# Will Customers Hang Up on an AI Receptionist?

> Canonical: https://dialzara.com/blog/will-customers-hang-up-on-ai-receptionist  
> Published: 2026-11-16  
> Summary: Some callers still prefer a person, and a minority will hang up. Most will stay if the greeting is clear, the pause is short, and they can reach a human.

"If I put an AI on the phone, will people hang up?" That's the question we hear most from owners who are otherwise ready to try an AI receptionist. A few callers will. Most will stay if the first few seconds feel like a competent front desk, not a maze, and if they can reach a person when they need one.

This post looks at what independent surveys actually show, why the hang-up numbers disagree, and how to set up the greeting, the pause, and the human handoff so more callers stay. We'll also be honest about the limits of the data: most of it is about chatbots, large contact centers, or people who already knew they were talking to AI, not a small business's after-hours line.

> **The bottom line**
> 
> -   **A minority of callers will hang up** when they realize it's AI. One 2025 survey put that share at 29%; a 2026 voice-agent study put immediate hang-ups at 11%, with about 30% skipping the AI and asking for a person.
> -   **Most people will give an AI a chance** if they know they can reach a human and if the AI can actually finish the job, not just collect a name.
> -   **People are less comfortable with AI on the phone than in website chat**, so a voice greeting has to work harder than a chatbot banner.
> -   **Compare AI to voicemail, not to a perfect receptionist.** In one older business-line study, 69% of callers hung up rather than leave a voicemail. An AI that answers, takes a message, and offers a transfer or callback is usually the better of those two options.
> -   **You can reduce hang-ups** with a short, natural greeting, almost no dead air, a brief disclosure, and a reliable path to a human.

## Will customers hang up on an AI receptionist? What the research says

There isn't a single official hang-up rate for small-business AI receptionists. The studies below come from reputable research firms, but they ask different questions of different people. Treat the percentages as a range, not a promise of what your line will do.

### Independent public opinion is mixed, not a panic

Pew Research Center is the most independent source here, and it's also the most cautious. In a [survey of 5,410 U.S. adults in August 2024](https://www.pewresearch.org/wp-content/uploads/sites/20/2025/03/pi_2025.04.03_us-public-and-ai-experts_topline.pdf), Pew asked whether AI would do better, worse, or about the same as people whose job it is to provide customer service. Only 19% said AI would do this better, 47% said worse, 19% said about the same, and 16% weren't sure.

That is not the same as "47% will hang up." It's a general opinion about how well AI would do the job, not a measure of who stays on the line. In a [June 2025 follow-up](https://www.pewresearch.org/short-reads/2025/09/17/from-political-speeches-to-songs-how-would-americans-react-if-they-found-out-ai-was-involved/) (5,023 U.S. adults), Pew asked people to imagine a helpful customer-service chat after which they learned they had been talking with an AI chatbot. 41% said that would make them feel worse about the experience, 53% said it would not change how they felt, and 6% said it would make them feel better. Americans aren't panicked. They're skeptical, and they'd rather not be surprised after the fact.

### Finding out later that it was AI can sour the experience

A [No Jitter report on Metrigy's Q1 2026 Consumer CX Index](https://www.nojitter.com/contact-centers/how-ai-voice-agents-handle-peak-customer-demand) (1,000 consumers in North America) found that when a company didn't disclose that a voice agent was AI, 64.6% of people who later found out said it had a somewhat or very negative impact on their experience. Only 35.4% said it made no difference, and almost no one said it was a positive surprise. So "sound so human they never notice" is a risky strategy, even if the voice is excellent. For more on that judgment, see our post on [the risks of talking to AI on phone calls](https://dialzara.com/blog/talking-to-ai-phone-calls-and-how-to-avoid-the-risks).

### The closest "will they hang up?" numbers still disagree

The most direct "will they hang up?" number we found comes from a [ServiceForge / OnePoll survey of 6,000 adults in the U.S., Canada, and the U.K. in October 2025](https://assets.serviceforge.com/serviceforge/pdf/ai-survey-report-serviceforge.pdf). When asked what they would do "the first time they heard an AI voice answer a company's phone," 29% said they would hang up, 33% said they would continue the call, and 33% said they weren't sure or it would depend on the situation (the published figures add to 95%; the remainder is rounding). ServiceForge, which sells an AI answering product, summarized that as "1 in 3." That's a useful headline and a worst-case reading of a first-impression question. It does not measure what people actually did on a live call, and the "not sure / it depends" group is as large as the "I'd stay" group.

Metrigy's 2026 consumer study, reported by [No Jitter](https://www.nojitter.com/contact-centers/how-ai-voice-agents-handle-peak-customer-demand) and in [Metrigy's own research](https://www.metrigy.com/product/consumer-cx-index-q1-2026/), is more encouraging and more specific to voice agents. 59.1% of consumers said they were willing to give an AI voice agent some time to solve their issue, but only if they know they can escalate to a human. About 30% skip that "give it a try" step and jump straight to a person. 11% hang up the moment they realize they aren't speaking to a person.

Those two studies don't cancel each other out. One asked a first-impression question to a general online sample; the other asked about behavior with voice agents among people thinking about customer-service calls. Together they say: a real minority will bail immediately, a larger group will stay if a human is available, and a sizable group is still making up its mind in the first few seconds.

### People are less comfortable with AI on the phone than in chat

A [Vonage survey of 1,000 U.S. adults in September 2024](https://www.vonage.com/about-us/newsroom/press-releases/Vonage-Survey-Reveals-Increased-Acceptance-of-AI-in-Customer-Support-with-Nearly-Two-Thirds-of-Consumers-Admitting-More-Efficient-Interactions/f12589bd-c842-4491-9967-6766f38aba91/) (conducted with Pollfish; respondents had used a brand's AI-powered customer service in the prior 12 months) found that 87% were comfortable using AI-powered website chat, but only 33% were comfortable using an AI-powered voice assistant. 91% said companies should tell them when they're interacting with an AI rather than a human. 85% said it was "very important" that an AI-powered system make it easy to reach a live agent.

That's the gap you have to close. People will type to a bot on a website who would hesitate to talk to one on the phone. A voice greeting that sounds like a cheap menu will push those hesitant callers toward the door.

### Willingness goes up when the AI can finish the job and a human is available

When the question is "would you use this if it actually helped?", the numbers rise. A [PolyAI survey of 1,000 U.S. consumers, published February 2025](https://poly.ai/blog/hear-me-out-consumer-study-reveals-phone-still-reigns-in-customer-service) (conducted with Dynata), found that 71% were willing to interact with an AI voice assistant if it could fulfill their request. The same study found that 55% of consumers are likely to ask for a customer service representative immediately if they hear a robotic Interactive Voice Response (IVR) system. That's a point against old-fashioned phone trees, not against a natural conversation, but it explains why a stiff, menu-like greeting will lose people fast.

A [Salesforce survey of 15,015 consumers in 18 countries, conducted with YouGov in July and August 2024](https://www.salesforce.com/news/stories/ai-customer-research/), found that nearly 75% of consumers want to know if they're interacting with an AI agent, and 45% said they would be more likely to use AI-powered customer service if they had a clear path to escalate to a human. [Twilio's Inside the Conversational AI Revolution report](https://www.twilio.com/en-us/report/Inside-the-Conversational-AI-Revolution) (4,800 consumers in 15 countries, August–October 2025) found that 78% said the ability to switch to a human is "critical" to trusting a company using AI, and 72% would gladly use AI if it were faster than a human. The same report found that 40% of consumers say AI "repeats itself or gets stuck in loops," and that 39% say they won't wait at all for an AI agent, versus 88% who will wait for a human.

The pattern is consistent across vendors: disclosure, competence, and a real human backup turn skepticism into use. Loops, delays, and no escape hatch do the opposite.

### Compare AI to voicemail, not to a perfect receptionist

The fair comparison for a shop that already misses calls is not "AI versus a warm person who always picks up." It's "AI versus voicemail or a missed call."

A [Moneypenny UK Small Business Call Report from November 2016](https://resources.moneypenny.com/wp-content/uploads/2020/01/Value_of_a_Call_Report.pdf) found that 69% of callers who were answered by voicemail declined to leave a message and hung up. That study is older, vendor-sponsored, and based on UK micro-businesses (a survey of 300 plus call data from 10,000 businesses), so treat it as directional, not current U.S. gospel. The direction still matches what most owners already know: a lot of people will not talk to a machine that only records a message. For more on that side of the math, see our post on [the cost of a missed call](https://dialzara.com/blog/missed-calls-hidden-costs-and-ai-solutions), and [whether missed calls affect Google ranking](https://dialzara.com/blog/do-missed-calls-affect-google-ranking).

An AI receptionist that answers, takes details, books an appointment, or offers a transfer or callback is a different offer than voicemail. Some callers will still prefer a person, and you should make that easy. But "a minority may hang up" is usually a better outcome than "most people who hit voicemail never leave a message."

## Why callers hang up or ask for a person right away

From the research above and from the calls we review, hang-ups and "let me talk to someone" requests usually come from a short list of frustrations, not from a blanket refusal to use AI:

-   **They didn't expect a machine.** A stiff or overly casual opener, or no mention of AI when the voice is clearly not a person, makes people feel tricked. Salesforce, Vonage, and Metrigy all show that surprise disclosure later is worse than a brief, early one.
-   **The pause is too long.** After someone says why they called, even a couple of seconds of silence feels like the system didn't hear them. Twilio's finding that 39% won't wait at all for an AI agent, versus 88% who will wait for a human, is the hard version of this.
-   **It sounds like a phone tree.** "Press 1" energy, or a robotic IVR, is what drove 55% of PolyAI's respondents to ask for a representative immediately. A conversation that actually answers the question is a different experience.
-   **They can't get to a person.** 85% of Vonage's respondents said easy access to a live agent is very important. 78% of Twilio's said the ability to switch to a human is critical to trust. If "representative" or "someone in the office" leads nowhere, the next move is to hang up.
-   **The AI gets stuck.** Repeating the same question, failing to book the appointment, or looping after a simple request is the 40% complaint in Twilio's report. People will give a system one or two chances, not five.
-   **The call feels high-stakes, personal, or already frustrating.** Billing disputes, complaints, and "I need to speak to the owner" calls are where even willing callers ask for a human. That's not a failure of the AI. That's the AI doing its job by getting out of the way.

If you run a trade shop, this is the same worry that comes up on a job site.

## Practical ways to reduce hang-ups

You can't talk every skeptical caller into staying, and you shouldn't try to trap them. You can remove the reasons people leave in the first ten seconds.

### Write a greeting that sounds like a helpful front desk, not a phone tree

The first sentence has to do three jobs: identify the business, set a calm tone, and invite the caller to say why they called. A fourth job, disclosure, belongs in the same breath if you're going to do it (and we think you should). Keep it short. Here's a pattern that works:

> "Thanks for calling Ridgeline Electric. This is the after-hours desk, an AI assistant. I can take a message, book a visit, or connect you with the on-call tech. How can I help you?"

That's about six seconds. The caller knows who they reached, that it's AI, and that a person is available. Then you stop talking and listen. For more greeting and prompt language, see [best AI voice receptionist prompts](https://dialzara.com/blog/best-ai-voice-receptionist-prompts).

On Dialzara, the greeting is fully customizable. You can refine the agent with example conversations and real-time feedback during test calls (see the [Dialzara FAQs](https://dialzara.com/faqs)). Pick a voice that matches your shop. If your customers expect a local, conversational tone, don't choose the most formal option in the library. Preview the voices at [dialzara.com/voices](https://dialzara.com/voices).

### Cut the dead air

Set the agent to start speaking quickly, and write replies that stay short until the caller has finished explaining. If the AI needs a moment to look up a slot, have it say so: "Let me check the calendar." Silence is what makes people think the call dropped. A brief status line is better than a perfect, delayed answer.

### Say it's AI, briefly, then help

You don't need a legal monologue. One clause is enough: "I'm an AI assistant," or "this is the after-hours AI desk." Then do the work. The research is aligned here: most people want to know, and finding out later is what sours the experience. We've written more on how to do this without making the call awkward in [how to disclose AI use on customer calls](https://dialzara.com/blog/disclose-ai-use-customer-calls).

What you should not do is hide it, or spend the first twenty seconds apologizing for being a computer. Callers came to get something done. Disclose, then help.

### Make it easy to reach a human, and actually connect them

Build a clear escape hatch into the greeting and the prompt: if they ask for a person, a technician, or "someone in the office," transfer them or take a message and schedule a callback. Don't make them say it three times. Don't pretend the AI can handle a complaint it's going to lose anyway. For the mechanics, see [best practices for AI call transfers](https://dialzara.com/blog/best-practices-ai-call-transfers).

On Dialzara, you can transfer a live call to a teammate, or have the agent take a message and trigger a callback. The [FAQs](https://dialzara.com/faqs) recommend a direct instruction for cases the AI can't finish, for example: "If the caller asks something you can't answer, tell them you'll pass it to a specialist who will follow up." You can also enable live-call transfer to a human teammate on request.

### Don't trap people in a loop

If the agent can't book the time, find the order, or answer the question after one clarification, it should offer a human path. That's how you avoid the "repeats itself or gets stuck in loops" complaint that 40% of Twilio's respondents described. Test this on purpose: call in with a messy request, an after-hours emergency, and a "I need to talk to Mike" request, and listen for loops.

### Measure hang-ups on your own line

The published surveys are a starting point, not your scoreboard. In Dialzara, listen to recordings and read the summaries after a handled call. Watch for calls that end in the first 15 seconds, callers who ask for a person and don't get one, and greetings that get talked over. Adjust the script, then check the next week's calls. Your number will be more useful than anyone's survey.

## Getting started with Dialzara

If you want callers to reach a live answer instead of voicemail, you can try Dialzara on a 7-day free trial. Per the [pricing page](https://dialzara.com/pricing) (November 2026):

-   Inbound plans start at **$29/month** (Business Lite, 60 minutes), then Pro at $99 (220 minutes), Plus at $199 (500 minutes), and Elite at $349 (1,000 minutes).
-   Every inbound voice plan has a 7-day free trial, with no setup fees, no long-term contract, and cancel anytime.
-   Every inbound plan includes 24/7 availability, call recordings, call summaries, email/SMS notifications after each handled call, and 10 languages, including English and Spanish.
-   You can forward or port your existing number. Porting is free.

Set the greeting, the transfer rules, and a test call before you point your published number at the agent. If you'd rather keep your current number, here's [how to forward calls to an AI receptionist](https://dialzara.com/blog/forward-calls-to-ai-receptionist).

Learn [how the AI receptionist works](https://dialzara.com/solutions/ai-receptionist), or [book a demo](https://dialzara.com/meet).

## Frequently asked questions

### Will customers hang up if they reach an AI receptionist?

Some will, especially on a first impression. One 2025 survey found 29% said they would hang up when they first heard an AI voice; a 2026 voice-agent study found 11% hang up immediately, while 59% will give the AI time if they know they can reach a human. The rest stay, ask for a person, or decide in the first few seconds. A clear greeting and a real human option reduce hang-ups.

### Do customers mind talking to an AI answering service?

Many are cautious, especially on the phone. People are much more comfortable with website chat than with an AI voice. They mind being stuck, surprised, or unable to reach a person more than they mind a competent assistant that answers quickly and can hand the call off.

### Does an AI receptionist sound human enough that people will stay on the line?

Modern AI voices can sound very natural, but sounding human is not enough. If the pause is long, the greeting feels like a menu, or there's no way to reach a person, callers will still leave. A natural voice plus a short, useful conversation is what keeps people on the line.

### Should I tell callers they're talking to AI?

Yes. Independent and industry surveys consistently find that most people want to know, and that finding out later can sour the experience. A brief line in the greeting, "I'm an AI assistant," is enough. Then help them with the reason they called.

### What if a caller asks for a real person?

Transfer them, or take a message and schedule a callback. Don't make them ask twice. You can set this as a rule in the agent's instructions so "representative," "technician," or "someone in the office" starts a handoff right away.

### How do I know if callers are hanging up on my AI receptionist?

Listen to recordings and read the call summaries. Short calls that end during the greeting, repeated "hello?", or a request for a person that never gets fulfilled are the usual signs. Check a week of calls after you change the greeting to see if the pattern improves.

### Will clients talk to an AI receptionist, or just hang up?

Most will talk if the greeting is clear and a human is available. Surveys of people willing to use an AI voice assistant when it can complete the request come in around 7 in 10. The callers who hang up immediately are a minority, and you can keep that group small with a better opening and a real escape hatch.

### Is an AI receptionist worse than voicemail for hang-ups?

Usually no. An older study of small-business lines found 69% of callers hung up rather than leave a voicemail. An AI that answers, takes details, and can transfer or book is a better offer than a recording that only asks for a message, even if a minority of callers still prefer a person.

---

_By Dialzara Team._
