CompuSys In the early 1980s, the manufacturing division of CompuSys, a large computer
company, began to develop an expert system called Config. Config was designed
to help the company’s sales representatives produce error-free configurations
prior to providing price quotations. Traditionally, configuration errors had
been causing significant problems for the firm’s sales representatives.
Hopes ran high among the project’s leaders, Tom Jones and George Smith,
who planned to base Config on one of their earlier successful projects called
Verifier. Jones and Smith invited a small group of sales representatives to join the
project team. Between November 1981 and January 1982 this group experimented
with a prototype of Config and registered a series of complaints. They pointed
out that Config did not integrate with the system that salespeople normally
used to prepare price quotations, that it was slow, and that it was not user
friendly. Jones and Smith promised to address these complaints and have Config operational
in sales offices during 1983. In 1982 the project team prepared a discounted
cash flow analysis that indicated a net present value of $43.9 million (20
percent discount rate) for the five-year period FY82– FY86. Operating
and development costs for the project were projected at $10.4 million between
FY82 and FY86. At the end of 1983, Config was still not operational in the firm’s sales
offices. Sales representatives involved in the project complained vigorously
that the engineers had not responded to their complaints. Yet engineers on
the team continued to be enthusiastic. The project team agreed to continue
with the project and move forward with a full-scale implementation of Config.
In 1985, the project engineers finally managed to link Config and the price
quotation system used by the sales representatives. However, those representatives
who used the system found it more time consuming to produce price quotes than
the representatives who did not use the system. Notably, a second financial analysis was undertaken in 1985, and it showed
the net present value of Config to be $55.7 million, discounted at 20 percent,
for the five-year period FY85–FY89. Retrospectively, CompuSys executives described the reactions of Tom Jones
and George Smith: When we have done surveys of people in the field, the information we got wasn’t
what Jones and Smith wanted to hear, but their denial has been so powerful.
Their response has been: “Wrong answer, we don’t like that answer.” They
never listen to negative feedback. When I approach Smith with negative feedback
from the field, he gets defensive about it. He says: “Well, that’s
hearsay.” So we have to go out and conduct additional surveys. Sometimes
he asks us to go back and survey the same individuals again, and even if the
information is still negative, Jones and Smith find some way to put a positive
spin on it. A third financial analysis conducted in 1987 indicated that the net present
value of Config was at least $41.1 million for the five-year period FY87–FY91. By 1990, Config had yet to catch on with the sales force. Many executives
outside of manufacturing were arguing that Config should be abandoned. The
following remarks were typical: I wrote three whistle-blowing memos to three vice presidents. I couldn’t
stop it. It was a sacred cow project. There were some very strong champions inside. My arguments fell on deaf ears. I pointed out that based on the usage patterns, I didn’t think anybody
would miss Config very much if we fumed it off tomorrow. If CompuSys has spent millions on this project, we have really missed the
boat. I argued that we should pull the plug on this effort immediately. The people responsible for developing Config are trying to breathe life into
something that should be allowed to die. We have proof today that Config is
not successful. It has failed miserably. The problem is nobody is willing to
kill it.
On the other side, one manager who had been on the project team since its
inception argued for its continuation, saying: “I don’t think we
want to get rid of Config. I don’t want to just throw out 10 years of
work.” In 1992, Tom Jones died, and George Smith left CompuSys. A new vice president
evaluated the Config project and concluded: “What are we doing? We’re
spending a lot of money on something that nobody really wants. Stop.” And
so the project was terminated. Subsequently, managers provided their impressions
about why Config had been adopted and why it continued. One manager stated: What I saw were Jones and Smith building upon a previous success with the
Verifier project and saying we can hit that home run again. I think it was
a combination of optimism which you could call undue or not and a sense that
this was a new technology that we were applying to this problem and that experimenting
with it could yield the results that we wanted even if we couldn’t see
them in front of us at the moment. So there was a kind of technological optimism. A second manager offered the following view: Even though we sometimes got a cold reception from Sales, it continued
to be funded anyway as an R&D project. As long as Tom Jones was in charge,
Config never got reviewed. It didn’t get into the official budget review
process until 1992. Senior management had developed so much faith in the technology
that they continued to fund it regardless of what the data said. Case Analysis Questions - Draw up a list of the behavioral features you perceive in the case of
Config, and indicate where in the case these features occur. If you perceive
there to be multiple occurrences of a particular feature, list each occurrence
separately.
- Discuss any parallels between the case of Config and examples described
in the chapter.
|