Ten years ago, I faced what seemed to be an insurmountable challenge. I had just been appointed head of student-success programs at Georgia State University, one of the most diverse public research universities in the nation. Among our 52,000 students, 67 percent are nonwhite and 58 percent are eligible for Pell Grants. Each year, thousands of students were enrolling with high hopes and quickly confronting a tangle of bureaucratic obstacles, academic challenges, and financial pressures. With little institutional support to help them navigate the maze, two-thirds of students were dropping out, with the highest attrition suffered by our low-income and minority students.
We knew what we needed to do. We needed to intervene earlier and far more proactively. We needed to deliver timely, individualized support for our students at scale. We just didn’t know how to do it. At the time, the kind of personalized attention our students required was the exclusive purview of institutions with hefty endowments and low student-to-faculty ratios. It was found at elite private colleges and a handful of well-resourced public flagships. Georgia State was neither.
The job is becoming more professionalized, holistic, and high tech. But colleges are just beginning to learn how to use new masses of data to help their students thrive.
How times have changed. Last year, we tracked every student at Georgia State for more than 800 analytics-based risk factors on a daily basis. We monitored each time a student registered for classes, and let students know immediately if they had signed up for a course that did not apply to their degree programs. We tracked academic performance as early as three weeks into each semester, assessed final grades earned by our undergraduates, and systematically intervened with students whose performance in prerequisite courses put them (according to the historical data) at risk of struggling in their next-level classes.
Our academic advisers conducted 54,000 in-person interventions with students prompted by alerts generated by our analytics platform, and they delivered another 100,000 individualized contacts by email, text, and phone. Students were guided through complex registration and financial processes by one online portal, and they could view metro Atlanta job statistics for hundreds of careers, updated daily, on another. More than 250,000 student questions about registration, financial aid, and academics were answered automatically by a newly launched chatbot — an AI-enhanced texting platform — with an average response time of seven seconds.
I am often asked how we made so many technological changes so quickly. How did we adopt, integrate, deploy, and scale multiple new platforms in such a short period of time? While the transformation seems far from sudden from my perspective — it is the result of, no joke, participating in an estimated 10,000 meetings with faculty, staff, vendors, and others — I have learned several lessons over the years about the selection and deployment of new student-success technologies.
First, don’t wait for the newest technology to be perfected before you act. At Georgia State, we have been very willing to get in on the ground floor of new technological advances, often before all of the questions have been answered. We signed on with a vendor to support the use of predictive analytics in advising before our partner even had a demo product to show us, and we were the first university nationally to sign a contract with the start-up vendor that supports our chatbot. In both cases, this approach allowed us to help shape the development of the product to better fit our needs and to acquire the systems for lower costs.
More important, the approach has allowed us to make changes to our failed traditional approaches much more quickly. While I would have liked for some other campus to do the work of blazing the trail, waiting would have delayed implementation by years, disadvantaging thousands of our students in the meantime.
Second, be wary of claims that you can solve big, complex problems that have vexed higher education for decades with home-grown technologies. Large companies have considerable bandwidth not merely to develop solutions but to offer continual updates post-implementation. I have seen many universities delay to act on the promise that “we will figure it out for ourselves.” Too often, the home-grown solution never materializes or is outdated soon after deployment.
This does not mean that campuses should rush into partnerships with technology companies. If all goes well, your relationship with a vendor can last for years. You obviously need to choose your partners carefully. But how? We have always selected companies that show that they genuinely understand the problem at hand. If you need to explain the problem to a vendor, it is highly unlikely that the vendor has developed a comprehensive solution. While our IT staff members are critical partners in the process, they do not lead the selection. The staff in functional areas such as advising and financial aid will be using the technology (you hope) years after your IT team has finished putting it into place.
Third, make sure that the solution makes sense to the users. By placing the functional teams front and center, you also will help to ensure that the new system interfaces with end users effectively. An all-too-common shortcoming of today’s technology vendors is a lack of attention to how their technologies will be used day to day. When faced with the choice between higher-powered capabilities with weaker user interface, and weaker capabilities with stronger user interface, I choose the latter every time.
The reason is simple. The success of any new platform depends 10 percent on the features and 90 percent on the effectiveness of the implementation. Your students, staff, and faculty will ultimately determine the impact of the new technology, so put the bulk of your efforts into changing existing behaviors.
When we started predictive analytics in advising, we asked our president to meet personally with our advising staff to promote buy-in, and we worked with the human-resources department to revamp adviser job duties to reflect the use of the new technology. When we implemented our chatbot, we hired a respected external researcher to conduct a random control trial to document the impact of the new approach.
We also run return-on-investment analyses of all of our programs — a step that has illustrated, more often than not, that the platforms pay for themselves by keeping more students enrolled, thus generating additional revenue from tuition and fees. Universities are indeed slow to change. With each new technology, you need to develop a plan to change not only platforms but attitudes.
Is all the work worth it? You decide. In part through delivering greater personalized support to our students at Georgia State, we have raised graduation rates by 22 percentage points and eliminated all achievement gaps based on race, ethnicity, and income level. We are graduating 2,800 more students every year than we were before we began these efforts, and more African-Americans than any other nonprofit college in the country.
Adopting new technologies can be frustrating, even overwhelming. On the other hand, what is the alternative? How can we explain to thousands of students that we might have made the difference between their dropping out and graduating, but we were reluctant to tackle the hard work of change?