Most of what you read about AI and sales is theoretical. This isn't. This is six months of live data from a real client in residential home building, plus an honest model of what one more change could do.
Meet Denes Aldott, a sales consultant at EasyStart Homes here in Australia. He has been running one of our voice products in live operation since January 2026. His verdict after testing the voice himself, before a single real lead went through: "That's just so unreal, well done!" His leads still don't realise they're talking to AI.
Here is what the numbers look like, and where they could go.
What Denes runs today (and what it doesn't have)
Denes uses our outbound product: a structured calling cadence that works his lead database with up to nine touches across six days, blending calls, SMS and email. No human caller time required.
The results are consistent and strong. In the December 2025 pilot, the system contacted 60 leads and produced 10 qualified bookings, a 16.7% booking rate. Across the live engagement from January to May 2026, processing 50 to 100 leads a month, it has held a 17 to 25% booking rate.
But here is the important part. This system is not a speed-to-lead system. Leads are uploaded in batches, not triggered the instant someone enquires. First contact might land hours or even days after the person originally raised their hand. There is no real-time trigger from web-form submission to first call.
Which means that 17 to 25% is being achieved without the single highest-leverage variable in lead conversion: speed. The floor is already this good. The question is what the ceiling looks like with speed added.
The model: what speed-to-lead would add
We won't insult anyone's intelligence with fantasy maths. The research (see our speed-to-lead post) shows a 3x to 21x improvement in qualification when you respond in five minutes versus hours. But you cannot simply multiply 21% by 21 and claim 441%. That would be dishonest, and Denes already recovers some leads through persistence, which narrows the gap.
So we model a conservative uplift built on two real mechanisms.
Mechanism 1, higher first-attempt contact. Today, Denes's leads are called hours or days after they enquire. Many have moved on or forgotten they enquired. With speed-to-lead, the first call lands within 60 seconds, while they're still at their computer thinking about their home build. Velocify's data shows up to a 391% contact-rate improvement at one minute versus two. Even conservatively, first-attempt contact should improve 2 to 3x.
Mechanism 2, captured after-hours leads. Today, a lead who enquires at 9pm on a Saturday hears nothing until Monday. With speed-to-lead, they get a call in 60 seconds. After-hours enquiries are 35 to 50% of total volume, and in a delayed-cadence system they're largely dead weight.
The conservative projection
| Metric | Current | With speed-to-lead | Change |
|---|---|---|---|
| First response time | Hours to days | Under 60 seconds | 500x+ faster |
| First-attempt contact rate | ~30-40% | ~60-75% | +80-100% |
| After-hours coverage | None | 24/7/365 | New channel |
| Booking rate (conservative) | 17-25% | 28-35% | ~1.4-1.6x |
| Booking rate (moderate) | 17-25% | 32-40% | ~1.6-2.3x |
| Monthly bookings (75 leads) | 13-19 | 21-30 | +8-11 |
At Denes's typical volume of 75 leads a month, that's 8 to 11 additional qualified bookings every month, from the same leads, with no extra ad spend. In home building, where a single booking can become a build worth tens or hundreds of thousands of dollars, that isn't a marginal gain. It's transformative.
Why we didn't claim more
The MIT research implies a 9x conversion lift for sub-five-minute response. Applied naively to a 21% baseline, that would suggest a booking rate approaching 100%. We don't claim that, for three honest reasons:
Not every lead is reachable regardless of speed: wrong numbers, spam, no real intent.
Denes's cadence already recovers some leads through persistence, so the incremental gain from speed is smaller than a from-scratch comparison.
Real-estate leads have their own qualification barriers: finance readiness, location, timeline, that exist independent of how fast you call.
We'd rather under-promise and over-deliver. 28 to 35% is defensible, achievable, and still pays for itself many times over.
The floor is 17 to 25% without speed. Add the one variable the system does not yet have, and the honest, conservative ceiling is 28 to 40%. Same leads. Same spend.