HVAC Contractor — 94% Contact Rate, ~14× Projected ROI
The situation
A regional HVAC contractor with 6 service trucks and 11 total team members had a chronic missed-call problem. During business hours, when all 6 trucks were on calls, the office line went to voicemail. After hours and weekends — about 30% of total weekly call volume — every call routed straight to voicemail. Average call-back time was around 4 hours, by which point a substantial share of those callers had already hired a competitor.
Pre-pilot data showed: roughly 40% of business-hours calls unanswered, ~100% of after-hours calls missed. Estimated lost revenue from missed-and-not-recovered inbound: $8,000–$15,000 per month based on average HVAC service-call value.
What we did
1. Day 1 — Kickoff & voice agent training
30-min Zoom kickoff to align on: top 14 call types (no-cool, no-heat, maintenance, estimates, parts, billing, etc.), business hours, escalation rules, AI disclosure phrasing, and tone (warm, slightly conversational, never robotic). Voice agent trained on the client's FAQ document + a sample of past customer interactions.
2. Day 2 — Deploy + 48-hour go-live
Call forwarding configured so the AI agent answered when the team didn't pick up within 3 rings. Missed-call text-back wired to fire SMS within 60 seconds of any missed call. Web-form Speed-to-Lead: any form fill on the client's site triggered an AI callback within 60 seconds. All inbound calls + leads land in the client's existing CRM via API integration. AI disclosure scripted per CA AB 2013 compliance at the start of every AI-answered call.
3. Day 3-14 — 14-day pilot phase + daily script tuning
Listened to a sample of recorded calls every morning. Logged any agent missteps. Pushed script updates daily based on what the data actually showed (3 major updates during the pilot — see “what we learned” below).
4. CRM integration + call recording / transcription
Every call recorded, transcribed, and searchable in the client's dashboard. Every captured lead lands in CRM with full transcript + recording link + sentiment score + AI-suggested next-action.
The results
14-day pilot phase, 50 inbound calls. Breakdown:
| Outcome | Count | % of pilot |
|---|---|---|
| Answered live by AI agent → captured contact info | 23 | 46% |
| Answered live → appointment booked into calendar | 15 | 30% |
| Missed call → recovered via SMS text-back <60 sec | 9 | 18% |
| Customer escalation to human (handled correctly) | 2 | 4% |
| Agent error / script update logged | 1 | 2% |
| TOTAL CONTACT (live + recovered) | 47 / 50 | 94% |
ROI math, annualized against pre-pilot baseline of ~57% contact rate:
- Weekly booked appointments: ~8 (before) → ~14 (with AI agent)
- Weekly booked revenue: $3,400 → $5,950 (at $425 avg ticket)
- Annual booked revenue lift: +$132,600
- Annual engagement cost (at $899 setup + $699/month): $9,287
- Year-one ROI: ~14× return on incremental booked revenue
What we learned
Three script updates the pilot data forced us to make — none of which would have been obvious without 14 days of real call recordings:
- Day 4 — Same-day service over-promising. The agent was confirming “we can have a tech out today” during peak heat when dispatch was already overbooked. Updated script to check dispatcher availability before quoting timing.
- Day 7 — “What's wrong with your AC?” was confusing customers. Most people don't know how to describe HVAC failure modes. Replaced with a triage tree: “Is it running but not cooling, not running at all, or making a noise?” Conversion improved measurably.
- Day 11 — Financing questions were being escalated. Three callers asked about financing during week 2. Added the client's financing partner details directly into the agent's knowledge base so it could answer in-line instead of escalating.
The pilot phase is the entire reason this works. A static voice agent shipped on day 1 will sound robotic and miss real customer language patterns. 14 days of real calls + daily tuning is what turns a generic agent into one that sounds like it works at the company. We won't ship a voice agent engagement without it.
Want results like this for your business?
Start with the FREE AI Visibility Audit + Fix-It Kit (normally $199). Full visibility score, ready-to-paste fixes, 5-step action plan custom to your site. FREE this month only.
Get My FREE Audit →