Will Customers Hate Talking to an AI Receptionist When They Call My Small Business?

It's a Tuesday in August in Katy. The upstairs AC quit around 6:30 pm, the house is 84 degrees and climbing, and a homeowner is scrolling Google for an HVAC company. The first two calls ring out to voicemail. The third one picks up on the first ring: "Thanks for calling, I'm the AI assistant. I can get a technician scheduled or connect you with someone. What's going on?" Ninety seconds later she has a morning appointment and a text confirmation.

She didn't spend that call thinking about artificial intelligence. She was thinking about her AC.

That's the scenario I keep in mind whenever an owner tells me, "My customers will hate talking to a robot."

The short answer

An AI receptionist is a voice agent that answers your business line, understands normal speech, answers questions from a knowledge base about your business, books appointments, and transfers calls to a person. Customer frustration comes from bad outcomes like loops, wrong answers, and no way out. A well-configured AI receptionist that's upfront about being AI usually beats the alternative most callers actually get: voicemail.

What does the research say about unanswered calls?

The research on caller attitudes toward AI is still young and mostly vendor-funded, so I don't lean on it. The research on what happens when nobody answers is much clearer.

• Lots of calls go unanswered. Small-business call studies put the share of inbound calls that go unanswered somewhere between roughly 40% and over 60%, depending on the study and the industry. For a crew that's on roofs or under houses all day, I'd expect the higher end.
• Most people don't leave a message and wait. A commonly cited industry figure says about 85% of callers who don't get through won't call back. Even if your number is half that, it's a lot of jobs going to whoever answered.
• Speed decides who wins the job. A Harvard Business Review study of 2,241 US companies found that responding within an hour made a business nearly 7 times more likely to qualify a lead. The same study found the average response time was about 42 hours.

Put those together and the real comparison is clear. Your AI receptionist isn't competing against you on your best day. It's competing against the ring-out at 6:30 pm.

What does it cost if customers do hang up?

Here's simple math you can redo with your own numbers (I go deeper on it in how much revenue small businesses lose from missed calls). Say you're an HVAC company that misses 15 calls a week after hours and during jobs. If a third of those are real service requests, that's 5 jobs. At an average ticket of $400, that's $2,000 a week, or roughly $104,000 a year, walking to a competitor.

Now flip the fear around. Suppose a few callers do dislike the AI and hang up. If the AI books even half of those 5 weekly jobs, you're still ahead by about $52,000 a year compared with voicemail. The risk of a few annoyed callers is real, but it's small next to the cost of silence.

Your numbers will be different. Plug in your missed calls, your booking rate, and your average job value. That's the diagnosis. Prescription without diagnosis is malpractice, and that applies to buying AI too.

Which businesses should worry about customer perception, and which shouldn't?

Strong fit: the trades. HVAC, plumbing, pest control, roofing, pool service, junk removal. The caller has a concrete problem and wants a time slot. A clear, fast answer reads as good service. Nobody calling about wasps in the eaves wants small talk.

Good fit with care: medical, dental, and med spa scheduling. Booking and FAQs work fine, but call data is protected health information, so HIPAA applies. Use a vendor that will sign a BAA.

Weak fit: businesses where the conversation is the product. Counselors, funeral homes, high-end custom builders mid-sale, attorneys taking a first call from someone in crisis. There, a human voice is part of what the customer is paying for. Use AI for overflow and after-hours message-taking at most.

Mixed: businesses with a lot of repeat regulars. A pool route customer who has talked to the same office manager for six years may be thrown by a new voice. The fix is to let the AI cover after-hours and overflow while your regular person still answers during the day.

Do customers really hate AI, or something else?

This is the part most articles get wrong. They either promise "customers can't tell the difference" or warn that "customers hate AI." Neither is useful.

Some callers can tell. That's fine. What callers hate is being stuck: a loop of "I didn't understand that," an answer that's wrong, or no way to reach a person. A poorly trained human receptionist produces the same frustration.

The other mistake I see is owners comparing the AI to themselves on their best day, picking up on the second ring and knowing every customer by name. That's the wrong baseline. Compare it to what callers get today when you're on a job, at dinner, or asleep.

On my own business line, FoundOnline runs an AI receptionist that transfers to my cell. We're early and don't publish client results yet, so I'll only speak to what I saw while testing it. The weak spots were questions the knowledge base didn't cover yet, where the agent had to say I'd call back. The fix was more and better answers in the knowledge base. A more human-sounding voice wouldn't have helped.

How do you set up an AI receptionist so customers don't mind it?

This is the sequence I use. Each step depends on the one before it.

1. Measure first. Count your calls, when they come in, and how many go unanswered. Your phone carrier or call log usually has this.
2. List the top ten questions. Pricing ranges, service area, hours, "do you handle X," "how soon can someone come out." Pull them from memory and a few days of notes.
3. Write an honest greeting. Say it's an AI assistant in the first sentence and say what it can do.
4. Build the knowledge base from those ten answers, in plain language.
5. Set transfer rules. Who gets the call, under what conditions, and what happens if they don't pick up (usually a text summary and a callback task).
6. Try to break it. Call it yourself, interrupt it, mumble, ask odd questions. Have your spouse or a tech call it too.
7. Soft launch after hours only. Read transcripts weekly and fix gaps.
8. Expand to overflow and daytime once the transcripts look clean.

If you record calls, add a recording disclosure. Texas is a one-party consent state, but California and some other states require all-party consent, and a disclosure covers you either way.

What's one thing I can do this week?

Call your own business after hours from a number you don't recognize. Listen like a customer. Do you hear twelve rings? A full mailbox? A greeting from 2019?

Then ask yourself honestly: would someone with a leaking water heater wait for your callback, or call the next company on Google?

One tactic that works: when you do set up AI, put the disclosure and the escape hatch in the first breath. "I'm the AI assistant. I can book a visit or get you to someone." Callers who know they can say "let me talk to a person" and actually get transferred tend to relax and finish the call. Knowing the door exists does a lot of the work.

What changed with AI that makes this different from phone trees?

Five years ago, talking to a machine meant press 1, press 2, and shouting "representative." That memory is why owners brace for customer backlash, and the backlash was earned.

Today's voice agents run on large language models. They understand a sentence like "the upstairs unit is blowing warm but downstairs is fine" without a menu. They handle interruptions, answer follow-up questions, and pull from what you've told them about your business. They can book on your calendar, send a text confirmation, and transfer to your cell.

The technology moved a lot. People's memories of bad phone trees didn't. That gap is why honesty and an easy human handoff matter more than the voice itself.

Should my AI receptionist say it's an AI?

Yes. Say it in the greeting. Callers handle AI fine when they aren't surprised by it, and disclosure builds more trust than trying to pass as human.

Can callers still reach a real person?

They should always be able to. A good setup transfers to your cell or office on request, or when the question is outside what the AI knows, and texts you a summary if you can't pick up.

Will older customers struggle with it?

Some might. Natural-language agents are easier than phone menus because there are no buttons to press. Keep the greeting short, and make "talk to a person" work on the first ask.

Is an AI receptionist better than an answering service for customer experience?

It depends on what your callers need. I compare the two side by side in AI receptionist vs. answering service.

The bottom line

Customers don't hate AI. They hate getting stuck and they hate voicemail. Be upfront that it's AI, train it on the questions people really ask, and give callers a quick way to a human. Do those three things and most callers will care about one thing: whether their problem got handled.

If you want to hear what this sounds like before deciding, call the live AI receptionist demo on foundonline.pro and judge it as a customer would.