Key takeaways
- Half of US adults are more concerned than excited about AI in daily life; only 10% are more excited than concerned[1].
- Inbound calls are a different case from robocalls: the homeowner called you and wants a problem solved.
- Acceptance depends mostly on the call going well: fast, accurate questions, honest disclosure and an easy route to a person.
- Measure it yourself with early hang-ups, requests for a person, completed calls and what reviews say.
What the surveys can and can't tell you
There is no large, independent survey that asks homeowners specifically how they feel about an AI answering a plumber's phone. Vendor surveys exist, but they are designed to sell. What we do have is solid public-opinion data on AI in general, on how Americans handle phone calls, and on how they use reviews. Put together, those give a realistic picture, and point to what you should measure in your own calls.
The honest summary: Americans are cautious about AI, very cautious about unknown phone calls, and quick to punish a bad experience in reviews. None of that means homeowners refuse to talk to an AI when they have a leak. It means the call has to go well, and the first ten seconds matter.
Americans are wary of AI in general
In a June 2025 Pew Research Center survey of 5,023 US adults, 50% said they were more concerned than excited about the increased use of AI in daily life, 38% were equally concerned and excited, and 10% were more excited than concerned[1]. The concerned share was up from 37% in 2021[1]. The same survey found 76% think it is extremely or very important to be able to tell whether content was made by AI or by people[1].
- More concerned than excited 50%
- Equally concerned and excited 38%
- More excited than concerned 10%
Source: Pew Research Center, 2025[1]
Show the data
| Item | Value |
|---|---|
| More concerned than excited | 50% |
| Equally concerned and excited | 38% |
| More excited than concerned | 10% |
Read that carefully. Concern is not refusal. Most of those people use AI-powered tools daily, from map routing to spam filters to voice assistants on their phones. What the numbers show is a public that wants transparency and control. A plumbing company that discloses AI clearly and makes a person easy to reach is working with that instinct, not against it.
They are warier still of phone calls
Americans have learned to distrust the phone. Pew found that eight in ten don't generally answer their cellphone when an unknown number calls, and only 19% generally pick up such calls[2]. Years of robocalls have taught people that an unexpected call from a machine is probably a scam.
That history explains much of the resistance to AI voices, and it is why outbound AI calls face a much harder audience than inbound ones. When your AI calls a homeowner out of the blue, it is competing with every scam call they have ever received. When a homeowner calls you, the situation is reversed.
Why inbound calls are a different case
A homeowner calling a plumber has chosen your business, usually because something is wrong. Their goal is narrow: get someone to come, at a time that works, for a price they can accept. What frustrates them is not the technology but the obstacles: being put on hold, reaching voicemail, being asked the same thing twice, or being told someone will call back and then waiting.
An AI that answers on the first ring, asks sensible questions, confirms the address and gets the on-call plumber moving is removing those obstacles. Callers in that situation often care far less about who answered than whether help is coming. The same AI that is merely tolerated on a routine booking can be genuinely welcome at 11 pm with water on the floor, because the alternative is a voicemail box.
What raises and lowers acceptance
| Factor | Raises acceptance | Lowers acceptance |
|---|---|---|
| Greeting | Business name, honest AI line, straight to the problem | Long scripted welcome, no disclosure |
| Speed | Answers on the first rings, short questions | Menus, long pauses, repeated questions |
| Understanding | Gets the problem right the first time and reads back the address | Mishears and makes the caller spell everything |
| Honesty | Says when it does not know and what happens next | Guesses prices or arrival times |
| A person | Offered without argument, at any point | Refuses or loops the caller |
| Outcome | Booked time or a tech on the way | A vague promise of a callback |
| Language | Continues in the caller's language | Forces English on a Spanish speaker |
Most of these are within your control. The greeting, the questions, the approved answers and the route to a person are all things you write and test. Our guide on whether to tell callers they are talking to AI includes greeting scripts.
Older callers deserve a careful design
Plumbing customers skew toward homeowners, and many are older. Pew data shows 96% of adults aged 65 and older own a cellphone and 79% own a smartphone[3], so they can reach you easily, but they may be less used to voice assistants and more sensitive to fast speech or unclear prompts.
- Speak at a measured pace and ask one question at a time.
- Repeat the address and callback number slowly, digit by digit.
- Offer a person early if the caller sounds confused or asks twice for something.
- Avoid jargon such as diagnostic fee without a short explanation.
- Never end the call until the caller has heard clearly what happens next.
The reputation risk is lopsided
One bad AI call can cost more than many good ones earn. Pew found that 54% of Americans who read online reviews pay more attention to extremely negative reviews, compared with 43% who pay more attention to extremely positive ones[4]. A review saying a robot answered and nobody came will be read closely by your next customer. A review saying the call was handled quickly will rarely mention AI at all.
That asymmetry is the strongest argument for testing before launch and reviewing calls afterward. It is also why the no-answer step matters: the calls that turn into angry reviews are usually ones where the caller was promised help that did not arrive.
Measure acceptance in your own calls
Early hang-ups
Track the share of calls that end within 15 seconds of the greeting. Compare it with your old voicemail or answering-service rate.
Requests for a person
Count how often callers ask for a human, and at what point in the call. Early requests point to the greeting; late ones point to a question the agent could not handle.
Completed calls
Measure the share of calls that end with a booking, an urgent transfer or a clear handover, rather than an abandoned call.
Complaints and reviews
Tag any complaint or review that mentions the phone experience, positive or negative.
Listen weekly
Play five random calls in your weekly meeting. Tone problems show up in recordings long before they show up in numbers.
What to change when the numbers look bad
If early hang-ups or requests for a person are higher than you expected, the cause is usually specific and fixable. Work through these in order, changing one thing at a time so you can see what helped:
- Shorten the greeting. Anything longer than two sentences before the first question loses impatient callers.
- Ask about the problem first, then contact details. Callers relax once they feel heard.
- Check the voice speed and the pause before the agent answers; slow responses feel robotic even when the words are right.
- Look at where in the call people ask for a person, and write approved answers for the questions that trigger it.
- Make the handover faster for existing customers with billing or warranty questions, who often prefer a person.
- Listen for mispronounced local street names and town names, and add phonetic notes to the knowledge base.
Give each change a week before judging it. Call volume varies by day and weather, so compare like with like: weeknights with weeknights, storm weeks with storm weeks.
Example: reading the first month
Example: in its first month of AI after-hours answering, a shop sees early hang-ups on 6 of 120 calls, compared with roughly 30 of 110 calls that reached voicemail in the same month last year. Callers asked for a person on 14 calls, mostly existing customers asking about invoices. The owner adds an approved answer about invoice questions and a faster handover for billing. Two reviews mention the phone, both positive about the speed of response. The shop keeps the setup and widens it to lunchtime overflow.
Frequently asked questions
Do customers hate talking to AI?
Surveys show most Americans are cautious about AI, but inbound callers with a plumbing problem mainly want help. Acceptance depends on a fast, accurate, honest call with an easy route to a person.
Are older customers less accepting?
They can be less used to voice assistants. Speak slowly, ask one question at a time, read numbers back digit by digit and offer a person early if they seem confused.
Will AI hurt my reviews?
A bad call can, because negative reviews get more attention than positive ones. Test thoroughly before launch and review a sample of calls every week.
How do I know if my callers accept it?
Track early hang-ups, requests for a person, completed calls and review mentions, and listen to a few calls every week.
References
Pew Research Center (2025). How Americans View AI and Its Impact on People and Society (opens in a new tab) โ
Pew Research Center (2020). Most Americans don't answer cellphone calls from unknown numbers (opens in a new tab) โ
Pew Research Center (2026). Mobile Fact Sheet (opens in a new tab) โ
Pew Research Center (2016). Online Reviews (Online Shopping and E-Commerce) (opens in a new tab) โ