In the following year, little progress was made. Randy Pond, a director in manufacturing4 and eventual co-leader of the project, described the dilemma facing the functional areas in late 1993: We knew we were in trouble if we did not do something. Anything we did would just run
over the legacy systems we had in place. It turned into an effort to constantly band-aid our
existing systems. None of us were individually going to go out and buy a package The
disruption to the business for me to go to the board and say “Okay, manufacturing wants to
spend $5 or $6 million dollars to buy a package and by the way it will take a year or more to
get in . . .” was too much to justify. None of us was going to throw out the legacies and do
something big.
The systems replacement difficulties of functional areas perpetuated the deterioration of Cisco’s legacy environment. Incremental modification continued while the company sustained an 80% annual growth rate. Systems outages became routine. Product shortcomings exacerbated the difficulties of recovering from outages.
Finally, in January of 1994, Cisco’s legacy environment failed so dramatically that the
shortcomings of the existing systems could no longer be ignored. An unauthorized method for accessing the core application database—a workaround that was itself motivated by the inability of the system to perform—malfunctioned, corrupting Cisco’s central database. As a result, the company was largely shut down for two days.
Cisco’s struggle to recover from this major shutdown brought home the fact that the company’s
systems were on the brink of total failure. Solvik, Pond, and a number of other Cisco managers came
to the conclusion that the autonomous approach to systems replacement they had adopted was not
going to be sufficient. An alternative approach was needed. Solvik described what they did:
We said, “we can’t wait casually by while Order Entry, Finance, and Manufacturing go out
and make three separate decisions.” It would take too long to get those applications in place.
We needed to take faster action. At that point we got sponsorship from the SVP of
Manufacturing, Carl Redfield. He was with Digital before Cisco, in PC manufacturing. He
took the lead and said, “O.K., let’s get on with this.… let’s start from the manufacturing
perspective, and see if we can get the Order Entry and Financial groups in the company
interested in doing a single integrated replacement of all the applications, instead of taking a
longer time doing separate projects.” And so in February, about a month after the [company
shutdown], we went about putting together a team to do an investigation to replace the
application.
Redfield understood from previous large-scale implementation experiences at Digital how “monolithic” IT projects could take on lives of their own. He echoed Solvik’s concerns about project size and had strong views about how Cisco should approach a large implementation project.
I knew we wanted to do this quickly. We were not going to do a phased implementation, we would do it all at once. We were not going to allow a lot of customization either. There is a tendency in MRP systems5 for people to want the system to mirror their method of operation instead of retraining people to do things the way the system intended them. This takes a lot longer. Also, we wanted to create a schedule that was doable and make it a priority in the company as opposed to a second tier kind of effort.
Cisco’s management team realized that implementing to meet business needs would require heavy involvement from the business community. This could not be an IT-only initiative. It was critically important to get the very best people they could find. Solvik elaborated: “Our orientation in pulling people out of their jobs [to work on the project] was if it was easy then we were picking the wrong people. We pulled people out that the business absolutely did not want to give up.”
Consistent with the need for a strong Cisco team, the company would also need strong partners. Solvik and Redfield felt it was particularly important to work with an integration partner that could assist in both the selection and implementation of whichever solution the company chose. Great technical skills and business knowledge were a prerequisite. Solvik explained the choice of KPMG as the integration partner:
KPMG came in and saw an opportunity to really build a business around putting in these applications. They also saw this as kind of a defining opportunity, to work with us on this project. As opposed to some other firms that wanted to bring in a lot of “greenies,” KPMG was building a practice of people that were very experienced in the industry. For instance, the program manager that they put on the job, Mark Lee, had been director of IT for a company in Texas that had put in various parts of an ERP system.
With KPMG on board, the team of about 20 people turned to the software market with a multi-
pronged approach for identifying the best software packages. The team’s strategy was to build as
much knowledge as possible by leveraging the experiences of others. They asked large corporations
and the “Big Six” accounting firms what they knew. They also tapped research sources such as the
Gartner Group.6 By orienting the selection process to what people were actually using and
continuing to emphasize decision speed, Cisco narrowed the field to five packages within two days.
After a week of evaluating the packages at a high-level, the team decided on two prime candidates,
Oracle and another major player in the ERP market. Pond recalled that size was an issue in the
selection. “We decided that we should not put Cisco’s future in the hands of a company that was
significantly smaller than we were.”
The team spent 10 days writing a Request For Proposals (RFP) to send to the vendors. Vendors were given two weeks to respond. While vendors prepared their responses, the Cisco team continued its “due diligence” by visiting a series of reference clients offered by each vendor. Following Cisco’s analysis of the RFP responses, each vendor was invited in for a three-day software demonstration and asked to show how their package could meet Cisco’s information processing requirements. Cisco provided sample data, while vendors illustrated how key requirements were met (or not met) by the software.
Selection of Oracle was based on a variety of factors. Redfield described three of the major decision points:
First, this project was being driven pretty strongly by manufacturing and Oracle had a
better manufacturing capability than the other vendor. Second, they made a number of
promises regarding the long term development of functionality in the package.7 The other part of it was the flexibility offered by Oracle’s being close by.8
Cisco also had reason to believe that Oracle was particularly motivated to make the project a success.
Pond provided his impression of Oracle’s situation: “Oracle wanted this win badly. We ended up
getting a super deal. There are, however, a lot of strings attached. We do references, allow site visits and in general talk to many companies that are involved in making this decision.” The Cisco project would be the first major implementation of a new release of the Oracle ERP product. Oracle was touting the new version as having major improvements in support of manufacturing. A successful implementation at Cisco would launch the new release on a very favorable trajectory.
From inception to final selection the Cisco team had spent 75 days. The final choice was teambased. Solvik described how the decision was made and presented to the vendors:
The team internally made the choice and informed the vendors. There was no major process we had to go through with management to “approve” the selection. We just said “Oracle you won, [other vendor] you lost.” Then we went on to contract negotiations with Oracle and putting a proposal together for our board of directors. The focus immediately turned to issues of how long the project would take, and how much it would cost. The team decided “yes, we will do this and we ought to go forward with the project.” So now at the very end of April we were putting the whole plan together.