Do restaurant guests mind an AI answering the phone?
Some guests do mind, and published surveys put the share anywhere from about a quarter to about two-thirds, depending on the question asked. What the research agrees on is what turns resistance into acceptance: tell callers it is automated, get the order right, and make a person easy to reach in one step. Judge any phone answer against those three, then check your own call log.
Published November 19, 2026 · Written by the team at Scaled Growth Agency · 5 min read
The short version
- Published surveys put discomfort with AI anywhere from 27% for restaurant AI (PAR Technology, 2026) to 64% for customer service in general (Gartner), so your own callers are the only reliable measure.
- Guests want to be told it is AI and to reach a person easily: 43% in the PAR survey ranked transparency first, and 44% said AI goes too far when they cannot reach a person.
- In the 2026 QSR Drive-Thru Report, voice AI order accuracy was 89% and satisfaction 95%, but 19% found it somewhat difficult or worse, up from 7%, so test accuracy first.
- Track early hang-ups, requests for a person and wrong tickets for a few weeks before and after any change to your phone answer.
- Put the way out in the greeting and pass the caller's details to staff so nobody has to repeat themselves.
How many guests object? It depends on how you ask
The share of people who object to AI in service settings runs from about a quarter to about two-thirds, depending on the survey. None of the surveys below measures a restaurant phone call itself, so treat them as the nearest evidence, not a forecast for your line.
The widest-net number is from Gartner. In a survey of 5,728 customers fielded in December 2023, Gartner found that 64% would prefer that companies not use AI for customer service, and 53% would consider switching to a competitor if they learned a company would use it. That survey covered customer service across industries, not restaurants.
The restaurant-specific number is friendlier. A June 2026 survey of 1,000 U.S. adults, commissioned by restaurant technology company PAR Technology and fielded through Dynata's consumer panel, reported that average discomfort with restaurant AI has fallen to 27%. The same survey found that 47% say AI goes too far when it replaces most human interaction, and only 7% believe a restaurant cannot be too automated. Pew Research Center's June 2025 survey of 5,023 U.S. adults found that 50% are more concerned than excited about the increased use of AI in daily life, up from 37% in 2021.
Worked through for 100 callers: Gartner's general figure would mean 64 prefer no AI, while PAR's restaurant figure would mean 27 are uncomfortable with it. The gap between 27 and 64 is the real finding. It says a sizable minority will want a person whatever you do, and that your own callers are the only reliable measure.
Guests want to be told, and to have a way out
The most consistent finding is not about the technology. Guests want to know what they are talking to and to be able to reach a person. In the PAR survey, 43% ranked transparency first when it comes to trusting AI use at restaurants, and 44% said AI goes too far when they cannot easily reach a person. Pew found that 76% say it is extremely or very important to be able to tell whether pictures, videos and text were made by AI or by people. That Pew question is about content, not phone calls, but the instinct is the same.
Gartner's advice on hand-offs applies directly to a phone line. Gartner analyst Keith McIntosh advises that AI chatbots tell customers when they are being connected to an agent, and that the agent should pick up where the chatbot left off. On a restaurant line, that means the person who takes over should already know the caller's name, what was ordered and why the call was passed along. A caller who has to start over is likely to hang up annoyed.
Routine ordering scores well, but accuracy and friction still show up
The best recent data on voice AI taking restaurant orders comes from drive-thrus, not phones, and it is mixed. The 2026 QSR Drive-Thru Report, a QSR Magazine study with data partner Intouch Insight, reviewed 95 targeted AI interactions plus 79 unplanned ones at four national chains, according to MarketScale's summary. In those voice AI encounters, order accuracy was 89%, up from 84%, and overall satisfaction was 95%, up from 91%.
The same report found that guests rating the AI easy or somewhat easy slipped to 53% from 56%, while those rating it somewhat difficult, or difficult and frustrating, rose to 19% from 7%. Read together, most guests are satisfied and a rising minority is struggling. By the report's own accuracy measure, 11 of every 100 encounters still had a mistake. A wrong phone order costs a remake or a refund and a guest who may not come back, so accuracy is the first number to test on your own line.
Keep the limits in view. A drive-thru is not a phone call, the sample is 174 interactions, and the report's definition of accuracy is its own.
How to find out what your own callers think
Ask them, and watch what they do, because your callers are not the national average. A few weeks of simple tracking will tell you more than any survey above. Start before you change anything, so you have a baseline to compare against.
- Early hang-ups. Pull your phone log and count calls that end within about ten seconds, before anything is said or ordered. A rise after a change to your phone answer is the earliest warning.
- Requests for a person. Count how often callers ask for staff, and how often the person who takes over can finish the job without making the caller repeat themselves.
- Wrong or remade tickets. Tally phone-order mistakes for a few weeks before and a few weeks after. Without a baseline, an accuracy number means nothing.
- Counter comments. Ask whoever hands over pickup bags to jot down any remark about the phone, good or bad, on a sheet by the register.
- Ten test calls. Call at a slow time, during a rush and after close. Order something with a modification, ask about hours, and ask for a person. Write down what you heard.
A greeting and hand-off that follow the research
Keep the greeting short, say what is answering, and name the way out in the first ten seconds. These steps come from the findings above plus ordinary host-stand practice, not from a study of restaurant greetings.
- Say it is automated. For example: 'Thanks for calling. I'm the restaurant's automated assistant. I can take a pickup order, book a table or answer questions.'
- Name the way out. Add: 'Say staff at any time to reach a person.' Then make sure someone can take that call, including at the Friday rush.
- Decide which calls always go to a person. Complaints, allergy questions, large parties and private events, an upset caller, and anything the system cannot settle after one try.
- Pass the context along. Staff should see or hear the caller's name, number, what was said and why the call was passed.
- Read the order back. Items, modifiers, total and pickup time, before the caller hangs up. It is the cheapest accuracy check there is.
- Offer a text. A texted menu or ordering link lets a caller who would rather not talk finish the job another way.
Where an AI answering service fits, and where it does not
Scaled Growth Agency is an AI phone answering service for restaurants. The assistant answers the restaurant phone in under a second, around the clock, including during a dinner rush. It takes pickup orders by voice and sends them into the Toast point of sale, so the kitchen ticket prints like an online order. It books reservations, answers menu, hours and directions questions from the restaurant's own menu, and can text the caller a link. It screens spam calls and hands a caller to staff when the call needs a person, and English and Spanish assistants are available.
Measured on our own restaurant lines in 2026, 92% of calls are handled start to finish by the assistant, and the rest are handed to staff on purpose. An average complete phone order takes 1 minute 57 seconds, 1 in 3 phone orders is paid by a card link before pickup, and 56% of callers say thank you to the assistant. That last figure is a habit, not a satisfaction score, and it does not replace asking your own guests. You can hear real calls on the homepage or read how the AI voice agent works.
Whatever answers your phone, a machine, a host or a service, the research points to the same test. Does the caller know what they are talking to, does the order come out right, and can they reach a person in one step? If your own test calls and call log say yes to all three, they will tell you more than any survey.
Frequently asked questions
Do restaurant guests like talking to an AI on the phone?
Opinion is split, and no published survey we found measures it for restaurant phone calls specifically. Gartner found 64% of customers would prefer companies not use AI for customer service, while PAR Technology's 2026 survey found average discomfort with restaurant AI at 27%. Both surveys point to the same fix: be open about it and make a person easy to reach.
Should I tell callers they are talking to an AI?
The research favors saying so. In PAR's 2026 survey, 43% ranked transparency first for trusting AI use at restaurants, and in Pew's 2025 survey 76% said it is extremely or very important to be able to tell AI-made content from human-made content. Legal requirements vary and are not covered here, so ask your own counsel.
What should happen when a caller asks for a person?
The call should reach staff in one step, and the staff member should already have the caller's name, number and what was discussed. Gartner's Keith McIntosh advises that AI tell the customer when a hand-off is happening and that the agent pick up where the AI left off.
How accurate is voice AI at taking restaurant orders?
The best recent published figure comes from drive-thrus: the 2026 QSR Drive-Thru Report, with Intouch Insight, put voice AI order accuracy at 89%, up from 84%. That leaves 11 of every 100 encounters with a mistake by the report's measure. A phone menu may behave differently, so run your own test calls and keep a baseline of your current error rate.
How can I tell whether my guests accept an automated phone answer?
Compare a few weeks before and after the change on three counts: calls that end within about ten seconds, requests for a person, and wrong or remade tickets. Add ten test calls at slow, busy and closed hours, and ask counter staff to note any guest comments about the phone.
Sources
- Majority Prefer Companies Not To Use AI For Customer Service: Gartner (Insurance-Canada.ca reprint of the Gartner release)
- Pew Research Center: How Americans View AI and Its Impact on People and Society
- Franchising.com: Diners Revise Their Thoughts About AI (PAR Technology 2026 diner survey)
- MarketScale: summary of the 2026 QSR Drive-Thru Report by QSR Magazine and Intouch Insight