It's 6 p.m. on a Thursday in July. It's 100 degrees outside, and a homeowner's AC just died. They call the nearest HVAC company and get placed on hold. In a sales-queue analysis published by Call Centre Helper magazine, over 50% of callers had already abandoned the call by the 45-second mark. Whether or not this exact caller hangs up at 45 seconds, the direction is the same: the longer the hold, the more likely they hang up and call the next name on Google. That's not just a lead lost. That's a $2,000 install walking straight to a competitor, and it happens to trades businesses every single day, in every single market.
This isn't a hypothetical. ACHR News, one of the HVAC industry's main trade publications, reported in June 2026 that contractors nationwide are adopting AI voice agents specifically to stop this bleed — agents that answer instantly, book directly into field service platforms like ServiceTitan and Housecall Pro, and never put a caller on hold. This case study covers how one regional company built exactly that system, and what changed over the following 90 days.
The Hold-Music Problem
Home services companies live and die by the phone. Unlike e-commerce or software, there's no cart to abandon and recover later with an email — there's a ringing phone, and either someone answers it in the next few seconds or the customer moves on. The trades are especially exposed to this because so much of the call volume is urgent and same-day: a broken water heater, a dead AC unit in a heat wave, a sewer backup. Nobody in that situation waits patiently. They call, and if nobody picks up, they call the next company.
For most home services businesses, the phone is answered by whichever office staff member isn't already on another call, dispatching a technician, or handling a walk-in. That works fine at low call volume. It breaks down completely during the seasonal surges that generate the most revenue — the first heat wave of summer, the first cold snap of winter — which is exactly when call volume triples and staffing doesn't.
Company Profile
The company in this case study is a family-owned HVAC and plumbing business operating three locations across a mid-sized metro area, generating roughly $6.2M in annual revenue with 34 employees — the large majority of them field technicians. Before this project, two office staff split phone duty alongside dispatch, invoicing, and parts ordering. There was no after-hours answering service beyond a voicemail box that technicians checked (inconsistently) the next morning.
Ownership already knew calls were slipping through the cracks. What they didn't have was a clear picture of how many, or what it was costing them.
What Was Broken: A Two-Week Call Audit
Before building anything, the company ran a two-week audit using their phone system's existing call logs, cross-referenced against jobs actually booked in ServiceTitan. The findings:
- Roughly 1 in 4 inbound calls during business hours went to voicemail or a busy signal during peak call windows (early morning and late afternoon)
- After-hours and weekend calls — nearly a third of total call volume — had no live coverage at all beyond voicemail
- Voicemails left after hours were returned the following business day in the majority of cases, by which point the caller had frequently already booked with someone else
- Office staff spent significant time on calls that didn't need a human at all — appointment confirmations, "what time is my tech arriving," basic pricing questions — time that came directly out of their capacity to handle new-customer calls
The audit made the priority obvious: the highest-value fix wasn't hiring more office staff. It was making sure every call got answered immediately, regardless of time of day, without needing a human for the calls that didn't require one.
The AI Dispatch System: Four Components
The rebuild took about five weeks from kickoff to full rollout across all three locations, built around the company's existing ServiceTitan instance.
1. AI Voice Agent for Every Inbound Call
The core of the system is an AI voice agent that answers every call — 24/7, across all three locations — in under two seconds. It identifies the issue (AC down, no heat, leak, clogged drain), asks the qualifying questions a dispatcher would ask, checks real-time technician availability and location in ServiceTitan, and books the appointment directly onto the schedule. True emergencies — active gas leaks, flooding, no heat with a vulnerable occupant — are immediately escalated to an on-call technician rather than handled by the agent alone. Similar to what we've covered in our guide to AI voice agents for small businesses, the highest-ROI use case is almost always after-hours coverage, because that's where the company is starting from zero.
2. Missed-Call Text-Back
As a backstop for the rare call the voice agent doesn't fully resolve, an automated text-back fires within seconds containing a direct booking link and the office's callback hours. This closes the gap for customers who prefer texting over talking to any voice system, human or AI.
3. Automated Appointment Confirmations and Reminders
Every booked job now triggers an automatic text confirmation immediately, plus a reminder the morning of the appointment with the technician's name and estimated arrival window. This offloaded a category of calls that previously consumed real office-staff time — "just confirming my appointment" — freeing that capacity for new business.
4. Post-Job Invoicing and Review Requests via Make.com
When a technician marks a job complete in ServiceTitan, Make.com now handles the handoff automatically: the invoice is generated and sent, payment is collected via the customer's preferred method, and — 24 hours later, assuming no service issue was flagged — a review request goes out via text. The office manager's shared calendar and email triage for scheduling callbacks also moved onto Google Workspace, so all three locations share one real-time view of bookings instead of location-specific spreadsheets.
Losing jobs to hold music?
We build AI voice and dispatch systems for home services companies — HVAC, plumbing, electrical — that plug directly into ServiceTitan, Housecall Pro, and the tools you already run. Book a free call to see what's possible for your shop.
Book a Free Strategy Call →Results After 90 Days
The company tracked call and booking data closely for 90 days after full rollout. Here's what changed:
- Live call answer rate: from roughly 75% to effectively 100% — every call is answered instantly, day or night
- After-hours bookings: from near zero to accounting for approximately 18% of all new jobs booked
- Average time-to-booking for a new service call: from several minutes of hold/callback time to under 90 seconds
- Office staff time freed up: roughly 12–15 hours per week per location, reallocated toward parts coordination and handling the complex customer situations that genuinely need a person
- Online review volume: increased meaningfully month over month once review requests became automatic and consistent rather than dependent on a technician remembering to ask
Ownership's own estimate, based on comparing booked-job volume against the same 90-day window the prior year and isolating the after-hours and previously-missed-call segment, put the recovered revenue at approximately $30,000 per month — against an implementation and monthly tooling cost of well under 10% of that figure. As with any case study, isolating the AI system's exact contribution from seasonal demand and other marketing changes is imperfect, but the after-hours booking data — going from near-zero to a meaningful share of monthly volume — is the cleanest signal that the phone, not the marketing, was the bottleneck.
What AI Can't Replace
The AI voice agent is excellent at the transactional core of the job: answering instantly, qualifying the issue, checking real technician availability, and booking the appointment. It's consistent in a way a rotating cast of office staff, handling ten other things simultaneously, structurally can't be.
What it doesn't replace is judgment on the edge cases — the upset customer disputing a prior repair, the borderline emergency call that needs a human's read on urgency, the pricing negotiation on a large system replacement. The agent is configured to hand these off to a human immediately rather than attempt to resolve them itself. Ownership reviews a sample of call transcripts weekly, both to catch anything mishandled and to refine qualifying questions as new call patterns show up.
It's also worth being honest that a voice agent doesn't fix an underlying capacity problem. If a company is chronically understaffed on technicians, answering every call faster just means more customers hearing "the earliest we can get to you is Thursday" — better than voicemail, but not a substitute for enough trucks on the road.
How to Replicate This for Your Shop
If your business takes more than a handful of inbound service calls a day and doesn't have live after-hours coverage, you're very likely leaving money on the table. Here's the practical starting point:
- Pull your own call data first: Most phone systems and field service platforms (ServiceTitan, Housecall Pro) already log this. Compare calls received against jobs actually booked for a two-week window and see where the gap is.
- Start with after-hours coverage: It's the highest-ROI first deployment because most shops are starting from zero live coverage in that window.
- Integrate directly with your dispatch system: A voice agent that can't see real technician availability just creates double-bookings. The integration into ServiceTitan or Housecall Pro is what makes this actually usable, not a novelty.
- Automate the post-job loop: Invoicing, payment collection, and review requests are the easiest wins to bolt on once the intake side is working — tools like Make.com handle this without custom development.
- Review transcripts weekly for the first quarter: This is a managed system, not a set-and-forget one. The calibration in the first few months is what determines whether it feels helpful or robotic to your customers.
The shops winning the most jobs in 2026 aren't necessarily running the biggest ad budgets. Increasingly, they're the ones who pick up the phone in two seconds flat, every time, including the Saturday night the water heater decides to fail. That's a fixable problem, and it's considerably cheaper to fix than most owners assume. For a broader look at how this fits into daily operations, see our guide on AI for operations managers, or read how a law firm solved a strikingly similar intake problem in a completely different industry.