Avoiding the College Enrollment Cliff With AI

min read

Higher education institutions must do more than implement rudimentary digital approaches to address the looming enrollment cliff. Artificial intelligence tools can help colleges and universities optimize data and address real-world institutional capacity constraints.

 robot helping business people to climb to the top of the mountain
Credit: FGC / Shutterstock.com © 2023

Colleges and universities in the United States are facing a looming threat that's come to be known as the higher education enrollment cliff. By 2025, the number of traditional college-aged students is predicted to decline by more than 15 percent. The Great Recession of 2008 and 2009 resulted in a sharp decline in the birth rate, which has stretched on for years.Footnote1 Correspondingly, there will be fewer children who reach college age in 2025.Footnote2

In addition, fewer students are attending high school, and high schools in some states have significant retention problems, further thinning the ranks of prospective college and university students.Footnote3 A growing number of career alternatives are drawing people into the working world directly from high school or to online alternatives to traditional higher education.

The enrollment cliff poses a major existential threat to higher education. Colleges and universities that don't act now may be forced to downsize significantly, reduce the number of active classrooms, shrink their physical campuses, eliminate programs, and lay off faculty and staff. Some institutions will be forced to shut down completely, and some already have. Between 2016 and the end of 2021, sixty-one colleges and universities closed or merged due to enrollment decline pressures.Footnote4

Do Better Things

Many higher education institutions have adopted rudimentary digital approaches to confront the enrollment cliff. They've gone through the first phase of digital transformation—digitizing workflows, leveraging data integrations to create more complete datasets, and creating dashboards and spreadsheets to accelerate the recruitment process and improve analytics.

While the first phase of digitization was about using technology to do the same thing better, institutions need to do better things to overcome the enrollment cliff. They must take a systematic approach to student recruitment and retention and leverage advanced artificial intelligence (AI) analytics tools.

What does "doing better things" entail? First, it involves doing a much better job of gathering the requisite data and knowledge needed to increase enrollment and retention. College and university leaders need to know where students at their institution come from, where they are being recruited from, and what populations the institution reaches out to and the cultural and socioeconomic backgrounds within those populations. What groups are underrepresented in institutional recruitment efforts or are ignored altogether, and how can the institution attract students who have traditionally been left out of the recruitment process? What is the best way to reach out to, nurture, recruit, and retain students from underserved populations? Getting a handle on that data is a crucial first step.

Understand Hidden Relationships

However, data alone isn't enough. Colleges and universities must also understand the dynamics and relationships between various data. For example, institutions may discover that students who attend certain high schools or play certain sports are more likely to enroll and apply to their college or university. Or, an institution might find that some students are more likely to apply or enroll if they receive a personal call from a faculty member. Maybe some students would not have applied at all if they had not received a faculty call.

Uncovering hidden relationships in the data presents an exponential problem, and humans are not good at exponential problems. Even a small number of individual factors can result in tens of millions of variable combinations to consider. AI tools can analyze the complex dynamics and relationships within institutional data.

Act On and Optimize Data

But even understanding the dynamics and relationships in institutional data is insufficient. The next level is actionability. Knowing that certain relationships and dynamics exist between data isn't helpful if institutional leaders don't know how to act on that knowledge. For example, simply knowing that some prospective students would be more likely to apply for enrollment if they received a call from a faculty member is not completely actionable. Faculty members can't call every potential student. They simply do not have enough capacity.

Instead, institutions must identify the applicants whose desire to apply will increase the most if a faculty member calls them, taking into account the faculty member's capacity constraints. That's actionability. AI technology tools have the potential to clearly identify the factors that cause change and make actionable recommendations that consider real-world institutional capacity constraints.

Once an institution has actionability, it has one final step—optimization. Optimization helps institutional leaders determine which capacity constraints need to be expanded to get the highest impact in the most efficient way. If faculty calls constrain an institution (whether that means there aren't enough faculty members to make calls, faculty members don't have enough time to make calls, or both), that area needs to expand. Then, given faculty members' capacity, who should they call to optimize the outcome? With AI-enabled optimization, faculty members won't, for example, end up calling prospective students who would be responsive to an email marketing campaign. Optimization helps institutions understand exactly where to apply constrained resources and collaboratively increase the most-constrained resources. For instance, certain faculty members might be more likely to call prospective students when presented with clear evidence of how such calls impact enrollment.

Get Started with AI

Colleges and universities can't wait until 2025 to start using AI. Institutions should test some hypotheses that come out of an AI analysis. AI tools need time to learn, and institutional leaders and others need to understand what the AI is telling them, confirm whether the AI is correct, and adjust it if it is not. Now is the time to invest in AI. The cliff is fast approaching.


  1. Jill Barshay, "College Students Predicted to Fall by More than 15% After the Year 2025," The Hechinger Report, September 10, 2018; Brady E. Hamilton, Joyce A. Martin, and Michelle J. K. Osterman, Births: Provisional Data for 2021, Vital Statistics Rapid Release no. 20, (Hyattsville, MD: National Center for Health Statistics), May 2022. Jump back to footnote 1 in the text.
  2. The declining birth rate exacerbates the existing financial challenges higher education institutions face. See Hernán Londoño, "Avoiding the Edge of the Cliff," EDUCAUSE Review, May 4, 2020. Jump back to footnote 2 in the text.
  3. Susan H. Greenberg, "Enrollment Remains Top Risk Cited by Colleges," Inside Higher Ed, January 6, 2023. Jump back to footnote 3 in the text.
  4. "A Look at Trends in College Consolidation Since 2016," Higher Ed Dive, January 18, 2023. Jump back to footnote 4 in the text.

Hernán Londoño is Senior Strategist, Higher Education, at Dell Technologies.

Arijit Sengupta is Founder and CEO of Aible.

© 2023 Hernán Londoño and Arijit Sengupta. The text of this work is licensed under a Creative Commons BY-NC-ND 4.0 International License.