We took one AI voice agent, pointed it at a list of leads nobody had called in months, and let it dial. 11,000 calls in 15 days.
This is the full breakdown. Every number, including the two it closed.
But I'll tell you the punchline first, because it's the part that made me change what I do for a living. This ran on a list everyone had given up on. If it works there, it works almost anywhere.
The list nobody was calling
Some context. I spent five years running a high-ticket coaching business. We scaled it past $3 million, with over 15,000 clients.
After a few years of that, you end up with something every business has and nobody works properly: a big pile of leads sitting in the CRM, going cold. People who bought something once. People who raised a hand and then went quiet.
Everybody knows you should call them. Almost nobody does, consistently, at scale. Either you don't have the time, or a sales team costs more than the return while you're still figuring it out. I've been there.
What the agent actually did
This was not cold calling. Every person on the list had bought from us before. They weren't warm, some hadn't heard from us in years, but they knew who we were.
The agent did four things:
- Called existing buyers from the list, on autopilot.
- Followed a script that was already proven by our human appointment setters.
- Qualified with the BANT framework: budget, authority, need, timing.
- Booked a call only when the lead was actually qualified. If not, it tagged the interest and notified a human rep for follow-up.
The voice was ElevenLabs. The logic and the connections to the CRM and calendar ran in n8n. Nothing here was reinvented. We copied a process that already booked calls, and let software run it without getting tired.
The numbers
Fifteen-day window. Here's what came back.
| Metric | Result |
|---|---|
| Spend (tools, credits, telephony) | ~$457 |
| Leads called | 2,825 |
| Total dials (~4 per lead) | 11,000+ |
| Pickups | 542 |
| Meetings booked by the agent | 22 |
| Deals closed by a human rep | 2 |
| Revenue | $6,800 |
Two deals. One at $4,800, one at $2,000. One of them was a lead who had entered our CRM about two years earlier. Nobody was ever going to call him. That's the moment I sat there and thought, alright, this is what I'm doing for the next ten years.
The comparison that made me switch
The raw revenue isn't the interesting part. The comparison to a human team is.
- Response rate from dials: about 3.5% with our human setter, 4.7% with the agent. Not because the AI is smarter, but because it actually calls every lead, four times, without a bad week.
- Cost per booked call: $50 to $200 with a human setter, down to about $20.
- Monthly cost: an appointment setter runs around $3,000 a month. The agent ran the whole campaign for roughly $500.
A human setter on a good week does maybe 200 calls a day. Then they get sick, or coached, or life happens. The agent just works, like clockwork.
Why it actually worked
Read this part twice, because it's the part people skip.
The agent didn't create that revenue. Three things did, and they were true before AI touched anything:
- The offer was proven. We'd sold it for years.
- The appointment-setting script was validated by humans first.
- We had closers who actually close.
AI removed the inconsistency from a system that already worked. That's the whole trick. It's leverage on something real, not a substitute for it. If you want the exact economics and the three lessons, I broke them down here. If you want to see this specific dead-list play, it's here.
-Ruben
Grow your revenue or cut your costs in 90 days. Or you don't pay.
Done-for-you AI systems, the same ones I run in my own agency and build for 7 and 8-figure coaching and consulting businesses. Want them installed for you? Let's talk.
