3 Lessons From Rolling Out AI at 250 Colleges

I’ve helped over 250 universities roll out AI for enrollment. Here are the top 3 things I’ve learned thus far. The first one is a make-or-break:

1. AI Readiness lives or dies on your data (aka data silos are bad).

The schools that stall all have the same problem. Their student data sits in systems that don’t talk to each other. The CRM has one version of a student stuck in time, the texting vendor has an even older one, the SIS has its own record, and there’s probably also a spreadsheet someone updates by hand on Fridays.

That happens because schools buy one tool at a time. A chat widget one year (“The board says we need a chatbot”), a texting vendor two years after that. Each one does its job. None of them share what they learn, so the student who told your chatbot she’s worried about affording nursing school gets an email the following week asking if she’s considered applying without acknowledging her concerns about affordability. Of course she notices. She doesn’t write in to complain about your integration strategy; she just stops answering altogether.

If your digital comms aren’t managed by a single platform, you will lose over time. AI needs data for context to be helpful. Most point solutions work, but they create data silos. Data silos are bad.

2. When the data connects, the individualization and outcomes are almost hard to believe.

A counselor can pay real attention to a few dozen students at a time. That’s the ceiling, and no human can avoid that. AI gives thousands of students a version of that attention at once, at 11 p.m. on a Tuesday, when the counselor is asleep, and a kid is staring at a half-finished application wondering if it’s worth finishing.

Pull the aggregate across a hundred schools, and the students who engaged with AI outreach deposited at a rate 2.5x that of the students who didn’t. AI is on when you can’t be.

3. Human touch still reigns supreme.

The teams that win point AI at the busywork so their counselors get time back for the conversation that actually matters, the one where a kid decides whether they belong in college at all. I was that kid! First-generation student, and nobody at my house had been through it. One counselor paid attention to me, and it literally changed the direction of my life. AI was never going to be that person. It just has to keep that person off inbox triage so she’s free when the moment shows up.

If you’re weighing AI for your team, look at your data before you sit through a single demo. PLEASE LOOK AT YOUR DATA. Tell me where your student context lives right now, and I’ll tell you how ready you really are.