Data & Analysis | Feature

Effective BI on the Cheap

Ohio's Miami University is using software it already had on hand as part of a successful business intelligence program, one that's designed to help boost enrollment and contain costs.

• By Linda L. Briggs
• 06/17/10

Jump-started by a basic goal of measuring and containing costs, a project at Ohio's 200-year-old Miami University is using existing software tools and staff as part of a successful business intelligence (BI) rollout.

The project, now in its second summer and using only software the university already had on hand, focused overall on increasing summer enrollment. This year, with the advantage of some solid numbers from the BI solution, the focus is driving up enrollment to contain costs--a delicate balance of good academics and better financial management.

With about 20,000 undergraduates and four campuses, Miami University is a long-standing liberal arts college in southwest Ohio. Robert Frost once called the classic tree-lined main campus in Oxford "the most beautiful college there is."

In an effort to assess the overall cost of summer courses, Miami began a project in 2009 called the Summer Challenge. "The provost and the CFO were interested in determining the profitability of summer [courses]," according to Assistant Director of Business Intelligence Phyllis Wykoff. Administrators wanted a profit-loss statement at the course section level, Wykoff explained, which involved working with data that included registration information, the course charge versus the actual amount paid, and the overall course expense including faculty salary. Most of the information was in the university's ERP system, SunGard Higher Education Banner, but wasn't arranged to easily support that kind of analysis.

With no suite of BI tools on hand, Wykoff and her team elected to bring in a BI consulting company and to use software it already had. In just three months, and with a budget under six figures, Wykoff's team worked with the consultants to quickly create a data mart using Oracle Warehouse Builder. (The university already had an Oracle enterprise license, so there was no additional cost.) Microsoft's database management system, SQL Server, which the university already had, is used for reporting, along with Microsoft Analysis Services--a part of SQL Server--to build cubes (sets of prebuilt measures used to analyze data) that query the data mart. "We used what we had," Wykoff said of the low-budget project. "If necessity is the mother of invention, we had a huge necessity."

Other than a single person that the university had hired the previous year with specific BI experience--a data modeler--Wykoff and her staff had no one specializing in business intelligence and data warehousing. Instead, her newly formed BI team of seven, drawn from existing staff, worked closely with a consulting firm that specializes in BI and analytics in higher education, Ohio-based Incisive Analytics, for the first three months of the project. Wykoff, whose background is in application development, had worked closely on Miami's ERP system earlier, moving into her current BI position four years ago. "It's been drinking from a firehose ever since," she commented.

Wykoff said she realized that Incisive's chief architect, Christina Rouse, truly understood higher education when Wykoff mentioned needing a student count, and Rouse fired back, "Headcount or FTE?" In contrast, Wykoff said, "Most people say, 'What's so hard about that?'--and then you know they don't understand higher education and how significant counting can be."

During the three months of intense work with Incisive, Wykoff said she and her team learned about data modeling, what Microsoft Analysis Services can do, how to move data to and from the data mart, how to use analytics, and what users might be looking for in the data.

Also working to give input and feedback to Wykoff's team was a committee of various users. That committee, which has met weekly for a year and continues to meet and give input, includes representatives from budgeting, finance, the registrar, the bursar, and other academic personnel. As the group gains an understanding of just what BI solutions can do, Wykoff said, excitement is building. "There's starting to be a little line [of people] saying, "How can we get on the BI list and be next?'"

One lesson Wykoff offered as the project moves into its second year is this: Delivering the right data to end users through BI is an iterative process. "Once [users] see the data, they're going to ask a different set of questions [than the first time around], or realize some assumptions," she noted. That makes BI different from a more traditional software development project, in which user needs are researched, then a system is installed, modified, and left to run. In that sort of development process, the path and next steps are pre-determined. That's not the case in BI, Wykoff noted, where the team has had to establish that the project is directionally correct, then repeatedly turn to users for feedback at each step.

Wykoff also recommended building plenty of data audits and crosschecks as part of the project, as well as ensuring that IT staff can quickly explain the data's validity, not just its reliability. The same data delivered every day may be reliable, she explained, but it isn't necessarily correct. Building in plenty of checks helps correct that, as well as address initial skepticism and questions from users who aren't used to seeing the sort of data that a good BI solution can produce. "We've learned [to] make sure that we can quickly explain how we got [data] and how we validated it during the process," Wykoff said.

As the BI team's confidence increases, Wykoff said, auditing and cross-checking the data will be designed into every project they plan. "It may be as simple as counting people in the ERP system, counting them in the star [schema--part of the database design], counting them in the cube, and counting dollars to make sure they balance."

As useful as information from the BI solutions is proving to be, Wykoff pointed out, it isn't regarded as the deciding factor in whether or not to offer a course. Instead, it's simply another factor in decision-making on the value of various courses. "There are a lot of reasons to teach," Wykoff said, "and profitability is not the only criteria.... But at least when a dean or a department chair makes a decision to offer a course, they now understand whether they're making or losing money."