The 2am Call That Used to Go to Voicemail — A Home Services Company Recovers $180K in After-Hours Revenue

The call comes in at 2:17am. It's January. The heat is out.

Before: the call rings to the dispatcher's cell, which goes to voicemail because she's asleep. Or it rings and she answers, exhausted, and tries to coordinate a job in the dark. Or — most commonly — it rings to the main business line, hits voicemail, and the homeowner hangs up and dials the next HVAC company in their search results.

After: the call is answered in two seconds. The agent asks what's happening. Confirms the address. Pulls up the service history — this homeowner had a tune-up eight months ago, Bryant furnace, model number on file. Determines it's a genuine no-heat emergency. Creates the job in ServiceTitan. Sends a push notification to the on-call technician with the address, the issue summary, and the equipment type. The tech is on-site in 45 minutes.

That's the difference. Here's how it happened.

The Situation

The company ran eight technicians and one dispatcher. The dispatcher worked 7am to 5pm, Monday through Friday, with a modified Saturday schedule. Outside those hours — evenings, nights, weekends — calls went to the main line, which rolled to voicemail.

After-hours calls represented about 25% of total inbound call volume. The company's best estimate, based on two years of voicemail data and average job values, was that they were losing 3–4 emergency calls per week to voicemail abandonment. At an average job value of $4,500 for an emergency HVAC service call, that was $13,500–$18,000 per week in recoverable revenue. Annualized: roughly $180,000.

That number felt abstract until the owner pulled a specific week in February of the previous year — a cold snap, four no-heat calls after 6pm, all went to voicemail, all called back the next morning and went with whoever answered first. Three of those four had been customers before.

The dispatcher, for her part, didn't want to be the on-call answering service. That wasn't her job, and it was burning her out. She had started talking about leaving.

The owner's constraint: "I don't want to hire a night-shift dispatcher for 25% of my call volume. That math doesn't work. But I need to answer those calls."

The Approach

Mode 1 deployment with ServiceTitan integration. The agent went live in approximately two hours — most of that time was the ServiceTitan API configuration, not the agent setup itself.

The provisioning was straightforward: Workforce Wave crawled the company website, pulled the service areas, the equipment brands serviced, the service types (HVAC, plumbing had been added the previous year), and the general pricing structure ("free estimates on replacements, $89 diagnostic fee applied toward repair"). The knowledge base was complete enough to handle most caller questions without any manual addition.

The ServiceTitan integration gave the agent four capabilities:

1. Customer lookup by phone number (returning caller? Service history pulls automatically)
2. Job creation with pre-populated fields (customer record, service type, equipment notes)
3. Dispatch trigger (push notification to the designated on-call tech)
4. Appointment scheduling for non-emergency calls (routed into the next available slot)

The on-call rotation was configured as a simple variable: the owner updated a single field in the agent dashboard each week with the on-call tech's name and notification preference. The agent referenced that variable for every emergency dispatch.

The Configuration

The triage logic was the most important design decision. The agent needed to distinguish between:

• True emergencies (no heat, no cooling in extreme temps, gas smell, flooding) — immediate dispatch, on-call tech notified
• Urgent but schedulable (system making noise, reduced performance, pilot light issues, non-critical component failures) — next available slot, usually same-day or next morning
• Standard scheduling (tune-ups, estimates, quotes, maintenance agreements) — standard booking flow, business hours

The triage questions were simple and specific: "Is anyone in the home without heat/cooling right now?" "Is there any smell of gas?" "Is there water coming from the unit?" Three questions covered 90% of the triage decision. The agent was explicitly instructed not to pretend to diagnose — if it couldn't determine urgency from the answers, it escalated to emergency by default.

Equipment data from ServiceTitan was used where available: the agent could reference the customer's equipment type, age, and last service date when explaining what to expect on the call ("Looks like we serviced your Carrier heat pump about 14 months ago — the technician will take a look at the previous work when he arrives").

The Results

After-hours calls answered: 100%, starting day one. The voicemail abandonment rate went to zero because the calls were answered.

Emergency jobs dispatched after hours, first 90 days: 67 jobs. Of those, the owner estimated approximately 40–45 would previously have been lost to voicemail abandonment. At average job values, the recoverable revenue in the first quarter alone more than covered the annual cost of the platform.

Dispatcher retention: The dispatcher is still with the company. Her job no longer includes being woken up at 2am. In the owner's words, this was "not a small thing."

On-call tech experience: Measurably improved. The tech no longer receives a disoriented 2am phone call — he receives a push notification with the job address, the issue description, and the equipment type. He knows what tools and parts to bring before he gets in the truck. The average time from dispatch to arrival dropped from 58 minutes (when the dispatcher was coordinating) to 43 minutes.

The Intelligence Loop

Thirty days in, Workforce Wave flagged a pattern that the owner found counterintuitive: the agent was routing too many calls to on-call dispatch.

Specifically, about 30% of calls that the agent classified as "emergency" were actually better classified as "urgent but schedulable." The callers used emergency-sounding language — "it just stopped working," "it won't turn on," "something's wrong with it" — but when the agent asked the specific triage questions, the answers indicated non-emergency situations (a system that wasn't heating well but was producing some heat, a unit that had shut off but the homeowner was comfortable, a pilot light issue in a dual-fuel system with a backup).

The on-call tech was being dispatched for calls that could have waited for an 8am appointment. That meant overtime pay, wear on the on-call rotation, and tech frustration.

Workforce Wave generated a revised triage prompt that added one clarifying follow-up question for ambiguous situations: "On a scale of 1–10, how urgent is this — is anyone uncomfortable right now, or is this something you noticed but can manage until morning?" Callers who responded with lower urgency were offered first-morning appointments with a call confirmation. Callers who pushed back were still dispatched.

After the adjustment, on-call dispatch usage dropped by 40%. Not a single true emergency was missed. The on-call tech's weekend overtime went from a near-certainty to an occasional event.

What They'd Tell You

The owner's take, six months in:

"The tech gets a push notification, not a phone call. He sees the job address and the issue. He knows what to bring. That alone saves 15 minutes per call. The part I didn't expect was the triage learning — it figured out that half of what it was calling emergencies could wait, and it asked us before changing anything. I approved the change in about two minutes. Now the on-call rotation is manageable and my best tech isn't threatening to quit every February."

The dispatcher added one observation worth noting: "I was worried it would make me feel replaced. It doesn't. The calls it handles, I didn't want to handle at 2am anyway. The calls it sends to me are the ones that actually need me — complicated situations, upset customers, jobs that need real judgment. It filters the noise."