The best AI prompts for restaurants and negative reviews, 21 to copy (2026)

To answer a negative review with AI, give the assistant the review text, your restaurant’s tone and the specific thing that upset the guest. Review the draft before posting and close with an invitation to come back. This library gathers 21 prompts to copy, and in the Founderz AI program you learn to apply them with your team.

Add Founderz as a preferred source

What you’ll take from this

This guide gathers tested templates for answering Google, TheFork and Tripadvisor reviews, and this is what you take away:

• 21 prompts to copy and paste into ChatGPT, Claude, Copilot and Gemini: negative, positive, doubtful reviews and monthly analysis.
• What a review reply prompt needs: context, tone, the specific fact and a call to book.
• How to spot patterns (waits, allergens, service) in your Google, TheFork and Tripadvisor reviews.
• What changes in each assistant and which guest data you should not paste.
• A method to write your own prompts and a ten-minute daily routine.

Managing reviews is a daily task for any restaurant. An unanswered complaint weighs more than it seems: whoever reads it decides whether to book or look elsewhere. The problem is not the complaint but the time it takes to reply well when you have lunch service at two and suppliers waiting. According to a BrightLocal study (2023), 88% of consumers say an owner’s reply to a negative review influences their decision to visit. With fifteen minutes a day and a good prompt, artificial intelligence takes the writing off your plate and leaves you the judgment.

They work for the restaurant owner with a single venue and for restaurant owners and managers running several: anyone handling customer reviews, online reviews on Yelp or Google, and customer feedback can quickly see what each guest asks for and reply in a realistic way, without sounding like a template. Use the AI tools below for restaurant management in 2026, keep your brand voice and avoid repeating the same phrase in every reply.

Swap what is in brackets: cuisine, venue name, dish, tone and the review text. For a similar approach in another sector, see the AI prompts for hotel reviews and upselling emails.

ChatGPT prompts for restaurant review replies

ChatGPT is the most used assistant for reviews: it drafts fast, adjusts tone and accepts files with your exported reviews. These prompts cover the most common situations on Google Business Profile, TheFork and Tripadvisor.

Negative reviews

1. Negative review about the wait

Act as the floor manager of a [type of cuisine] restaurant in [city].
Context: the guest complained they waited [time] for a table or a dish. Our busiest hours are [Friday and Saturday 9:00 to 10:30 pm].
Task: write a public reply to this review. Acknowledge the wait, apologize in one sentence without long excuses and invite them to book ahead next time.
Constraints: [warm and honest] tone, 80 words maximum, do not blame other guests or the kitchen, do not promise discounts.
Format: a single paragraph ready to post, signed as [name and role].
Review: [paste the text].

2. Negative review about a dish

You are the floor manager of [restaurant name], a [type of cuisine] restaurant.
Context: the guest criticized the dish [dish name] (for example: "the rice was overcooked" or "the steak arrived cold").
Task: write a reply that thanks them for the feedback, acknowledges the experience without arguing with their taste and explains you have passed it to the kitchen team.
Constraints: do not get defensive, do not justify the recipe, offer that the kitchen team will look after their next booking, 90 words maximum.
Format: the final reply and, below it, one line with a shorter alternative.
Review: [paste the text].

3. Negative review about the service

Reply as the manager of [restaurant name] to this review about table service.
Context: the specific issue was [distracted server / forgotten order / curt manner]. I have not spoken to the team yet, so do not state what happened.
Task: acknowledge the specific failure, say you will review it with the team and invite the guest to write to [email or phone] so you can look after them personally.
Constraints: sober, responsible tone, do not name the employee, no excuses, 80 words maximum.
Format: reply ready to post on Google.
Review: [paste the text].

4. Review about allergens or food safety

Act as the director of a [type of cuisine] restaurant. A guest describes in their review a problem with [allergen or suspected food poisoning].
Task: write a brief public reply that shows we take it seriously, apologizes and offers a direct contact to investigate.
Constraints: do not admit legal fault or deny the facts, give no medical details, do not mention the guest's name, 70 words maximum. Mark with the word DOUBT any sentence a person must validate before posting.
Format: the public reply and, separately, three internal questions I should ask the kitchen and floor staff.
Review: [paste the text].

5. Unfair or doubtful one-star review

Act as the manager of [restaurant name]. I think this one-star review is unfair or does not come from a real guest (I cannot find a booking for that day).
Task: write a polite public reply saying we cannot locate the visit, asking for more details via [contact channel] and making clear we want to fix it.
Constraints: do not accuse the author of lying, no irony, 60 words maximum.
Format: the public reply and, separately, the steps to ask Google to review the review if it breaks its policies.
Review: [paste the text].

Positive and ambiguous reviews

6. Positive five-star review

You are the owner of [restaurant name], a [type of cuisine] restaurant.
Task: reply to this 5-star review with a grateful note that mentions the detail the guest highlighted and invites them back to try [seasonal dish or daily menu].
Constraints: 50 words maximum, do not repeat the thank-you phrase used in recent replies ([last phrase used]), no exaggeration.
Format: two variants in different tones (warm and more formal).
Review: [paste the text].

7. Ambiguous review: ask for more context

Act as the manager of [restaurant name]. This [2 or 3]-star review does not explain what went wrong.
Task: reply kindly asking for more detail about what happened so you can fix it, offer a direct contact channel ([email or WhatsApp]) and invite them to book once it is resolved.
Constraints: warm tone, do not assume what the problem was, 60 words maximum.
Format: one paragraph ready to post.
Review: [paste the text].

8. Analyze a month of reviews from a file

Act as a reputation analyst for a [type of cuisine] restaurant. I am attaching a file with [month]'s reviews from Google, TheFork and Tripadvisor.
Task: group the reviews by theme (food, service, wait, price, atmosphere, allergens), count how many are positive and negative in each theme and identify the three most mentioned dishes.
Constraints: do not invent data that is not in the file; if a pattern rests on few reviews, say so; flag whether wait complaints cluster in a time slot.
Format: table with theme, number of mentions, tone and a short quote, followed by three actions for next week.

Claude prompts for analyzing and answering reviews

Claude fits when you work with long texts: a year of reviews, a tone manual or several replies you want to draft carefully. It accepts long documents in a single request, so you can paste everything at once.

9. Quarterly reputation summary for management

You are a hospitality consultant. I am pasting [number] reviews of [restaurant] from the last three months: [paste the reviews].
Task: write a one-page report for the owner with what works best, the three problems that repeat most and their likely cause.
Constraints: base every claim on specific reviews (quote a short phrase), distinguish repeated facts from isolated opinions and do not propose spending without justifying it.
Format: title, three blocks (what works, what fails, what I would do) and a final list of decisions that depend on me.

10. Tone guide for replying to reviews

I am attaching 10 review replies I think are well written and 5 that feel too cold: [paste the examples].
Task: extract the voice of [restaurant name] and write a one-page tone guide so anyone on the team replies the same way.
Constraints: include opening and closing phrases we do use, phrases we avoid and how to handle a one- to three-star review.
Format: guide with short sections and three "before and after" examples improving my cold replies.

11. Batch replies to the week’s negative reviews

Act as the manager of [restaurant name], a [type of cuisine] restaurant. I am pasting the [5] most recent negative reviews, separated by "---": [paste the reviews].
Task: write a different reply for each, citing the specific issue in each review and closing with an invitation to book again.
Constraints: [warm] tone, 80 words maximum per reply, none starting the same way, no excuses or discounts.
Format: numbered list in the same order as the reviews and, at the end, which review deserves a personal call from me.

Copilot prompts for managing reviews inside Microsoft 365

Copilot works inside Excel, Outlook, Word and Teams with the files in your work account. If you already run the restaurant on Microsoft 365, it is the easiest way to handle reviews without leaving your tools.

12. Analyze the reviews spreadsheet

Analyze the table in [sheet name] with [month]'s reviews: columns date, platform, rating and comment.
Task: create a pivot table with average rating by platform and by week, and a column classifying each comment as food, service, wait, price or atmosphere.
Constraints: do not change the original data, work in a new sheet and explain the formula you use in each column.
Format: new sheet with the table and three conclusions in text below.

13. Write in Outlook to the guest who left the review

Draft an email to [guest name], who left a [1- or 2]-star review about [reason] on [date]. Use the booking email I have in [folder or thread].
Task: apologize for [specific issue], tell them in one sentence what we have changed and invite them for a second visit with [agreed gesture, for example an appetizer].
Constraints: warm yet formal tone, 120 words maximum, do not ask them to delete the review.
Format: subject and email body, signed as [name and role].

14. Weekly review summary for the team meeting in Teams

Use the file [name of the reviews spreadsheet] and this week's chat in the [channel name] channel.
Task: prepare a five-point summary for Monday's meeting: reviews received, average rating, the most repeated praise, the most repeated complaint and who handles it.
Constraints: do not quote guest names, 150 words maximum, mark as unowned anything without an owner.
Format: bulleted list with one assigned task per complaint.

Gemini prompts for reviews on your Google listing

Gemini fits if your restaurant lives in the Google ecosystem: your Google Business Profile listing, Gmail, Docs and Sheets. These prompts take advantage of moving the result straight into a document or a sheet.

15. Reply to Google reviews in your listing’s tone

Act as the manager of [restaurant name], in [neighborhood and city]. This Google review has [rating] stars.
Task: write the public reply. Mention the specific fact in the review, naturally include the phrase [type of cuisine restaurant in neighborhood] only once and invite them to book via Google or WhatsApp.
Constraints: no more than 75 words, no hashtags or emojis, no links.
Format: reply ready to paste into the listing.
Review: [paste the text].

16. Turn reviews into a tracking sheet

I am pasting [month]'s Google reviews: [paste the reviews].
Task: create a table for Sheets with the columns date, stars, main theme, dish mentioned, specific issue and reply status (unanswered, answered, escalated).
Constraints: one row per review, do not merge several reviews into one, leave a cell empty if the data does not appear.
Format: tab-separated table, ready to paste into a sheet.

17. Reply protocol in a document for the team

Act as the operations manager of [restaurant name].
Task: write in a document a one-page protocol for answering reviews: who reads them, at what time ([before 12:00]), maximum reply time ([24 hours]), which reviews are escalated to the owner and which guest data is never mentioned.
Constraints: short sentences, one owner per step, no legal language.
Format: title, numbered list of steps and a final box with three model sentences.

More AI prompts for restaurants: asking for reviews, social media and chaining steps

Beyond replying, AI helps you ask for reviews, reuse the good ones on social media and check your own replies before posting.

18. Ask satisfied guests for reviews

Act as the customer care lead of [restaurant name].
Task: write three short messages asking for a Google review from guests who just dined with us, one to hand over at the table, one for WhatsApp and one for the thank-you email.
Constraints: do not offer an incentive in exchange for the review, do not ask for five stars, include the link [listing review link], 35 words maximum each.
Format: list of three messages with the channel in bold.

19. Turn a good review into an Instagram post

Act as the community manager of [restaurant name], a [type of cuisine] in [neighborhood].
Context: a guest left this 5-star review and has given permission to quote it: [paste the text].
Task: write the text of an Instagram post that quotes one phrase from the review, without naming the guest, and invites people to book for [dish or event].
Constraints: 60 words maximum, warm tone, three hashtags, no promises we cannot keep.
Format: post text and a suggestion for a real photo of the dish.

20. Chain prompts: extract themes and prioritize

We will work in three steps with [month]'s reviews for [restaurant]. Do not move to the next step until I say "continue".
Step 1: read these reviews and list the themes that appear, ordered by frequency: [paste the reviews].
Step 2: group the comments by category (food, service, wait, price, atmosphere) and tell me which are repeated complaints.
Step 3: for the [five] most recent negative reviews, write a reply in this tone: [tone].
Constraints: do not invent data and warn me if there are too few reviews to see a pattern.

21. Review a reply before posting it

Act as the reply editor for [restaurant name]. I am pasting the review and the draft reply I wrote. Review: [paste the review text]. Draft: [paste your draft].
Task: check the draft against this list: acknowledges the specific fact, is not defensive, contains no personal data and closes with an invitation to come back.
Constraints: do not rewrite everything, mark only the sentences that fail and explain why in one line.
Format: table with sentence, problem and proposed change, and at the end the corrected version.

How to answer negative reviews with AI without sounding robotic

Answering negative reviews with AI works when you follow a process, not when you copy and paste the first output. The goal is twofold: care for the guest’s experience and not lose someone who already took the time to write.

1. Identify the specific thing that bothered the guest: wait, dish, service or price.
2. Decide whether it is a one-off complaint, a venue mistake or a mismatched expectation.
3. Use one of the prompts above with context and tone.
4. Add the name, a real service detail and adjust any sentence that sounds generic.
5. Check that you acknowledge the problem, are not defensive and close with an invitation to book again.

A reply reaches everyone who will read it before deciding whether to book, not just the person who complained. These four mistakes cost credibility:

• that would fit any venue.
• in the review: it shows nobody read it.
• or long justifications, which escalate the conflict.
• and losing the chance to win the guest back.

AI review analysis: what patterns you can spot in your menu

AI review analysis turns hundreds of loose comments into useful information about your menu. Where a manual read only sees isolated complaints, AI groups themes and shows what repeats. It is the basis of good menu engineering: you know which dishes work and which to revisit.

• Signature dishes: the most mentioned in a positive way, with name and frequency.
• Allergens: insufficient information or lack of gluten-free or lactose-free options.
• Waits: whether complaints cluster in specific time slots.
• Guest satisfaction: the month’s tone and how it moved from the previous one.

In the restaurant industry, customer satisfaction shows up in wait times, cold food and slow service comments, and sales data tells you which menu items repeat. A concrete case: a manager exports a month of reviews into a document, passes them to the AI and asks for a summary by theme. In one afternoon they get what would take days by hand. They spot, for example, that wait complaints cluster on Fridays from 9 to 10 pm, a pattern invisible when you read reviews one by one. With that data they adjust shifts and fix the problem at the root.

How I built these prompts and how to write your own

Every prompt above follows the same five-field structure. If you understand what each one does, you can write a prompt for any situation in your restaurant without copying mine.

• Paste the whole review, as the guest wrote it, and the name if it appears.
• you decide what gets published.
• “No excuses” and “no discounts” prevent the most typical AI replies.
• and pass them as a tone example next time.
• If the result does not convince you, say which sentence to change instead of starting over.

Before and after: a weak prompt made strong

Weak prompt

Reply to this review: "They took forever and the rice was cold".

The result is usually a generic text that would fit any venue: no role, no tone, no word limit and no invitation to come back.

Strong prompt

Act as the manager of [Casa Pepe], a rice restaurant in [Valencia].
Context: the guest dined on a Friday at 9:30 pm and says they waited a long time and the rice arrived cold.
Task: write the public reply to this review, acknowledging both problems and inviting them back.
Constraints: warm tone, 80 words maximum, no excuses or discounts.
Format: one paragraph ready to post on Google.
Review: "They took forever and the rice was cold".

With the five fields, the reply cites both facts, sounds like your venue and needs no rewriting.

How to adapt the prompt to each AI assistant

The five-field structure works in any assistant, but each one responds better to a different tweak, and this is what is worth changing.

ChatGPT

It works well with long prompts and files. Attach the exported reviews spreadsheet and ask for the theme table; ask for tone variants in the same answer.

Gemini

Best if your data lives in Google: ask for output as a table for Sheets or as text for Docs, and give one instruction per message.

Copilot

Name the file, sheet or email thread in the prompt. It works with the documents in your work account, so you do not need to paste anything.

Claude

Use its capacity for long texts: paste a quarter of reviews at once and ask for a report with quotes. It is very good at drafting with nuance.

How to build AI prompts into your daily routine and save time

A fixed routine turns the tool into a team habit. Every morning, one person reviews the previous night’s new reviews on Google Business Profile, TheFork and Tripadvisor, generates drafts with the templates, personalizes them and posts them before opening. Ten to fifteen minutes, depending on the venue’s volume.

• Define who reviews the reviews and at what time.
• Save the prompt templates in a shared document.
• Set a daily reminder so no review goes unanswered.
• Review each reply before posting: specific fact, tone, call to book.
• Run a monthly analysis to spot patterns and adjust the menu.

Privacy and data: what to consider when using AI with guest reviews

Working with reviews means handling personal data. Even though Google or Tripadvisor reviews are public, the guest’s name is still personal data, and restaurants in the EU are subject to GDPR.

• Do not paste phone numbers, emails or full names if not needed; use “the guest”.
• and see whether your plan lets you turn off the use of your data for model training.
• or screenshots with contact details or spending history.
• of replies published with AI in case a guest exercises their right to erasure.
• in the prompt unless necessary.

Careful use protects your guests and avoids penalties. You can read the full text of the GDPR on the official European Union site.

Frequently asked questions about AI prompts for restaurants and reviews

These are the questions people search most about AI and restaurant reviews, with the answer first.

How do I answer a negative review with AI without losing the guest?

Give the AI the full review, your restaurant’s tone and the specific thing that upset the guest. Ask for a draft that acknowledges the problem, is not defensive and ends with an invitation to come back, and review it before posting.

What ChatGPT prompt can I use for my restaurant?

Use one with role, context, task, constraints and format, like the ones in the ChatGPT section: for example, “act as the manager of a [cuisine] restaurant and reply to this review in 80 words, no excuses”. Swap the brackets for your details.

Can AI answer Google reviews automatically?

Technically yes, but publishing without review is not advisable. AI does not detect irony or delicate cases such as allergies. Use it to draft and let a person decide what gets published.

Which AI is best for answering reviews: ChatGPT, Gemini, Copilot or Claude?

There is no single best. ChatGPT drafts fast and accepts files, Gemini fits your Google listing and Sheets, Copilot works with your Microsoft 365 files and Claude handles long texts well. Try two with real reviews and keep the one that needs the least editing.

How do I know what ChatGPT says about my restaurant?

Ask it the way a guest would: “where to eat [cuisine] in [city]?” and “what do people say about [your restaurant]?”. Note whether your venue appears and what it says. Recent, well-answered reviews and a complete Google listing are a solid base for your reputation.

How much time can I save with these prompts?

A draft goes from several minutes to seconds, and the monthly review analysis from days of reading to one afternoon. Always count review time: a routine of ten to fifteen minutes a day is usually enough.

Which guest data should I not paste into AI?

Phone numbers, emails, booking data, screenshots with contact details and full names if not needed. Replace the name with “the guest” and check your assistant’s data policy.

Can I use AI to ask my guests for reviews?

Yes: it can write the message for the table, WhatsApp or the thank-you email. Do not offer incentives in exchange for the review or ask for five stars, because platforms penalize it. See prompt number 18.

Your next step with AI prompts for restaurants and reviews

A negative review no longer has to cost you a guest. Start today with one template, save it in a shared document and apply it every morning. If you want your whole team to reply to the same standard, the Founderz AI program offers a practical approach focused on real work. Founderz is an online business school specialized in applied AI and a Microsoft Worldwide Training Partner.