The Enrollment Cliff in Higher Education Is an Operations Problem

Quick answer: The enrollment cliff in higher education is the sustained decline in US high school graduates that begins after the 2025 peak and continues through 2041. Most institutions answer it by buying more applicants and discounting deeper. Both moves work on the top of the funnel. Neither reaches the students who already said yes and then failed to arrive, who are the cheapest enrollments an institution will ever have the chance to keep.

Why enrollment spending misses the students you already won

Every fall, somewhere between census day and the first cabinet meeting, a version of the same conversation happens. Enrollment came in under plan. The room works through the familiar list. Search names cost more than they did three years ago. The discount went up again. A competitor opened a new scholarship. Somebody proposes a bigger media buy for spring.

What almost never comes up is the four days an admitted student waited for a financial aid decision in April. Or the housing assignment that landed after the deposit deadline. Or the transfer credit evaluation that finished in August, for a student who had already registered somewhere else in June. None of those students appears in the marketing report. They said yes. They are counted as won until the day they are written off.

The demographic pressure is real. WICHE projects that the nation reached its highest-ever number of high school graduates in 2025 at 3.9 million, and that the figure falls to 3.4 million by 2041, a decline of 13 percent. That is the environment every enrollment plan now operates inside. It is not, by itself, a plan.

The strategic question is narrower and more uncomfortable. When the pool shrinks, does an institution spend to buy back volume, or does it first stop losing the students it has already won?

Three numbers that should change where enrollment money goes

Take them together instead of one at a time.

Applications are growing faster than applicants. Through the end of the 2025-26 season, Common App reported applications up 6 percent while distinct applicants rose 2 percent. The same students are applying to more places. Volume at the top of the funnel now says less about who will enroll, and it costs more to generate.

Yield has been drifting down. NACAC reported an average yield rate of 30.2 percent at four-year not-for-profit institutions in fall 2022, against 32.1 percent in fall 2019. That remains the most recent national figure published, and it predates the current cycle. Treat it as a direction of travel.

A meaningful share of admitted students never arrive. EdResearch for Action puts summer melt at 10 to 20 percent of college-intending students nationally, with higher rates among low-income students. Melt has financial, personal, and academic causes. Operational friction is one contributing factor among them, and it is the one an institution controls directly.

Now set those against sector spending. NACUBO estimates the first-time undergraduate discount rate at private nonprofit institutions reached 57.1 percent in 2025-26, up from 54.5 percent a year earlier. NACUBO also reports that net tuition revenue from all undergraduates fell 1.9 percent after inflation in 2024-25.

Read plainly, that describes an institution paying more to acquire each student while losing an unmeasured share of the ones it acquired, for reasons unrelated to price or fit.

Why the interval between yes and day one has no owner

It is not that nobody cares. It is that no single office owns the interval.

The CRM manages recruitment and accurately reports the funnel up to the point of admission. The student information system owns the record once enrollment is complete. Between them sits a sequence of decisions that four or five offices have to make in a specific order, and that work lives in email, shared spreadsheets, and PDFs.

Admissions releases the decision. Financial aid packages are awarded, which sometimes require a verification document. Housing assigns a room, which sometimes waits on the deposit. The registrar clears transfer credit, which determines what the student can register for. IT provisions the account that makes any of it usable. Each office does its job competently. None can see where the student actually stands.

Because the interval has no owner, it has no budget line, no cycle-time metric, and no name in the strategic plan. Marketing spend gets attributed to enrollment. Operational delay is not attributed to the melt, so nobody funds its repair.

What the leak costs the CFO, enrollment, aid, and IT

For the CFO, this is the only enrollment intervention with no acquisition cost. The student has already been recruited, reviewed, admitted, and discounted. Every dollar of that investment is spent. Losing the student in July writes it all off.

For the VP of enrollment management, it is the difference between a yield number you can explain and one you cannot. A funnel report shows how many admitted students did not enroll. It does not show how many were waiting in your offices when they decided.

For the registrar and financial aid, it is a capacity constraint that reads as a performance problem. NASFAA found that 91 percent of aid administrators reported greatly or somewhat increased time and resources per application over the past five years. NASFAA also found that 52 percent experienced moderate or severe resource shortages during peak periods. Of those, 68 percent described the shortfall as permanent. Peak season pressure is structural.

For the CIO, it is where shadow systems come from. When an office cannot get a process supported, it builds one in a spreadsheet or buys one on a card, and the institution acquires another unreviewed application holding student data.

What closes an operational leak in the enrollment journey

Closing it requires neither a CRM replacement nor a Banner or Workday project. Those systems are doing what they were built to do. What fails is the work between them, which no system of record was designed to carry.

That work needs a governed place to run. A governed execution layer sits alongside the systems of record. The office that owns a process builds and maintains the workflow. IT governs the standards it runs under. Each step routes to the correct approver, and every decision is recorded with an owner and a timestamp, so a complete file is available on demand rather than on request. That last point pays twice, once for the student waiting in April and once for the Title IV program review three years later.

Kissflow is the governed execution layer at the edges of the enrollment stack. It connects to the systems your institution already runs, including Banner, Workday, and Ellucian, and replaces none of them. Its AI generates governed, human-readable applications instead of code. When policy changes, the director of admissions can change the review step without a vendor ticket or a developer queue, and IT can audit what moved.

Four steps to take before the next admit cycle

• Start with a single interval. The most common candidate is admitted to the award because it is short, it is measurable, and directly tied to the deposit decision.
• Map who actually approves what, and where the file truly waits. The answer is rarely the step people assume.
• Move that interval onto a governed workflow and leave the CRM and the SIS untouched.
• Measure cycle time in days before and after, then report it beside the yield. One measured interval funds the argument for the next.

The institutions that hold their numbers through the next decade will be the ones that stopped losing students they had already won.

Frequently asked questions

What is the enrollment cliff in higher education?

It is the projected sustained decline in US high school graduates following the 2025 peak of 3.9 million, falling to 3.4 million by 2041, according to WICHE. It shrinks the traditional-age applicant pool nationally, and the effect varies sharply by region.

Is summer melt really an operations problem?

Partly. Melt has financial, personal, and academic causes that no institution controls. Operational delay is one contributing factor, and it is the one that an institution can measure and reduce within a single cycle without new acquisition spend.

Do we need to replace our CRM or SIS to fix this?

No. These failures occur during handoffs between those systems rather than within them. The fix is to give that in-between work a governed place to run while the systems of record stay where they are.

What should we measure to know it is working?

Cycle time in days for each named interval, captured before and after, reported alongside yield and melt. Days to aid decision and to transfer credit evaluation are the two most useful starting metrics.

How long does one interval take to move?

Scope a single process instead of a program. One interval with a clear approver map is term-scale work, which is what makes it fundable without net-new headcount.