systems development is well managed.

Conclusions
... but many Finnish companies have still lot to
learn -Realistic goals what do they want to
reach? -Correct timing when to start measurement
program? -Best practices and methods what should
they measure?
Lets keep learning from each other! Thank you!