It was the 1990s. A boom time, at least in technology. And Stanford University was eager to dump its 20-year-old mainframe software and move to the latest financial and human-resources systems from Oracle and PeopleSoft.
The prestigious institution, located in the heart of Silicon Valley, enjoys access to the best and the brightest minds in technology. But Stanford administrators discovered that even they are not immune to the pain of delays and budget overruns that accompany complex enterprise resource planning (ERP) systems.
“Sometimes I look back and wonder if this wave of ERP software … wasn’t a collective hallucination,” says Stanford CIO Chris Handley, a former psychology instructor who joined Stanford from PricewaterhouseCoopers in 1999. “Just buying the software does not solve the problem. You have to change the institution, and that’s something Stanford struggled with.”
Stanford has spent more than seven years transferring its financial systems onto applications from Oracle called Oracle Financials. The project was supposed to be finished in 1999. The delay has been caused in part by Oracle itself, which helped Stanford customize the software so heavily—changing Oracle Financials to accommodate the way Stanford redistributes overhead costs across its grants, for instance—that together they broke continuity with future versions of the software, rendering portions of what they put in place unusable.
In 2000, Stanford got part of Oracle Financials running and set the remaining projects aside. The university then started installing PeopleSoft to manage student records and human-resources systems. The PeopleSoft projects were delivered on time. But faculty and students found them so disruptive—interfering with Stanford’s ability to issue correct paychecks and assign student housing, for example—that in 2002 the Faculty Senate called for independent oversight of Stanford’s information technology department.
Stanford has spent a lot of money on software and still has work to do. According to the university’s annual budget plans, the board of trustees since 1999 has been asked to approve $93.4 million in capital expenditures for applications and infrastructure . The trustees had approved $60 million in 1994 to overhaul Stanford’s entire administrative information systems, a project they expected would take five years, even though controller Susan Calandra says some of the projects in the original plan were never started.
What makes Stanford’s troubles all the more ironic is the institution’s proximity to Oracle and PeopleSoft. Stanford, with its gracious red-tiled roofs, and Oracle, with its gleaming metal-and-glass towers, sit just 10 miles apart along Route 101, the main thoroughfare through Silicon Valley. Three Stanford professors serve on Oracle’s board of directors, and CEO Larry Ellison has pledged $10 million to the university as director of the Ellison Medical Foundation. Across San Francisco Bay behind a range of hills is PeopleSoft, which has been fighting Oracle’s hostile takeover attempt for the last year.
But such close relationships have made no difference in Stanford’s struggles with Oracle’s and PeopleSoft’s software, according to Handley. In fact, the high-profile business battle between the vendors complicates matters. Each company’s software is known to interfere with the other’s, to the detriment of customers like Stanford.
For Handley, a big problem is that the software is designed to be used by public companies, not decentralized educational institutions. He notes that every ERP package he’s worked with—Oracle, PeopleSoft and SAP—has a single ship-to address in the purchasing module. That’s great for a company like IBM, which is organized around a central receiving unit, but tough for a 14,000-student research university like Stanford that receives packages directly in offices and labs.
Feeling the Pain
At times, enterprise resource planning vendors can be slow to recognize the pain their software can cause customers. PeopleSoft, the market leader in higher education with more than 700 customers, was sued in January by the state of Ohio after a botched installation at Cleveland State University. PeopleSoft had no comment on the lawsuit, but senior vice president Jim McGlothlin says the company is proud of its work at Stanford and notes that PeopleSoft software is more robust than it was in 2000. He says PeopleSoft “can now look customers in the eye and come up with customization figures of 5% or less.”
Oracle, meanwhile, says it has 70 higher education customers and is now rolling out a new student-administration module called the Oracle Student System that forms the core of its software. Oracle senior vice president Ron Police says that while it was Stanford’s choice to customize Oracle Financials, the university has since made “management changes” and is now happy with its progress installing Oracle.
By the time Handley was hired to oversee the Oracle and PeopleSoft projects, Stanford had decided to change itself rather than the software. This meant relinquishing forever the convenience of technically superior mainframe software customized for Stanford. “If you needed an extra three days in August to close the books, we could give you that,” recalls Dick Guertin, a Stanford developer who helped create the software in 1970. “Vendors [won’t] devote time to giving you those bells and whistles.”
But Handley contends that Stanford should distinguish itself through teaching and research, not unorthodox administrative processes that software vendors could standardize and ultimately make cheaper. He also stopped using vendors as consultants, selectively hiring outside firms to maintain what he calls “the appropriate distance” between the two roles.
So far, however, each installation of Oracle and PeopleSoft at Stanford has been different enough that it’s hard to establish best practices. The university must cope with what Handley calls “version upgrade gridlock”—installing Oracle v. 11.5.9 requires changing PeopleSoft v. 7.6, upgrading to PeopleSoft v. 8 requires changing Oracle v. 11.5.9, and so on.
Learning From the Past
In 2000, the information technology department got the general-ledger portion of Oracle running. They also started on PeopleSoft Student Administration, which includes Stanford’s student records system, but introduced it gradually in a series of nine phases spaced two or three months apart. This schedule turned out to be too tight for Stanford’s technology staff , which had three PeopleSoft projects going. But the software tended to creep up on users, who felt PeopleSoft was being forced on them and they had not been adequately prepared, Handley says.
Stanford changed tactics a year later. Officials installed PeopleSoft Human Resources Management System in only two phases. But the university found it had gone live with “new software and old processes,” Handley says. Student-aid checks, for example, printed in departmental rather than check-number order. After the printer went down during the first big printing, the staff had to figure out which checks were missing.
By May 2002, computer users at Stanford were getting upset. Minutes from a meeting of the Faculty Senate on Jan. 23, 2003, cite “a significant uprising within the user base,” along with a lack of appreciation for Stanford’s developers. In Handley’s view, the campus-wide Systems Governance Group that resulted to oversee information technology was a way of letting the university’s schools feel as though they had more control over the projects.
Nevertheless, in the summer of 2003 Stanford’s Systems Budget Group temporarily denied funding for all technology projects. “What bad thing is going to happen if we don’t do this?” information technology finance director Bill Clebsch remembers committee members asking about what he thought was a routine workflow project. He was told: “Unless you can show costs or benefits to the institution, we can’t keep funding this stuff.”
Stanford’s most stressful experience still lay ahead. This time it was with Oracle Financials, which finally went live on Sept. 2, 2003. In what is known in information technology circles as a “big bang,” roughly 4,000 users went live at once. That way everybody could move simultaneously to the new chart of accounts, which records the account names and numbers used in Oracle General Ledger.
The campus spent months trying to prepare for potential problems, such as the need to translate account numbers into the format accepted by Oracle. Still, transactions were delayed, data was missing and system performance was slow. Handley says the experience exposed weaknesses in project management because the team missed deadlines and “got swept away in all innocence” into supporting the software without enhancing the code. Calandra says the central offices—accounts payable, procurement, and so on—were learning the software along with the schools. When the schools bypassed the overwhelmed help desk and called for advice on processing transactions, they got inconsistent answers.
A New Curriculum
During the last three months, technology projects are going more smoothly at Stanford, according to Handley. Although the campus is still adapting its processes to Oracle, the “behavioral changes” required to use PeopleSoft are complete. In addition, an upgrade to Oracle v. 11.5.9 in April finished a day early because the project manager was assigned to Calandra, who could tell technology staff exactly what the controller’s office needed. “By September, we should have most of the functionality people would have expected [from Oracle] in the beginning,” Handley says.
Now faced with budget cuts and layoffs, Stanford’s information technology department has successfully sent coding and maintenance work to outsourcing firms in India, which are helping with Oracle report writing and an upgrade to PeopleSoft v. 8. Jobs that require deeper knowledge of Stanford, such as writing specifications, have been kept at home.
Communication has improved between the information technology and budget departments. In anticipation of what Clebsch calls the budget group’s “edict,” the technology department approached Chip Gliedman, an analyst at Forrester Research, who helped Stanford develop a system for evaluating projects called Total Economic Impact (TEI).
Corporate customers have been using the system for six or seven years, Gliedman says, but it needed modifying for higher education, which uses different metrics, such as the cost of attracting an application for enrollment, to measure success. So before the technology department submits a project for funding, the department must now calculate costs (how long will it take the staff to learn the new software?), benefits (how much time will the staff ultimately save?) and risks (what if the staff wants more features and the provost gives in?) over a five-year period.
No project approved under the new system is yet finished, and the process is still being tweaked. Calandra, who sits on the budget group, says TEI does not address how many things people can work on at one time or how much change the campus can absorb. She has also asked the information technology department to simplify its presentations.
But all agree that TEI has been very helpful in forcing the university to find and quantify costs and benefits. If one administrative support person working with either Oracle or PeopleSoft can save five minutes per transaction, the time will add up and that person can be free to do something else.
And should Oracle succeed in acquiring PeopleSoft and force Stanford to change its technology, TEI may prove critical. That’s because the work that Stanford has gone through to get the software functioning would then have to be done again.
Stanford Base Case
Headquarters: Stanford, CA 94305
Phone: (650) 723-2300
Business/Purpose: Teaching and research
Size: 6,654 undergraduate, 7,800 graduate students enrolled in 2003
Chief Information Officer: Chris Handley
Challenges: Replace a mainframe system with Oracle software for financial applications and PeopleSoft for human-resources applications.
Baseline Goals:
- Finish adapting financial systems to Oracle software by September, after five-year delay.
- Maintain quality of information system operations, amid budget cuts of 5% to 10% per year.
- Cut annual software ownership costs, which run up to 18% of purchase price, by outsourcing and thus licensing fewer modules.
