Higher ed is sitting on mountains of data about student finances, program costs, enrollment and recruitment, student behavior, retention, graduation rates, and more. But the institutions don’t know what to do with it. Skills gaps, data silos and limited budgets to invest in modern systems are barriers to colleges using their data effectively. The explosion of AI in fields across campus and new employee data needs from federal and state agencies mean an urgent solution is needed.
At ASU+GSV, the largest convocation of educators, investors and ed-tech entrepreneurs this month, every conversation I had was dominated by AI. There are huge opportunities to improve institutional development, student experience and student support. And many student information systems, customer relationship management platforms and workflows for learning management systems have been equipped with agents and AI-powered systems.
Yet the promise of those tools is only as good as the purity and accuracy of the data that flows through them—and the human skills to analyze the results. Bad data in + bad data out = bad decisions all around. There are too many institutions in this situation. Just one percent of Educause community members surveyed said their institution’s data systems are fully modernized, and nearly two-thirds (68 percent) said some systems are being modernized, and another third said they are in early discussions about data modernization (24 percent) or not talking about it at all (7 percent). In short, many colleges are buried in data they cannot fully utilize.
It’s what Mark Milliron, president of National University, calls “bad pipes.” The National team spent a year establishing inclusive data management and building a comprehensive data warehouse, including mapping their data and cleaning it (for example, making sure entries are consistent and in the right format) before they integrate it into their software systems. “I don’t think we could have scaled some of the strategies we’ve done without doing the plumbing work beforehand,” he said at our Student Success event a few months ago.
The National example can hold courses in other institutions. Fixing data pipelines will look different for each college, but getting their data—in areas like student performance, enrollment costs and program costs—in order is critical for institutions before they sign software deals with eye-watering price tags. Data-savvy administrators will know how the platform works when it’s not talking to other technical systems, and can pass on good data skills to new employees. That foundation can even put institutions on the path to greater technological independence.
Now that National has developed data capabilities across teams, Milliron said they are moving to design thinking and “domain expertise”—a deep understanding of National’s extraordinary, active and military students—to create bespoke programs for their campus. All colleges across the country have that background knowledge when it comes to their students. With clean data systems and big data insights, colleges can build their AI-powered platforms easier than ever.
These information management problems are not limited to institutions. They are increasing, leaving state governments in the dark. The sector does not have the right data on the management of qualifications and industry jobs in the right areas to respond to the needs of the labor market or to track the return on investment in individual programs.
In a session at ASU+GSV, Chris Mullen, director of data strategy and measurement at the Lumina Foundation, gave a preview of a new project to deliver basic buckets of organizational data—collected under the Job Creation and Opportunity Act and the Perkins Career and Technical Education Act, as well as the Registered Apprenticeship Partners Information Database System and provide the Education Data Database System with real-time government information on talent pipelines and industry demand.
The numbers are about to rise. Linking education and workforce data will be key to Title IV funding under new program-level accountability measures that go into effect July 1. As it stands, “do no harm” rules require data sharing between institutions, states and the federal government to understand if graduate programs are producing graduates who earn more than a local working adult with only a high school degree. Graduate programs will be judged on the earnings of bachelor’s degree holders. Workforce Pell eligibility will require similar follow-up from providers and state leaders before funding begins to flow.
The effort to coordinate data across agencies may sound herculean, but it’s not unheard of. In the ASU+GSV session, Kristin Hultquist, CEO and founding partner of HCM Strategies, pointed out that the IRS data feed on federal student aid applications makes the FAFSA process more accessible to more students. Lumina’s proposal will “rethink the collaboration of provincial data,” he said, and reveal what data is missing and where it should be cleaned up. Arkansas, he added, is an example of a state that already compiles data well. Its LAUNCH program connects job seekers and employer data in a free, AI-powered environment to address the needs of government workers.
Data on higher ed is available. The tools to use it to make real, evidence-based decisions are here. What’s missing are the pipelines to connect the systems and the investment in poor work to clean up the data that flows through them. It’s difficult but urgent: You can’t build an AI-powered future with dirty data and broken pipelines.
