Every trade and vocational school we talk to has the same asset sitting in its CRM, and nobody touches it. Hundreds, sometimes thousands of people filled out a form, booked a tour or a call, said “I’m in,” and then went quiet. That list should be mined and reactivated with AI, not by an admissions rep with a spreadsheet on a Tuesday afternoon.
Here is exactly how it worked last week for one of our newer clients: the real text, who got it, and where every one of the 30 replies landed.
The problem: ads off for a week
An electrician school had to turn its ads off for a week because the sales team couldn’t take the calls. The normal move is to wait. Pause the ads, catch your breath, restart next week from cold, lose seven days of pipeline, and pay to rebuild it.
We didn’t wait. We pointed our AI reactivation agent at the CRM and told it to go back to everyone who had already raised their hand.
The results
| Metric | Result |
|---|---|
| Texts sent | 230 |
| Replies | 30 (13%) |
| Booked calls | 5 |
| Time to get there | 24 hours |
| Ad spend during the test | $0 (ads were off) |
The first booked call landed on the calendar the same afternoon, and two more came the next day.
Who the agent texted
The agent did not blast the whole database. It went after the people who had shown real engagement in the last few weeks, in three pipeline stages: new lead, Zoom no-show, and ready to sign up. All of them had booked an information call with the sales rep in the past few weeks and never enrolled.
That’s 230 people who already know the school, already took an action, and are already some of the way to yes. It is the warmest audience the school will ever have, and it costs nothing to reach.
The message
It was written in the owner’s voice and sent from the owner’s number, the same one these leads had already texted with.
“Hey [name], it’s [owner] from [school]. You reached out a while ago about training with us, but we lost touch. Are you still looking to get into the electrical trade, or did something change?”
Look at what isn’t there: no link, no fake urgency, no “spots are filling up,” no “just following up.” It is one question with two exits. “Still looking” and “something changed” are both easy to type with a thumb, and both tell you something you can act on.
222 of the 230 got that message. Eight people got a different one. They were sitting in “ready to sign up,” told the owner they were in, and never showed up to class:
“You were all set to start with us, but we never got you into class. Did something come up, or was it just the timing thing?”
“We never got you into class” puts it on the school, not on them. “Was it just timing” hands them the easy answer, and timing is a yes. Timing means “text me when the next class starts.”
Why the sends were paced
The 230 texts did not go out at once. They were spaced across about a day, starting Thursday afternoon and finishing Friday around lunch. If 230 people reply in the same hour, nobody can hold the conversations, and a number that fires 230 identical texts in 10 minutes gets flagged by the carrier. The final follow-up was personalized from any notes on the lead, with an angle of urgency.
All 30 replies, sorted
30 of 230 replied. That’s 13%, from people the school had written off as dead leads.
- About 16 are still interested — “Yes, still looking. Call me.” One wrote “nothing’s changed,” which is a yes in two words.
- About six are interested but stuck — One lost a job. One got sick and is saving up. One has no funding lined up. One moved out of state until next year. These aren’t no’s, they’re “text me in January,” and the AI is holding every one of them with a reason attached and a note to follow up at specific times in the future.
- Four said no — One is already an apprentice somewhere. That cleans the database.
- Two wrote in Spanish — The AI didn’t guess whether the school teaches in Spanish. It said someone from the team would reach out and flagged a human, and a Spanish speaker is following up.
- One sent a referral — A friend who wants the electrician course.
The math
Do the math on a school like this. At $8,000 a seat, if two of those five calls enroll, that’s $16,000 from a text batch that costs less than a sandwich. And that’s before the 16 who are still interested get their appointment times offered, and before the six who are stuck come back in January. The ads were off, the sales team was catching up, and the pipeline still moved.
One thing to do before you launch
Give the AI the answers before it needs them. Before we launch, we train our AI on an entire knowledge base about the school. When someone asks what it costs, we usually have the AI give the answer along with payment options if possible. Making someone sit through a call to hear a number they can Google is how you lose them.
How to do this with your dead leads
First, make sure your leads are segmented. Then put a reactivation agent on the list. Your list is already built and the ad spend is zero. This school got 30 conversations and five calls from leads it already owns.
If you want the AI reactivation agent running on your CRM, reach out to us.