As colleges and universities are increasingly using data to support critical decisions, they are also grappling with ways to address both the cultural and capacity barriers to integrating analytics. Although there is evidence that both large and small institutions alike are turning to data and analytics to inform their decision-making, some smaller colleges may lack the technical talent, infrastructure, and budget to harness these resources.
Yet Hamilton College, a private liberal arts college in Clinton, N.Y., has found an approach that works for the small institution serving nearly 2,000 students. In 2017, at the urging of its president and trustees, the college overcame some of the initial obstacles often faced by institutions when tackling big data.
“Our trustees and board are used to having analytics available to them,” explains Michael Sprague, director, business intelligence and web services, Hamilton College. “They provided support and encouragement to investigate what was possible because they thought it would be very useful for decision-making.”
Hungry for detailed information, leaders at the top pushed for more—and better—analytics in two key areas:
Students’ well-being and success. “What we are seeing across the country is that students are experiencing more anxiety and depression and struggling with mental health,” says Terry Martinez, vice president and dean of students. By identifying the right metrics, she hopes to define Hamilton’s student success model and foster the connections to activities, internships, volunteer opportunities, and job offers that often lead to better mental health.
She adds that retention isn’t part of the problem for Hamilton College. “We have a wonderful retention rate,” Martinez says. “Many other small schools are taking a look at retention as a measure of success, but we have a 96 percent first-year to second-year retention rate and a 94 percent overall retention rate. Retention is not the issue. We need to look at different measures of success.”
Alumni/donor preferences and motivations. “We have 1,850 students, which means we don’t have a huge number of alumni,” says Lori Dennison, vice president for advancement. “We weren’t coeducational until 1978, and our class size wasn’t at the 400 [students] level until the ’70s. Because we were so small for so long, we tended to know everybody. Then 10 years ago, class size began inching up toward 500 [students]. We started the analytics program to understand this generation, which we didn’t know as well as we needed or wanted to.”
Starting From Scratch
In spring 2017, a 10–member working group explored the possibility of putting together a seamless business intelligence platform. Members began by talking with their peers at other institutions, evaluating their needs, and examining potential hurdles.
“We investigated what we needed; came up with a proposal; and went to the president, vice presidents, and trustees for approval,” Sprague says. “The initial proposal—without salaries—was $150,000, which included consulting and the software services necessary to use cloud technology.”
A data governance firm provided software and advice for setting up the data warehouse structure. Another consultant helped the school set up an initial proof of concept, provided training, and recommended appropriate cloud technology.
Gordon Hewitt, assistant dean of faculty for institutional research and assessment, explains that Hamilton also used outside support in developing common data definitions for its campus. “This involves listing the data variables on campus, breaking down data to basic elements, and building up definitions to ensure they are defined correctly; we take into account decades of different uses by different offices and divisions so that we have a common language and definition for each data point, rather than having several definitions across campus,” he says.
Sprague admits getting through the governance process for a project can be tedious. “It took a lot longer than we expected,” he says. “Once we got some experience, we were able to move more quickly. Our first project, which involved donor giving, took about six months to get to conclusion and opened the door to future projects.”
With implementation of the new platform, Hewitt says, users have more flexibility in data extraction and more refined data at their fingertips. “In the past, we put up PowerPoint slides on data. With the new data warehouse and Tableau reporting system, we can do interactive reports and filter data by different demographics. It gives a wide range of people, all the way down to the casual user, more interactivity and access to data.”
An Agile Approach
Both Sprague and Hewitt credit success thus far to two early decisions: choosing cloud computing and an agile approach. “Cloud computing is really the only way we could have done it,” Sprague says. “For us to move large volumes of data between systems without the cloud, we would probably need a million-dollar-plus infrastructure.”
The advantage of a cloud solution, he says, is that institutions with limited budgets can pay as they go. “For our first project, we weren’t using that many computing resources,” Sprague says. “We were able to use the same tools that a massive research university would use, but we were able to do it at a very low cost. We didn’t have to buy the infrastructure. We just had to buy time on Amazon infrastructure, which was key from a technology perspective.” He estimates that current cloud services with Amazon.com Inc. cost about $25,000 a year, and the selected integration product is another $30,000 annually.
According to Sprague, an agile approach—or addressing only the need that has been identified—fits perfectly with cloud computing because both strategies help the college only spend the time, and money, it needs. “Take, for example, the dashboard that shows cash giving,” he says. “We looked at what the need was for the dashboard and identified the data points. We built out the datasets and dashboard just using those data points. If something changes, we can add or delete data points, and we can address the need on a small scale.
“As a small school without a lot of staff, it’s better for us to build dashboards a piece at a time,” Sprague says. “It takes us less time to get to a product, and we can shift quickly to something else or make adjustments. Success is quicker.”
Without this agile approach, IT and IR could spend a year on a specific project for one Hamilton department, Sprague says. “By taking the agile approach, we might do a project for advancement that takes a short amount of time and do another for admissions. We can build out our data warehouse as we go through the process.”
The data governance committee, which has a representative from each of the seven campus divisions as well as from IT and IR, determines the priority of projects.
“Right now, we are looking for a dashboard on student success,” Hewitt says. “There are many definitions of student success and many different buckets of data can be used to measure it. Instead of trying to get every single variable defined and every data point on the dashboard, we picked five and are developing a dashboard based on those five. Once those five get defined, implemented, and used by the vice presidents, we will go to another five or 10 buckets. We have to take intense, data-heavy topics and chop them into little bits.”
The More You Know
With a couple years of progress completed, Hamilton’s analytics trailblazers offer this advice to their peers in similar small institutions:
Take manageable steps. “Definitely don’t bite off more than you can chew,” Hewitt says. “Take small steps at a time, don’t promise more than you can deliver, keep deliverables within reason, and don’t stretch the resources with quantity so that quality suffers.”
He recommends that institutions go through multiple iterations and improve upon each result. “Small colleges don’t have the capacity and resources to take a big bite out of such a large project at once,” he says. “Building something huge at first would take forever. With incremental progress, we can evaluate and redo if we have to.”
Garner leadership support. Sprague believes that trustees and senior staff must give preference to the analytics effort, even if it means that department projects requiring IT and IR assistance are relegated to the back burner. “Our time is limited,” he says. “We don’t have a lot of people, and we have a lot of projects. We have to prioritize, and that prioritization coming from the top down was key.”
Educate stakeholders. Institutions may encounter staff members who are perfectly content to have an Excel spreadsheet on their desktops, Sprague advises. “It serves their purposes,” he says. “When we say, ‘We need to get this to the data warehouse and make it accessible so it can be used with other data,’ the people holding the data don’t necessarily understand the big picture. It takes education and changing the culture.”
Don’t fear the cloud. While some traditional IT professionals may worry about the security of the cloud environment, Sprague does not. “We have found we’re able to keep our data more secure in a cloud environment than we could on our own campus,” he says. “Plus, it lets you use the most advanced technology without having to invest massive amounts of money.”
Expect expanded workloads. With increased data come increased workloads. Hamilton College was able to add an FTE to IT’s expanded duties, but other departments weren’t so fortunate. “We don’t have the ability to add FTEs willy-nilly, so we had to restructure the development office and create opportunities to grow analytics in advancement,” says Dennison.
Hewitt, who chairs the data governance group, is also juggling additional workloads without adding staff. “We’re doing the IR work—such as data reporting, analysis, and accreditation work—and helping to implement this new platform. It’s tough because people are demanding data. We’re trying to change the system and still feed the masses hungry for data.”
Allow time for data governance. “The first thing we learned is that you cannot do this if data are segregated on campus,” Dennison says. “If one database exists in the alumni office, another in the registrar’s office, another in the dean of students’ office, and another in athletics, the data become inaccessible. You need a data lake or warehouse, but that can’t exist until the existing data have been cleaned and defined.”
She estimates that defining all data on donor giving took almost a year. “It sounds so simple, but each question leads to another question,” Dennison says. “Be patient. This process is painful and long.”
Use analytics for decision-making. “Data analytics is a good investment as long as it is used to help the institution make better decisions,” Sprague says. “If you build a data warehouse and use it for ‘here’s-what-happened’ reporting, it’s not.”
Hewitt predicts that Hamilton’s data analytics efforts will pay dividends in the long run. “The products we are trying to develop will be game changers. They will be so far advanced from where we are now. We just need to marshal the resources and the will to get there. We certainly aren’t seeing cost savings yet, but we are getting good feedback on the projects that have come off the assembly line and feel really positive about the eventual rollout.”
MARGO VANOVER PORTER, Locust Grove, Va., covers higher education business issues for Business Officer.