Oregon Embraces Customer-Driven Approach

By John Makulowich, Senior WriterTake a mainframe environment in a major state agency and demand from end users for a simple front-end system and no-nonsense answers to financial questions, and you quickly find yourself on the verge of a migraine-size problem seeking an imaginative IT solution.That's not far from the truth in Oregon, where the answer came in the form of a data warehouse/data mart that will allow users from 125 agencies, in their own individual operation areas, to gather information specific to their program needs. It is another case of mass customization on the state level. John Radford, administrator for the Oregon Controller's Division in Salem, said the impetus for the warehouse was the installation of a mainframe-based, statewide accounting system. It did not take long before concern with ease of access and reporting started to brew."Early on, we found three barriers: The canned reports were not as user-friendly or complete as our accountants or our field managers would have liked. That is, they just did not meet user needs. Second, maintaining the data in the mainframe and allowing access presented security problems. Third was the cost of letting users access the system thousands of times each day," he said.Radford did what he called literary and analytical research on the Internet to uncover what functional and technical solutions were available. And in what is becoming common practice among IT professionals, he contacted other states and users to see the extent to which mainframe systems were used and how they approached the problems he faced."We found a lot of cases like ours. ... That is, thousands of transactions per hour, and huge production databases for online transaction processing facing the issues of 'audit-ability,' security and control," Radford said.The approach was to seek a cost-effective solution that allowed access to production data and could be ported to the desktop. The answer was a data mart using a suite of decision support system tools from Brio Technology Inc., Palo Alto, Calif. "We realized we needed to present the data in a meaningful way that allowed easy access," Radford said. "We did not want to have everyone be a SQL programmer to get to their information. Our approach has always been customer-driven, simply because in our state, the customer pays the bill." For John McGinn, administrative services manager for the state Department of Agriculture, getting users access to the data was a two-way street. His department uses the data mart for financial reports."We had to teach them what counted as a good financial report, and about the financial data they were looking at. They, in turn, had to teach us their needs and the data they required to do their jobs," he said.XXXSPLITXXX-"Once past those thresholds, we now have first-line accounting technicians accessing the data mart to get their work done, preparing reports for payables and receivables and finding bills that have not been paid," he said. He said a new model is emerging for the traditional owners of the data. They are no longer guardians but rather facilitators of getting the information out."The traditional owners of the data have a level of stewardship they would like to maintain. In a distributed environment, that is harder and harder to do," McGinn said. "Clearly, the approach we are adopting in Oregon is changing people's jobs. No longer can you be a guardian of the data. You must be a facilitator, getting the data into an intelligible format for the end user down the road." The data mart is refreshed Sunday evenings and updated Monday mornings. Anyone in the state system connected to the state network with the proper security can make connectivity over the Internet and attach to the database. Radford's next step is to "Web enable" the data mart so functional professionals like accountants or budget officers can create scripted reports to run each week. They then can distribute the reports to staff via e-mail.Wayne Eckerson, vice president of technology services for the Data Warehousing Institute in Gaithersburg, Md., said offering a Web component is now a requirement for doing business in the data warehousing space."A lot more users are accessing and analyzing the data, and the Web is spurring that growth," Eckerson said. "In principle, it allows delivery of [digital satellite systems] and business intelligence to potentially 100 percent of knowledge workers. "The tools for sophisticated users are already developed. On the horizon is the development of tools for the masses. The next big move is personalized delivery of information on demand," he said.When the Oregon system was installed a year and a half ago, there were 12 users in the first month. Now there are 370 users. Agencies are getting more sophisticated, more and more users are being added and interest is growing exponentially, according to Radford.The data now available is mainly financial: budgets and budget reports, general ledgers, revenues and receivables. Radford's office is migrating payroll data and connecting the two databases."Our approach remains one of staying on the right side of the growth curve. We continue to be demand-led and user-led," he said. "Admittedly, other states may have an approach that is centrally controlled, following the model that if you build it, they will come. Our model is, the customer is in charge." Radford learned five important lessons from building Oregon's data warehouse: 1) Let the customer lead the way and make the investment decision. 2) Proceed slowly and don't go too far out on the technology limb. Keep the project well-managed and under control.3) Recognize you will never have enough resources to do training. Provide facilities or forums to share training and services in order to lower costs. Begin with the attitude that you will share knowledge. 4) Build flexibility into your approach. Try not to design too rigidly, and do not focus on one user base, for example, Mac and PC. The key focus is access, not the tool to get it.5) Many administrators fear putting tools in the hands of neophytes, believing they will misinterpret the data or misunderstand the results. To overcome that, focus on requiring users to learn the data before allowing them access. In that way, they gain some idea how data is organized.