The best AI prompts for hotels, reviews and upselling, 20 to copy (2026)

A good prompt for replying to hotel reviews gives the AI the role, the guest details, your brand tone and the limits, not just the review. With that, ChatGPT, Gemini, Claude or Copilot draft a personalized reply in minutes. Below, 20 prompts to copy; to bring it to your whole team, see the Founderz AI program.

Add Founderz as a preferred source

What you’ll take from this

This library gathers the prompts a hotel needs for reputation and sales, with real examples to copy and adapt:

• Prompts to reply to Booking, Google and TripAdvisor reviews: positive, negative, mixed and unfair, keeping your hotel tone.
• Sentiment analysis prompts to spot the complaints that repeat across hundreds of reviews.
• Upselling emails that use arrival date, room type and reason for the stay.
• Versions for ChatGPT, Claude, Copilot and Gemini, and the method to write your own.

Managing a hotel’s online reputation eats hours every week. A 50-room property can get dozens of reviews a month across Booking, Google and internal surveys, and replying one by one, with judgment and without sounding like a template, is slow. This practical guide to artificial intelligence for the hotel and tourism sector shows you, step by step, how to use AI to analyze reviews and personalize each online reply. Well-designed AI prompts keep the quality of each reply and give time back to what needs a human, such as welcoming guests at the front desk or handling check-in.

In the hotel sector, AI review analysis lets you process thousands of reviews, spot the strengths guests praise most in comments on Booking, Google and TripAdvisor and improve guest satisfaction. Because these are repetitive tasks, you need no technical team or technical knowledge: ChatGPT, from OpenAI, has a free version and a specialized AI review platform connects to the PMS if you need it later.

What AI prompts for hotels are and who they are for

A prompt is the instruction you give an AI platform such as ChatGPT to produce a specific text. In a hotel it is the order you use to ask for a review reply, an upselling email or a summary of opinions. The more detailed it is, the more useful the draft. No technical knowledge is needed: you write in plain language, the language models interpret the instruction and return the text.

• Front desk and customer service staff, who answer reviews and inquiries daily.
• Revenue managers, who look for upselling opportunities and want to personalize offers.
• Owners of tourist accommodation, apartments or rural houses without a big team behind them.
• Hotel marketing managers, who handle reputation across several platforms at once.

AI speeds up the writing; the judgment about what to publish stays with the team. Before you start, a rule: remove names, booking numbers and any data that identifies a guest.

ChatGPT prompts for hotel reviews and upselling

These are the prompts I would use in ChatGPT for the daily reputation and sales work of a hotel, sorted by task. They also work in other tools with small changes. Text in brackets is an example: replace it with your case and keep the rest. Library reviewed in November 2026.

Prompts to reply to hotel reviews on Booking, Google and TripAdvisor

A review-reply prompt has four parts: role, context, tone and limits. With all four, the reply comes out personalized; with only the review, it comes out generic.

Positive review that mentions specific details

Act as the guest relations manager of a [3-star family] hotel in [Santander].
Context: a guest left this [5]-star review on [Booking]: "[paste the review without the name]". They praise [the cleanliness and the breakfast].
Task: write a reply that names those two details, thanks them personally and invites them back.
Constraints: maximum [80] words, [warm and professional] tone, no stock phrases like "thank you for your feedback", no promised discounts.
Format: one text ready to publish, signed by [name and role].

Negative review about noise, cleanliness or service

Act as the general manager of a [40-room boutique] hotel in [Valencia].
Context: [2]-star review on [Google]: "[paste the review without the name]". The problem is [noise in the room in the early morning]. We have already spoken with [the night team] and we will [revisit construction hours and offer interior rooms].
Task: reply with real empathy: acknowledge the problem, apologize without excuses, explain the specific action and offer a direct channel [reservations@hotel.com].
Constraints: do not promise financial compensation, do not blame the guest, no defensive tone, maximum [90] words.
Format: one text ready to review and publish, plus one line with what the team must confirm before publishing.

Mixed 3-star review

Act as the online reputation manager of a [4-star city] hotel.
Context: [3]-star review on [TripAdvisor]: "[paste the review without the name]". It praises [the location] and criticizes [the slow check-in].
Task: thank them for the positive with a specific detail, acknowledge the criticism without justifying it and say what we are doing to improve it: [a second desk at peak hours].
Constraints: [100] words maximum, [professional and warm] tone, do not mention other guests or competitors.
Format: one text ready to publish.

Unfair review or one with false claims

Act as the general manager of a [120-room beach] hotel.
Context: this [1]-star review arrives: "[paste the review without the name]". It claims [there was no air conditioning], but according to our records [every room has it and there were no breakdown reports].
Task: write a calm, firm reply that regrets the bad experience, states the facts without accusing anyone of lying and invites them to continue the conversation in private.
Constraints: no irony, no personal data about the guest, no dates or room number, maximum [100] words.
Format: one text to publish and a shorter alternative of [50] words.

Review in another language

Act as the guest relations manager of a [city] hotel with international guests.
Context: this review is written in [Spanish]: "[paste the review without the name]".
Task: reply in the language of the review with the same tone we would use in English: [warm and professional]. Then translate your reply into English so I can review it.
Constraints: simple sentences, no idioms, maximum [80] words per language.
Format: two blocks, "Reply in [Spanish]" and "English translation".

Prompts for upselling emails before arrival

To really personalize, the prompt needs the arrival and departure dates, the room type booked, the reason for the stay and the extra services you want to promote, from F&B to local experiences.

Room upgrade for a getaway

Act as the revenue manager of a [4-star boutique] hotel in [San Sebastian].
Context: [Laura] arrives on [Friday the 14th] and leaves on [Sunday the 16th] with a [standard] room booked for a [couples getaway]. We have [a sea-view room] available for [+$40 per night] and a [romantic dinner at the restaurant for $75 for two].
Task: write an email that offers the upgrade and the dinner without sounding like an ad, with a clear call to action.
Constraints: maximum [120] words, [elegant and brief] tone, a single reply address [reservations@hotel.com], no pressure or fake scarcity.
Format: subject line under 50 characters and the email body.

Late check-out, spa or local experience based on the reason for the stay

Act as the sales manager of a [4-star city] hotel in [Chicago].
Context: I am giving you three arrival profiles for next week, without personal data: [A: business trip, 1 night, single room; B: family with two kids, 3 nights, double room; C: couple, anniversary, 2 nights, superior room]. Services we can offer: [late check-out at 3 pm for $25, spa entry for $30, free kids' breakfast, anniversary cake for $18].
Task: for each profile, pick the offer that makes the most sense and write a short email.
Constraints: one offer per email, maximum [90] words, [friendly] tone, no pressure.
Format: three blocks with subject line and body.

Review sentiment analysis with ChatGPT

Sentiment analysis classifies opinions as positive, negative or neutral and groups them by theme. If 40 guests mention slow wi-fi, the AI spots it and groups it; reading one by one, that pattern would slip past you. With about 50 reviews, patterns start to show.

Recurring complaints and an improvement roadmap

Act as the online reputation analyst of an [80-room] hotel.
Context: I am pasting [60] reviews from the last [3] months on [Booking and Google], without personal data: [paste the reviews].
Task: 1) classify each review as positive, negative or neutral; 2) group them by theme (cleanliness, wi-fi, breakfast, staff, parking, value for money); 3) list the three most repeated complaints and the three most praised strengths; 4) propose one action for each complaint.
Constraints: do not invent data that is not in the reviews and flag any theme with fewer than 3 mentions.
Format: table with theme, number of mentions, sentiment, short example quote and proposed action; then a roadmap of 5 improvements ordered by impact and effort.

Claude prompts for hotel reviews and upselling

Claude stands out with long documents and careful drafting. Use it when you have a large export of reviews, a tone manual or a complaints protocol you want to put in writing.

Long review export into a report

Act as the reputation analyst of a [150-room resort] hotel in [Mallorca].
Context: I am attaching a CSV file with [500] reviews from the last year from [Booking, Google and TripAdvisor], with columns [platform, date, rating, text]. Personal data has already been removed.
Task: read the whole file and write a report with: trends by quarter, the five most repeated negative themes, the five most praised, differences between platforms and seasonality.
Constraints: quote at least one review per theme, do not invent figures and mark with [doubtful data] anything you cannot verify in the file.
Format: [two-page] report with a five-line executive summary at the top and a final table with priorities.

Tone guide from your best replies

Act as the brand editor of a [boutique] hotel in [Seville].
Context: I am pasting [10] review replies the team considers exemplary and [3] that do not represent us: [paste the texts].
Task: extract our brand tone and write a style guide for replying to reviews: words and phrases we use, the ones we avoid, how to open and close, ideal length and how to handle complaints, praise and unfair reviews.
Constraints: base each rule on an example from the texts, no generic rules.
Format: one-page guide with numbered sections, ready to paste as the first instruction in any AI tool.

Serious complaint handling protocol

Act as the operations director of a [60-room] hotel.
Context: our team replies to complaints without a common standard. Frequent complaint types: [noise, cleanliness, charges, staff attitude, safety]. Team: [front desk, housekeeping, management].
Task: write a protocol for serious negative reviews: who sees them, the time limit to reply, what is confirmed with the affected department, when to escalate to management and when to move to a private channel.
Constraints: clear language for staff without technical training, no promises of compensation without authorization, say when to consult legal counsel.
Format: numbered steps with owner and deadline, plus a review checklist before publishing.

Copilot prompts for hotel reviews and upselling

Copilot works inside Excel, Outlook, Word and Teams with the files in your work account, so it fits if your hotel already lives in Microsoft 365. Some features depend on your license.

Copilot in Excel: reviews by theme and rating

Analyze the table in this sheet [Hotel_reviews_2026], with the columns [Date, Platform, Rating, Comment].
Task: add a [Theme] column with one of these values [cleanliness, wi-fi, breakfast, staff, parking, noise, other] and a [Sentiment] column with [positive, negative or neutral]. Then create a pivot table with the number of reviews by theme and sentiment, and the average rating by platform.
Constraints: do not change the original columns and, if a comment fits two themes, choose the main one.
Format: the pivot table on a new sheet and a bar chart with the five most negative themes.

Copilot in Outlook: upselling email and guest reply

Write an email replying to this guest message [paste the latest email]. Use the booking in the thread: [standard double room, arrival October 12, 2 nights, couples trip].
Task: answer their question directly and, at the end, offer [an upgrade to a room with a terrace for +$35 per night] in one sentence, not a sales block.
Constraints: [cordial and brief] tone, maximum [110] words, do not invent hotel policies, mark in brackets anything the front desk must confirm.
Format: suggested subject line and email body, in the guest's language.

Copilot in Word and Teams: monthly reputation report

Use as sources the document [Reputation_report_September.docx] and the Teams meeting transcript [Weekly reputation review].
Task: write a monthly online reputation report for hotel management with: average rating by platform, three recurring complaints, three compliments, actions agreed in the meeting, owner and date for each.
Constraints: use only data from those two files, write [no data] if something is missing and do not include guest names.
Format: one-page document with a three-line summary, an actions table and a final paragraph with the month's priority.

Gemini prompts for hotel reviews and upselling

Gemini fits if your hotel works with Google Business, Gmail, Docs and Sheets. Google reviews are the ones travelers see most when they search for accommodation, so they are a good starting point.

Replies to Google Business reviews

Act as the guest relations manager of [Hotel Mirador], a [20-room rural] hotel in [Cantabria].
Context: these are the [5] new reviews on our Google Business profile: [paste the reviews without names, with their rating].
Task: write a different reply for each one, adapted to its rating: thank them with a specific detail on 4 and 5 stars and acknowledge the problem with an action on 3 or below.
Constraints: maximum [70] words per reply, [warm and direct] tone, do not repeat the same opening twice, do not promise compensation.
Format: numbered list, one reply per review, with the rating at the start.

Gemini in Sheets: rating and themes by month

Analyze the data in the sheet [Guest_reviews], with the columns [Month, Platform, Rating, Comment].
Task: calculate the average rating by month and by platform, detect the months where it drops and, for those months, summarize in one sentence the most repeated negative themes from the comments.
Constraints: do not invent data, state clearly if a month has fewer than [10] reviews and do not assume causes that are not in the comments.
Format: a new table with month, average rating, change from the previous month and main theme, and two concluding sentences.

Gemini in Gmail and Docs: upselling email with the seasonal offer

Use the document [Fall offers 2026] in my Drive as the only source of offers.
Context: [30-room mountain] hotel in [the Colorado Rockies]. Guest arriving on [Friday, November 7] with a [double room] for [a weekend as a couple].
Task: write an email that presents only the offer in the document that best fits their trip and explains how to book it.
Constraints: maximum [110] words, [friendly] tone, do not mention prices that are not in the document, do not invent services.
Format: subject line under 50 characters, email body and a final line with the offer deadline.

More hotel prompts: before and after the stay

The same structure works for other messages in the guest cycle, from pre-arrival communication to asking for the review.

Pre-arrival email that cuts front desk questions

Act as the guest communication manager of a [3-star city] hotel in [Boston].
Context: the guest arrives in [3 days]. Useful details: [arrival from 3 pm, parking by prior reservation at $20 a day, breakfast from 7:00 to 10:30, free wi-fi].
Task: write a welcome email that answers the five most frequent questions and offers [discounted breakfast] as the only offer.
Constraints: maximum [150] words, short sentences, no jargon, do not invent services, include a contact phone [+1 000 000 0000].
Format: subject line, body with a four-point list and a warm sign-off.

Post-stay message to ask for a review

Act as the guest relations manager of a [family] hotel in [Charleston].
Context: the guest left yesterday after [3 nights] and has not left any feedback. Review link: [Google link].
Task: write a thank-you email that invites them to leave a review without steering the opinion: no asking for five stars and no rewards.
Constraints: maximum [80] words, [grateful and simple] tone, a single call to action.
Format: subject line and body.

Analysis of your direct competitors’ reviews

Act as a hotel strategy consultant.
Context: I am pasting public reviews of [3] competitor hotels in [Miami], without guest names: [paste the reviews by hotel] and the [40] reviews of our own: [paste the reviews].
Task: compare what guests of each hotel praise and criticize on the same themes: location, cleanliness, breakfast, staff, value for money. Point out two opportunities where we could stand out and two risks.
Constraints: use only the texts I paste and do not attribute to one hotel what another one says.
Format: comparison table of themes by hotel and a final paragraph with the main recommendation.

Which AI tool to pick for your hotel

You do not need an expensive suite to get started. A general-purpose AI platform covers most use cases of a small hotel, a vacation apartment or a rural house, and all of these have a free version to try. They are a good AI for analyzing reviews and improving the guest experience without paying upfront. Check the current terms of each plan and each API before deciding, because they change often.

If you are looking for hotel-specific software, see the comparison of AI tools for hotels.

How to automate AI prompts in the hotel workflow

Automating does not mean letting the AI publish on its own: it means it drafts and a person reviews. A sensible workflow has three phases:

1. Capture: export reviews and booking data regularly or connect the AI to your PMS.
2. Drafting: save your prompts as templates and fill the brackets with the stay details.
3. Review: a team member approves every text before it is sent or published.

Assign who approves what: in a small hotel, the owner; in a larger one, front desk or marketing. No reply should go out without human review, especially complaints.

To measure whether it works, track a few KPIs for a month: average time to reply to a review, share of reviews answered, average rating on Booking and Google, and upselling revenue per booking. With that data you decide whether to extend AI to more revenue management or hospitality tasks, or to adjust the prompts. And know the common mistakes before they cost you a review: publishing without review, pasting the same reply on several reviews and pasting personal data.

How I built these prompts for reviews and upselling

I have been writing prompts since before ChatGPT existed, for ads, web content and the processes of entire teams. What I learned is simple: AI does not guess what you do not give it. Every prompt on this page follows the same six fields, in this order:

Five practical tips to write your own:

1. Start with the fact only you know: what the guest said, their room type, what you will do to fix it.
2. Ask for one change at a time in the same conversation instead of rewriting everything.
3. Save the reply you like most and paste it as a tone example in the next prompt.
4. Always tell it what not to do: promise compensation, invent policies, blame the guest.
5. Review each draft as if you were signing it, because you will be.

An example of a weak prompt made strong:

Before

Reply to this negative review: "[paste the review]"

After

Act as the general manager of a [40-room boutique] hotel. Reply to this [2]-star review about [night noise]: "[paste the review without the name]". Acknowledge the problem, apologize without excuses, explain that [we will review construction hours] and offer [reservations@hotel.com]. Do not promise compensation. [Professional and human] tone, maximum [90] words, text ready to publish.

The first returns a generic text any hotel could publish. The second returns a reply only your hotel could have written.

How the prompt changes in ChatGPT, Gemini, Copilot and Claude

• ChatGPT: accepts pasted text, CSV or Excel files and understands full sentences. This library works here as is.
• Gemini: works well with your Drive files. Name the document or sheet in the prompt and ask for one result per message.
• Copilot: say which app you are in (Excel, Outlook, Word, Teams) and which file it uses as the source; it does not invent data outside your files.
• Claude: handles long documents. Attach the whole file and ask for a report with direct quotes.

Limits of artificial intelligence and guest data protection

AI speeds up writing, but some decisions remain yours. For sensitive reviews, complaints with legal implications or delicate situations, reviewing the text and adjusting the tone is mandatory. Brand tone also needs supervision: AI imitates the style you ask for, but does not always get the nuances of your hotel right.

On privacy there is one basic rule: do not paste guest personal data into public AI tools. Also avoid scraping platforms without permission, because it can breach their terms of use. Common mistakes to avoid:

• Pasting reviews with the guest’s full name visible.
• Uploading spreadsheets with contact data that is not anonymized.
• Sharing booking information in free accounts without checking their terms of use.

Frequently asked questions about AI prompts for hotels and reviews

These are the questions hotel teams ask most, with the answer first.

What is the best AI tool for hotels?

It depends on your task. ChatGPT is the most versatile for replies, emails and sentiment analysis; Claude for long documents; Copilot if you work in Excel and Outlook; Gemini if you use Google Business and Sheets. For a small hotel, any of them on a free plan is enough to start.

Does using AI to answer your hotel reviews work?

Yes, when the AI drafts and a person reviews before publishing. You answer more reviews in less time and keep your brand tone. The mistake is publishing without review: replies come out generic and the guest notices.

What prompt do I use to reply to a negative Booking review?

One that includes the role, the review without personal data, the specific problem, the action you will take, the tone and the limits, such as “do not promise compensation”. The full version is in the ChatGPT section, ready to copy.

Can I use Gemini to reply to Google Business reviews?

Yes. Paste the reviews without names along with their rating and ask for a different reply for each one, adapted to the stars. Review each text before publishing it on your profile.

Can I use Claude or Copilot to analyze hotel reviews?

Yes. Claude is useful when you have hundreds of reviews in a long file and want a report with quotes. Copilot classifies by theme and sentiment inside Excel if you have the license. In both, remove personal data first.

Does AI replace hotel staff?

No. It automates repetitive tasks such as reply drafts or review summaries, but front desk service, delicate complaints and brand tone still need people.

How many reviews do I need for a useful sentiment analysis?

There is no magic number, but with about 50 reviews patterns start to show. With hundreds, the analysis reveals trends impossible to see reading them one by one. Start with what you have and expand.

What file format do I need to analyze reviews with AI?

Plain text pasted into the chat, a CSV file or an Excel sheet with the exported reviews. If you connect the AI to another system via API, the response usually arrives in JSON. For a manual analysis, CSV is the easiest: export, anonymize and upload.

What does ChatGPT say about my hotel when someone asks where to stay?

Check it yourself: open a new conversation in ChatGPT and in Gemini and ask about lodging in your city the way a traveler would. More and more travelers use these tools to find the best hotels, and the quality of your reviews and the up-to-date information in your profiles influence how they present you.

Your next step with AI prompts for hotels

Pick one of the prompts in the library, change what is in brackets and run your first test with a real review. If you want your team to master these workflows with method, the Founderz AI program brings these techniques to real work, with a practical, applied approach.