I'll be honest: I was skeptical.
My HVAC company has been in business for nine years. I built it on relationships and reputation. The idea of letting an AI answer my phone felt like handing my front door to a robot.
But I was also spending $40,000/year on a part-time office coordinator who handled calls, scheduling, and admin. She was great. The problem was, she worked 9–5. My customers do not.
Emergency HVAC calls — real ones, the kind where someone's heat is out in January — don't happen at 2 PM on a Tuesday. They happen at 10 PM on a Friday. And I was losing those jobs every week.
So I gave it 30 days. Here's what actually happened.
Week 1: Setup and Paranoia
Getting 24/7 OnCall configured was faster than I expected. They built a custom AI trained on my business — my service area, my service types, my emergency vs. non-emergency pricing, my scheduling windows.
I tested it obsessively the first week. I called pretending to be a customer with a broken furnace. The AI asked the right questions: What type of heating system? Gas or electric? When did it stop working? Is it making any unusual sounds?
What I didn't expect: it didn't sound robotic. It sounded like a competent, calm office person. Good enough that I wasn't embarrassed to have it represent my business.
I kept my coordinator handling daytime calls for the first week and routed only after-hours to the AI. I wanted to compare directly.
Week 1 Results
Of 14 after-hours calls in the first week:
- 11 connected and qualified by AI
- 3 hung up before completing (two were likely spam)
- 7 turned into scheduled jobs the next morning
- 1 was a true emergency — furnace out, elderly homeowner, January. AI flagged it urgent, texted me immediately. I called back within 8 minutes and was on-site within 90.
That last one mattered. That customer has been with me for six years. If she'd hit voicemail, she would have called someone else.
Week 2: Going All-In
I made the call at the end of week 1: route all calls through the AI, day and night. My coordinator would focus on outbound follow-up, estimates, and jobs that needed human attention.
Three customers in week 2 mentioned — unprompted — that they appreciated being able to reach "someone" after hours. None of them knew it was AI. One said, "Your team has really good phone manners."
Week 3: The Data Gets Interesting
By week 3, I started looking at the logs differently. The AI wasn't just answering calls — it was building me a database.
I could see:
- What time of day most calls came in — peak: 7–9 AM and 5–8 PM, exactly when I'd historically been the worst at answering
- What types of jobs were calling — 67% HVAC service, 18% new installations, 15% other
- Where callers were located — 3 zip codes made up 40% of volume. I should be targeting those in my ads.
- Which calls came in after hours — 31% of total
I hadn't tracked any of this before. I just answered the phone (or didn't) and did the work.
Week 4: The Number That Matters
At the end of 30 days, I sat down with my bookkeeper and looked at the month:
- Jobs booked from AI-answered calls: 34
- Average job value: ~$520
- Revenue from AI-captured leads: ~$17,700
- AI receptionist cost: $99
But here's the number I actually care about: 11 of those 34 jobs were calls I would have previously missed entirely. After-hours calls that would have hit voicemail, from customers who wouldn't leave one.
Those 11 jobs = ~$5,700 in revenue that simply didn't exist before. $5,700 minus $99 = $5,451 net gain in the first month.
What I'd Do Differently
Start with after-hours only, then expand. The first week of parallel comparison was valuable. It let me verify quality before going all-in.
Customize more aggressively. The AI is good out of the box, but the more detail you give it — specific services, pricing tiers, service area boundaries, emergency criteria — the better it performs.
Tell your team. My coordinator felt nervous when I first mentioned it. A month later, she loves it. She spends less time on basic intake and more time on follow-up and customer care.
What I'm Still Figuring Out
A few situations the AI doesn't handle as well:
- Long-time customers who call just to chat. Handled professionally but not warmly in the same way.
- Very complex commercial jobs. AI captures the info and flags these as priority — but I need to get on the phone quickly.
- Very specific technical questions. The AI correctly defers these to a technician callback, which is the right answer — just not what I'd ideally want.
None of these are dealbreakers.
My Bottom Line
Would I recommend an AI answering service to another HVAC contractor? Yes, without hesitation — with two conditions:
- You have to set it up properly. Give it your business information. Test it. It's not plug-and-play magic.
- It's not a replacement for relationships. It's a filter. It handles intake so you can focus on the work that requires you.
If you're losing after-hours calls — and most HVAC shops are — the math is undeniable. One extra emergency job covers the monthly cost. Everything after that is profit.
The 30-day test became permanent. I'm not going back.
Start your free two-week trial at 24-7oncall.ai — $99/month flat, no contracts, no per-minute charges.