Managing Information Technology

where he stood and work out an alternative that he could
support. Watkins did not try to force a recommendation on
the committee; rather, she presented the problems in busi-
ness terms along with a number of possible alternatives.
Because the agenda was well organized and all the infor-
mation was in the hands of participants ahead of time, the
steering group meetings were quite effective, usually end-
ing before the scheduled hour was up.
Several of the steering group members were the
sponsors of other projects, and after the PAS steering
group meetings were finished, they would stay around
and discuss these projects and their departmental prob-
lems with Watkins. She was pleased that she was viewed
as a Disbursements Department colleague and not as an
outsider.

Project Planning

The Stradis methodology requires that the project director
estimate two costs at the end of each phase of the project: the
cost of completing the rest of the project and the cost of the
next phase. At the beginning, estimating the cost of the proj-
ect was mostly a matter of judgment and experience.
Watkins looked at it from several perspectives. First, she con-
sidered projects in her past experience that were of similar
size and complexity and used their costs to estimate what the
PAS system would cost. Then she broke the PAS project
down into its phases, did her best to estimate each phase, and
totaled up these costs. When she compared these two esti-
mates, they came out to be pretty close. Finally, she went
over the project and her reasoning with several experienced
project managers whose judgment she respected. This initial
estimate was not too meaningful, however, because the scope
of the project changed radically during the early stages.
Estimating the cost of the next phase requires that
the project director plan that phase in detail, and then that
plan is used to set the budget and to control the project.
According to Watkins:

The Stradis methodology provides an outline of all

the steps that you go through to produce the deliver-

ables of a stage. I would go through each step and

break it down into activities and then break down

each activity into tasks that I could assign to people.

I would estimate the time that would be required for

each task, consider the riskiness of that task, and

multiply my estimate by a suitable factor to take the

uncertainty into account. I would also ask the people

who were assigned the task what kind of effort they

felt it would require and would consult with experi-

enced people in the IS area. Finally, by multiplying

my final time estimate by the hourly rate for the

person assigned to the task I would get a cost esti-

mate for each task and add them all up to get a total

cost for the phase. Again I would go over this with

experienced project managers, and with Peter and

Ted, before making final adjustments.

Then I could start scheduling the tasks. I always

included the tasks assigned to user department people,

although I did not need them for controlling my

budget and many other project managers did not

bother with them. I wanted Peter and Ted and their

people to see where they fit into the project and how

their activities impacted the project schedule.

To help with the scheduling, Watkins used a tool called

Project Manager’s Workbench that included a PERT module

and a Gantt Chart module. With the possibility of time con-

straints and different staffing levels, she often had to develop

several different schedules, for discussion with Shaw and

Anderson and for presentation to the steering group.

Staffing the Project

In addition to Watkins, Arnold Johnson was assigned to

the project at the beginning. Johnson had worked for

Carter as a maintenance programmer for many years.

Carter valued him highly as a maintenance programmer

and therefore only assigned about 20 percent of Johnson’s

time to the PAS project. According to Watkins:

Arnold did not see any urgency in anything he did,

and being primarily assigned to maintenance, he

never had any commitment to our deadlines, and he

would not even warn me when he was going to miss

a deadline. When you are on a project plan that has

tasks that have to be done by specific times, every

person must be fully committed to the project, so the

project plan was always in flux if we depended on

him to get anything done.

Johnson had a detailed knowledge of the existing

CIMS system, and Watkins had planned for him to docu-

ment the logical flow of the 14,000 lines of spaghetti code

in the main program of the CIMS system. Watkins reported:

He knew where things were done in the existing

program, but he never knew why they were being

done. He would never write anything down, so the

only way to get information from him was verbally.

We eventually decided that the only way to use him

on the project was as a consultant and that we would

have an analyst interview him to document the

existing system.

438 Part III • Acquiring Information Systems