"One more such victory..."

The Wall Street Journal (subscription required) ran a story today on the implementation of a new Oracle payroll system for Arizona State (there is an earlier story here from California State University). The approach taken was a bit different, or in the words of Emeritus Professor Colin Tully, a planned failure.

On the advice of Arizona State's IT department, instead of a designing and planning for a project that would take an originally estimated 4 years and cost $70 million, the University would instead try to do it for $30 million ($15 million for development, $15 million for maintenance over 5 years) in just 18 months. The reasoning was that since there would be operational glitches whether the project took the 4 years or 18 months, so what the hey, let's go for short approach and fix everything in the wash.

To hit their target date, the University increased the number of programmers and consultants on the project, as well as opted for a very "vanilla" payroll system (i.e., absolutely minimal customization) and if required, Arizona State business processes would be changed to conform to the payroll system requirements rather than vice versa. For instance, instead of being paid on the 15th and 30th of the month, employees would now be paid every other Friday.

Well, the University got exactly what it hoped for - a buggy payroll system in need of fixing.

The minor change in pay dates meant that 26 paychecks would be issued per year instead of 24, which has created recurring problems. There is a more detailed of explanation of why this seemingly simple change has caused recurring issues here. At least 700 out of 15,000 Arizona State faculty and staff have experienced paycheck problems.

Of course, the people are really responsible for all the "cost savings" have been the 700 Arizona State workers, who, like the LA Unified School District employees, paid the price in terms of incorrect or no paychecks. This hidden but very real sacrificial cost was not factored into the equation, and employee morale is not in the best of shape, to say the least. Affected Arizona State employees got an apology from management and a $25 give card for their trouble.

However, the new payroll system has been declared by one and all (or at least the Board of Trustees, University Management, the IT Department, and Oracle) as a major IT success for saving so much money. It is also being touted as the "wave of the future."

I am a bit puzzled, however, if the 4-year, $70 million originally proposed system was also a vanilla payroll system. If it wasn't, then I am a bit skeptical that this project can be declared a success, since it's an apple and oranges comparison.

Furthermore, using a vanilla approach to installing payroll systems to keep costs and schedule down is not a new idea by any stretch. Software vendors have been arguing that organizations should conform to their software for at least twenty years.

In addition, the project didn't seem to have a very robust risk management effort in place, given some of the reported errors. I guess you don't need to have one if the problems are expected.

I also wonder why they didn't try to do the project in 16 months or 14 months instead of 18 months, given their cost/benefit logic. Seems to me that by targeting say 20% of the faculty and staff for payroll problems, ASU could have saved millions more.

The situation reminds me some of what King Pyrrhus supposedly said after defeating the Romans, "One more such victory and we will be undone."


Risk Factor

IEEE Spectrum's risk analysis blog, featuring daily news, updates and analysis on computing and IT projects, software and systems failures, successes and innovations, security threats, and more.

Robert Charette
Spotsylvania, Va.
Willie D. Jones
New York City