77 dials. 16 conversations. 9 emails captured. 17 prospects wanted a demo. Zero meetings booked. The funnel tells you exactly which lever to pull.
The list is healthy — 54.5% dial-to-connect, 1.3% wrong-number rate, gatekeepers receptive enough to hand over direct decision-maker emails 4 times today. Targeting is fine.
The offer is structured to terminate at email-send, not at a booked meeting. Marcus said "I'll email you the demo" 17 times and 17 prospects said "yes, send it." He's performing the offer exactly as written — and the offer's natural end-state is "demo sent, follow up later," not "meeting on the calendar." Zero meetings is not a Marcus problem. It's an offer-design problem.
Pull the data Trellus surfaces directly. Each stage tells you whether the lever above it is broken.
| Stage | Today's rate | What it tells you |
|---|---|---|
| Dial → Connect | 54.5% | Healthy. Above HVAC cold-call benchmark (30-50%). 1.3% wrong-number rate. List is fine. |
| Connect → Pitch | 54.8% | Healthy. Gatekeeper-getting works. Marcus opens enough doors. Opener is fine. |
| Pitch → Conversation | 69.6% | Strong. People LISTEN to the offer. They engage. They give direct emails. Offer is interesting. |
| Conversation → Meeting | 0% | THE WALL. 16 conversations, 0 bookings. The offer never asks for a meeting — it asks for an email send. Prospects say yes to the email and the call ends. |
If the LIST were broken, dial→connect would be 20-30% (lots of dead numbers, wrong contacts, closed shops). It's 54.5%. If the OPENER were broken, pitch→conversation would be 30% (people shut down after hearing it). It's 69.6%. Every stage above the wall is performing within or above benchmark. The wall is the conversation→meeting transition — which is dictated by what the offer asks the prospect to commit to.
"I built an AI tool for [Owner] for free, based on their website. I'll send you the 60-second demo audio so [Owner] can hear it on his time."
That offer terminates at "I'll send you the demo." There is no meeting ask. The prospect's possible answers are yes (send it), no thanks, or send it to [email]. None of those are a booking. Marcus is closing the offer perfectly — the offer just doesn't include the booking.
Removes the email-send step entirely. Prospect either listens on the line or doesn't. If they listen, immediately stack the meeting ask.
No demo audio at all. The OFFER is "15 minutes of my time + a free working system installed on your phone today." That's a meeting on day one — no follow-up step.
Drop solo owner-operator shops with no inside sales. Today's signals: Goff Heating's 85-year-old Chuck "doesn't like talking to people," Estes Heating "prefers personal interaction," small shops generally deflecting with "we use an answering service" instead of recognizing the after-hours pain. These accounts won't book a meeting for AI regardless of offer quality.
Filter: require 5+ employees OR a "Careers" page on the website (proxy for org maturity) before adding to Marcus's dial queue. Estimated 10-15% improvement at this layer. Real wins still come from the offer change.
Based on today's funnel: 77 dials → 16 conversations → 0 meetings. If the offer change is real, tomorrow's same 77 dials → 16 conversations → 1 meeting booked. One. That's a 1/16 (6.25%) close rate at the conversation layer. Achievable, measurable, and the single delta between "pipeline filling" and "calendar filling."