AI in Home Services — What’s Actually Working for You?

Roger "in this industry people usually expect to speak with a real person" I agree, and this is why I have not picked up an AI receptionist yet. I'm hoping to hear from people who have been using it. Here in rural America, people just don't want to pick up the phone and talk to an AI.

We haven't used any AI tools yet and/or AI receptionist. We've learned from a lot of other people and people want to hear a live person on the phone. I personally, hate when an AI receptionist answers the phone. I want to speak to a real person. 

Roger, if you'd like like to talk, I'd love to tell you about my voice system that I tailor directly to you and your business.  hilltopados.com 1.541.945.3698

jrselectric​ 
The "people want a live person" instinct makes sense and isn't wrong. The real question is what's actually happening when no live person is available.

Most electrical and trades businesses miss 30 to 40 percent of inbound calls during job hours. Those calls don't leave voicemails. They just call the next company on the list. The cost of that isn't visible day to day, which is why it's easy to underestimate.

The math changes when you put a number on it. If your average job is worth $400 and you're missing four calls a week, that's a real number walking out the door every month, and none of it shows up as a loss anywhere in your books.

The goal isn't to replace the live conversation. It's to make sure the lead is still there when you're ready to have it.

I think the distinction between replacing the human and helping before the human gets involved is really important. I don't like the AI receptionist either because I expect to talk to a person. If I open a website or ChatGPT, I don't want to talk to a person.
I’m curious how owners here feel about AI handling the repetitive questions first, things like services offered, areas covered, pricing questions, availability, policies, etc., and then passing the customer to the business once there is real intent.

The missed-call text example is interesting because it makes the leak measurable.

I’d be curious to track the whole chain separately:

missed call → text sent → customer responds → estimate/appointment → booked job

Otherwise it’s easy to say “we’re answering more calls” without knowing whether that actually recovered revenue.

I’d also separate missed calls during business hours from after-hours calls. Those seem like two different operational problems.

Has anyone here measured how many missed calls actually turn into booked jobs after the automatic text?

Roger's missed-call text follow-up is exactly the right instinct, and ahmedoct2000 nailed what most people miss: make the chain measurable. Voice tool debates are noisy, but the missed-call-to-booked-job chain is trade-agnostic and usually the fastest lever.

A practical way to set this up without guessing.

1) Treat the missed-call text as the after-hours version of your voicemail. When a call doesn't get answered, a text within 1–2 minutes should ask for the address, what's going on (no heat / clicking noise / estimate), and the best time to talk. Keep it to three fields — anything longer and the customer bounces back to Googling your competitor. The message should also mention they can book directly through your jobber.com request form if they prefer, which is just a faster version of the same thing for people who'd rather not phone.

2) Separate business-hours misses from after-hours misses in your measurement. Most of the money is in business-hours missed calls, because those customers are deciding right then whether to call the next number. Track: missed call → text sent → customer replies → estimate/appointment sent → booked job, and do it twice (in-hours and after-hours). The in-hours chain almost always converts higher and closes faster; the after-hours chain is where you're measuring whether your greeting, request form, and first-callback promise hold up.

3) Put the callback promise in the voicemail greeting, not a feature list. Thirty seconds, plain: who's answering, that they can also text back or use the online request form, and when they'll hear back ("within 15 minutes during business hours, or first thing in the morning if it's not an emergency"). The greeting is a commitment, and the text-back is the backup so the promise survives a bad ring.

4) Define what an emergency is before the weather decides for you. Especially in the winter months, pick the conditions that actually trigger a dispatch (active water leak, no heat in a freeze, safety) and put that in the after-hours greeting. Everything else is queued for first-callback in business hours and the greeting says so. That keeps the missed-call text from turning a "is my furnace clicking at 10pm" conversation into an overnight dispatch.

5) Feed the transcripts into Jobber customer notes so the next human doesn't have to re-qualify. Roger's callout on transcripts into Jobber is one of the easier wins — once the missed-call text or voicemail transcript lands on the customer record, the callback is a continuation, not a restart.

On the voice-receptionist side, the reason a lot of tree/HVAC/electrical owners still prefer the text-back over the live AI voice is the same one jrselectric and ThatHandymanVan surfaced: in-home-services work, scoping and urgency and trust happen in the first conversation, and the missed opportunity on a wrong scope is pricier than the missed call itself.

If you're measuring it, the question that decides the next move is usually: what percent of missed calls are happening inside business hours vs after hours, and how many of the text-back responses are turning into booked jobs this month? That split tells you whether the fix is a routing/forwarding setup, a better greeting plus a request form, or actually adding a person.