So on Sunday I went into our Power BI to pull our application numbers. The report said 27,456 applicants for 2025-26. That’s a 74% jump in one year. Students were only up 7.6%. So that’s 11,700 more applications and about 500 more students.
For years, about half the people who applied to Forsyth Technical Community College in North Carolina became students: 48% in 2019-20, 51% in 2022-23, 42% in 2024-25. Last year it was 26%.
Two things are true at the same time here. On paper, it looks like a good enrollment year and a bad application year. You can’t trust that read, though, because a huge chunk of those applicants were never real.
What the ghosts look like on a dashboard
The fraud landed hardest on applicants 25 to 40 years old. Those almost tripled, up 157%. Above 40 almost doubled. And our conversion rates for those groups started looking awful, because so many of the “applicants” were bots.
Real adults still came, though. Adults went from 34% to 37% of our fall headcount, and their persistence held. So adult applications doubled, yet adult enrollment went up only about 18%. The gap between those two numbers is what we call ghosts.
Our admissions and records team looks at every application as it comes in. Anything that smells fraudulent gets set aside and goes into verification. That keeps the enrollment side of our reports clean. Even so, the application side still counts everybody who hit submit.
Where they came in
I figured this would show up in our CRM, that our leads would have doubled, too. Nope. We saw the same steady increase we expected from our digital marketing plan, and nothing more. In enrollment marketing, we’d call these stealth applicants. They never touched our site or our forms. They went straight to the application portal and submitted without ever entering our marketing workflows. Nothing my team owns was ever in their path.
This isn’t a California problem
It’s easy to think of ghost students as a California thing. The California Community Colleges reported that 31% of 2024 applications were fraudulent. In Iowa, Des Moines Area Community College found 6% of its fall applications were fake and started calling every enrolled student to make sure they were real.
North Carolina has it, too, and I’d bet most state systems are dealing with this onslaught. In March, the North Carolina Community College System’s chief information security officer told the State Board that the system is still figuring out how big this is and what it costs. Since February, colleges can flag suspect records into the system’s data warehouse, and the state application is getting a pre-screen to catch fraud before it ever gets to a college. In April, the FAFSA started checking identities inside the form. In June, the House passed the No Aid for Ghost Students Act.
All of that protects the money. None of it fixes the numbers already sitting in your reports. And the marketing and enrollment teams trying to figure out what worked last year are the ones stuck with it.
The denominator problem
Marketers live on metrics: cost per app, conversion rate, yield by program, melt, demand by age or county. Every one of those has applications in the denominator, and last year our denominator was mostly fake.
RAND’s 2025 study of community colleges in California, Texas and Kentucky found as many as two-thirds of applicants never become official enrollees. At that kind of volume, I haven’t seen anyone pull real loss apart from fake applications.
So a cabinet, or just a marketing officer on their own, could cut a channel because cost per applicant doubled, when what really happened is the bots left. Or call a campaign a huge win because it pulled 14,000 applications, when only about 2,000 of those became students. Neither mistake would feel like one in the room.
The access bill comes due
The usual fixes are ID uploads, callbacks, waiting periods and transcripts before the first class. All of them can work. Every one of them costs something at a college whose whole job is opening doors and knocking down barriers.
At Forsyth Tech, a flagged applicant gets asked for a photo ID, and admissions and records reaches out. But I keep thinking about a 38-year-old on her phone at 11 p.m. after the kids are finally in bed. Her transcript isn’t handy. She isn’t about to type her Social Security number into a form that popped up out of nowhere. To a filter built to catch bots, she looks like a bot. And she might just quit at the ID step for reasons that have nothing to do with fraud. If she does, she won’t show up anywhere as a student we lost. She’ll show up as fraud we caught.
We’re adding automated detection through our CRM and our refund provider. It’s hard. It should be a smidge hard.
Four things to do before spring budgets
- Restate your baselines. Pull your application totals for the last two years, before and after fraud removal. If you can’t produce a clean number, that’s your takeaway, and figuring out how is the next job.
- Report marketing at census. We keep a live count for operations, but the only enrollment number we actually report is census. I hope most colleges already do this.
- Measure verification by who drops out of it. Count the applicants who got asked for an ID and never sent one, then call a sample of them and find out how many were people.
- Get enrollment and marketing in the room where verification gets designed. We’d love to be in that conversation, because it has two sides: how many bots does this stop, and how many people does it stop?
Back to the dashboard
Looking at our dashboard right now, we’ve got a lot of applicants. Any other year, I’d be calling this a really, really good October and moving on. This year, with ghost applicants pumping up the numbers, I can’t tell you how many of them are people. Neither can your dashboard.
