Ottawa Software Process Improvement Network (SPIN)

Report on meeting of April 15, 1999
Topic: Software Estimation
Presenter: Gorden B. Wright
Affiliation: Quality Assurance Manager
Software and Systems Integration Group
Science Applications International Corporation

NOTE: Presentation material was not available at meeting.

Mr. Wright has more than 20 years experience in the software development field, focusing on Project Management and Software Process Improvement. This presentation was a brief overview of the full day course Mr. Wright taught in Ottawa the next day.

Mr. Wright prescribes a basic 4 step process:
(1) Create a technical baseline: write things down, description  of software & hardware, architecture, acquisition strategy, assumptions and constraints; no estimate at this stage.
(2) Do a preliminary cost estimate.
(3) Do a final cost estimate - document for defensibility.
(4) Revise the estimate.

Sounds easy, doesn't it!

However some quotes from Mr. Wright:

"If you don't have historical data you won't do a good job of  estimation"
"You can't do a good job of estimating without doing a good job  of requirements management"
"Its a tough sell to get a metrics database going"
"Corporate database takes years"
"Get an independent review of your estimate"
"Use 2 tools, get 2 estimates, examine parameters, do it  manually, compare results"
"Its an iterative process"
"Document everything, keep an estimation file(s), define this in  your software process(es)'
"Keep basic data - total effort/size/time"
"Do a post project review after the project has completed"

A few more quotes about the barriers to doing good estimating, tracking:

"management only hears cost and schedules"
"no training on what should be tracked"
"plans not realistic'
"we adjust the work but not the plans"
"size estimates often ignored or misunderstood"
"requirements change constantly"
"responsibility for requirements management not clearly established"

The solution:

It all starts with Senior Management. Since it takes long to see the payoff, senior management must see improving estimation as a strategic investment. Mr. Wright recommends getting pilot projects started to demonstrate these techniques work. Then publicize the successes.

Marvin Doran

Quality Specialist Communications Committee
Nortel Networks Ottawa SPIN

John F. Wilson 613 785-0711 phone
Screenphone Application Design jf.wilson@bell.ca
Bell Canada - Network Architecture
160 Elgin St., Room 630 613 785-4999 fax
Ottawa, ON K1G 3J4