Risely AI Research
What the public record says about the economics of education
Analysis of how institutions spend, price and retain. Every figure is either reported by a named source with its year or computed by us from one, and the working is shown on every page.
One student leaves. The bill depends entirely on who you ask.
the seat refills next term
the seat stays empty for years
A factor of 42, inside one sector. Retention is usually priced as though it were one number.
The analyses
What a student costs5
$5,300 to $224,000
What a lost student costs depends on whether the seat can be refilled
A departure removes about $5,300 of tuition at a community college and about $224,000 at a private medical school, and the difference is how many years the empty seat repeats the loss.
August 2026
Every college buys its class twice
The recruiting budget is the smaller of the two payments an institution makes to enroll each student; the larger payment is set one award letter at a time each spring, and it never appears in anyone's acquisition math.
$735
In Ohio, the median balance keeping a stopped-out adult from re-enrolling was $735
Eight Ohio institutions cleared old student balances, and the new tuition that followed was about 21 times what they would ever have collected.
$1,601
Where the IT dollar goes
Central IT, the technology department the whole institution shares, costs reporting institutions a median of $1,601 per full-time-equivalent student.
$12,800
The full price of a graduate has never been published
Every component of what a student costs, from first contact to diploma, appears in some published study.
Where the money comes from6
$20,500 cap
The new federal loan caps go by catalogue code, and the gap lands on the scholarship budget
On July 1, 2026, Grad PLUS ended and fixed borrowing caps took its place.
$7.07 billion
Universities now pay for a quarter of their own research
American universities spent $30.2 billion of their own money on research in fiscal 2024, and $7.07 billion of it was overhead they had earned under negotiated federal rates and never collected.
6 weeks vs 3 years
The same empty seat hits a Texas budget this year and a California budget in full three years later
California and Texas both fund school districts on attendance, but Texas reprices the money every six weeks while California pays on the highest of three years, so only one of the two superintendents can see the loss in this year's budget.
$80.27
The federal grant works out to eighty dollars a student, and the state money follows the federal paperwork
Money under the Carl D.
$27,617 vs $11,592
Two colleges can charge a family the same price and run on completely different money
For a family earning $30,000 and under, a private historically Black college (HBCU) and a private college outside the sector land on nearly the same net price, which is the amount the family actually pays after grants and scholarships.
$8,800 to $48,800
The same online institution can pay $8,800 or $48,800 a year for the same programs, depending on its home state
SARA, the State Authorization Reciprocity Agreement, lets a college enroll online students across state lines under one membership instead of separate approvals from every state.
Getting students in3
48% expect, 25% deliver
Online students expect an answer within three hours, and one program in four sends one
48% of online students expect a response within three hours of submitting an inquiry, and 25% of online programs report delivering one; both figures come from paired 2024 surveys by the enrollment research firm Ruffalo Noel Levitz.
8 to 40 percent
Every intervention that reduced summer melt could answer a student back
Between 8 and 40 percent of admitted students who intend to enroll never start, and the rate climbs as family income falls.
36% to 30%
Nobody keeps a national count of the students lost before census day
The enrollment count that state funding follows is taken on the census date, a day set a few weeks into the term when each college certifies who is enrolled.
Keeping them3
376 against 250
Every recommended caseload in education is public, and every measured caseload sits far above it
School counselors carry 376 students against a recommended 250, and school social workers carry 2,106 against the same recommended 250.
51% worked, 28% answered
The only published count of what happens after a student is flagged comes from the company that sells the flags
An early-alert system is campus software that lets an instructor, or an algorithm watching grades, flag a struggling student so an adviser can reach out.
11.4% less per degree
Reaching out first costs more per student and less per degree
In randomized trials, the programs that put a person on each student, an adviser, a coach, or a tax professional, kept more students enrolled and nearly doubled graduation.
Reaching families and donors3
$0.22
A postcard that goes to families beats a portal that waits for them
In a randomized trial published by the Institute of Education Sciences, a single $0.22 postcard cut student absences by about 2.4 percent (2017).
12.4% to 8.3%
The database grew faster than the donor file
Alumni giving participation fell from 12.4 percent to 8.3 percent between 2005 and 2014, the last year its publisher reported a national rate.
40 of 115
One major gift officer works about forty prospects a year
The published evidence converges on roughly forty prospects that one major gift officer actually works in a year, against assigned lists of 75 to 150 names.
Every piece carries a citation line, a dated source list, and a methodology section that shows the arithmetic behind each computed figure. Quote any finding with its source and year attached; that is what they are written for.