Guides

ChatGPT Prompts for Restaurants: Reports, Rotas, Reviews and More

Copy-paste ChatGPT prompts for restaurant reports, profit margins, staff rotas, menu descriptions, and Google review replies, with connected Super44 workflows for recurring tasks.

Alex Riesenkampff

July 10, 2026 · Updated July 11, 2026 · 10 min read · Markdown

The prompts below cover work an independent restaurant can hand to ChatGPT or Claude today. 86% of US restaurant operators say they are comfortable using AI and 81% plan to use it more, yet only 26% currently use an AI tool in their restaurant (Toast 2025; National Restaurant Association 2026, via Restaurant Dive). You will get copy-ready prompts for menu copy, reviews, reports, rotas, and supplier emails. Each task also shows what changes when Super44 has live venue data and operating context.

Why most "AI for restaurants" content doesn't help

Most search results offer software lists or broad advice instead of prompts an owner can paste into a chat. Menu-generator pages focus on layouts. General restaurant prompt lists rarely cover staff scheduling or supplier communication. This guide includes the prompt, the inputs it needs, and the checks to run before using the result.

What restaurant operators who use AI actually use it for

The usage data points in the same direction. Marketing is the most common reported AI use case among US restaurant operators: 19% of full-service and 15% of limited-service operators use it for marketing, compared with 10% for administrative tasks and 6% for AI-driven ordering (National Restaurant Association, 2026, via Restaurant Dive). Menu copy, social captions, and review replies are practical starting points.

The three-part formula behind every prompt that works

A useful prompt needs three things: role, context, and format. Missing one usually produces generic copy. OpenAI's own prompt-engineering guidance recommends giving the model a clear identity, explicit instructions, and the relevant context up front, rather than asking a vague question and hoping. For a restaurant, that breaks down into:

1. Role: "You are an experienced menu copywriter for independent restaurants" or "You are the owner replying calmly but firmly to a critical review."
2. Context: the actual facts. The dish, the ingredients, the prep method, the audience, the tone. Without this the model fills gaps with plausible-sounding filler; with it, the output sounds like your restaurant.
3. Format: length, number of variants, tone. "Three variants, under 25 words each, warm with a little dry humor" produces something different from "write a description."

The step most people skip is iteration. Treat the first draft as a starting point, not a finished product. A second, sharper instruction ("shorter," "less salesy," "closer to how we actually talk") almost always beats trying to nail it on the first attempt. Every prompt below follows this pattern: copy it, swap the bracketed placeholders for your own facts, then refine.

With Super44: The venue's approved tone and operating context are already available. The owner can ask for the decision or draft directly, without rebuilding the background in every prompt.

Menu descriptions that sell instead of sitting there

The gap between a menu description that sells and one nobody reads is specificity, not creativity. "Grilled chicken sandwich with honey sauce" is a fact; "crispy grilled chicken, spicy hot honey glaze, melted pepper jack, house pickles, toasted brioche" is a sales pitch. The prompt has to force that level of detail, not just accept a dish name.

You are an experienced menu copywriter for independent restaurants.
Write a menu description for: [dish name].
Ingredients and prep: [e.g. "dry-aged beef, 48-hour braise, celeriac purée,
roasted mushrooms, red wine jus"].
Audience/tone: [e.g. "neighborhood bistro, warm, no sales-speak"].
Length: max 25 words.
Give me three variants, each with a different focus (preparation / ingredient
sourcing / the eating experience).

A cafe, a bar, and a bakery should get different results from this prompt because their inputs differ. Check every variant against the dish before publishing it, especially the ingredients and allergens. Models fill gaps with plausible details, and plausible is not necessarily correct.

With Super44: Connected POS sales and review feedback ground menu copy in dishes that sell, language guests use, and the venue's established tone.

Answering Google reviews without sounding like a bot

A peer-reviewed study of 935,386 Google Maps reviews across 5,010 UK restaurants found that food, service, and atmosphere all affect the odds of receiving five stars. A separate Harvard study measured Yelp rather than Google. It found a 5–9% revenue effect per additional Yelp star for independent restaurants, while chains were largely unaffected. Google and Yelp are different datasets, but both show why owners should read the detail behind the score.

Guests are already asking AI before they walk in

26% of UK consumers now use AI tools like ChatGPT to research a venue before visiting, which is on par with Google Maps at 27%, and 60% say they trust AI-generated summaries of reviews. That's from a CGA by NIQ / Reputation study of 755 GB consumers, fielded September 2025. As Anthony Gaskell, Managing Director EMEA at Reputation, put it: "AI isn't a future trend in hospitality: it's already here... Operators need to adapt quickly to this new age of personalisation."

The problem with generic AI replies is that guests spot them instantly: "we're sorry to hear that, your feedback matters to us" reads as a template because it is one. The difference is whether the reply names the actual problem:

You are the owner of a [cafe/restaurant/bar] replying to a Google review , 
polite but personal, not corporate.
Review: [paste the review text].
Star rating: [number].
Our tone: [e.g. "casual and honest" or "formal and understated"].
Name the specific issue from the review, apologize only for what actually went
wrong, and, if relevant: offer one concrete next step (an invitation to talk,
not just "we'll follow up").
Max 60 words, no stock phrases like "we appreciate your feedback."

A Super44 analysis for a Cologne cafe read its full Google review history and quantified an estimated €4,000 monthly opportunity around 1- and 2-star experiences. The resulting plan identified roughly €3,600 across weekend staffing, service training, and reservation confirmations, with reply drafts for open reviews. The figures describe an opportunity found, not revenue already recovered.

With Super44: New Google reviews sync automatically and can trigger a notification. The response draft uses the venue's preferred tone, past review themes, approved replies, and hospitality research. The owner reviews the copy before posting.

How to upload restaurant reports to ChatGPT or Claude

For a one-off margin analysis, export the POS or accounting report as CSV or XLSX, remove personal data, upload it to ChatGPT or Claude, and ask for every formula. Define the columns before asking for conclusions. Gross margin and operating margin are different measures, and net sales cannot be compared cleanly with costs that include VAT.

Before uploading, remove guest and employee names, email addresses, phone numbers, and free-text notes. Then use a prompt like this:

I run a [restaurant/cafe/bar]. I have attached a [daily/weekly/monthly] report.
The currency is [EUR/GBP/USD], sales are [including/excluding] VAT, and each row
represents [explain the rows]. Relevant columns:
- Net sales: [column]
- Opening inventory, purchases, and closing inventory: [columns]
- Wages plus employer taxes, pension, and other on-costs: [columns]
- Other operating costs: [columns]

First check whether the data is complete enough. Calculate cost of goods sold
as opening inventory + purchases - closing inventory; do not treat purchases
alone as food cost. Then calculate gross-margin, food-cost, fully loaded
labour-cost, and operating-margin percentages. If an input is missing, say so
instead of estimating it. Then:
1. show every formula;
2. calculate the metrics by week and for the full period;
3. identify the three largest changes;
4. cite the exact rows behind each conclusion;
5. suggest three questions I should investigate next.

Use the same structure to analyse a Z-report, calculate labour cost, or compare food cost month to month. For the next report, the export, cleanup, explanation, and upload all have to be repeated. The answer is only as current as the file.

When repeated uploads stop making sense

Super44 keeps connected POS and operating data current, so it can analyse live numbers and surface a briefing without waiting for another upload. The prompt remains useful because it exposes the formulas and assumptions. Super44 removes the repeated export and preparation work.

Building a rota an AI can actually help with

A rota prompt must include the legal limits; otherwise the draft can be unusable. In the UK, workers are entitled to 11 hours of rest between shifts, a 20-minute break after more than 6 hours, and 24 uninterrupted hours off each week (or 48 hours per fortnight). Average weekly hours are capped at 48 over a 17-week reference period unless the worker has opted out (GOV.UK). The model also needs the team's actual availability and local rules.

Help me draft a staff rota for [period, e.g. "next week"].
Team: [names/roles, e.g. "Anna (front of house, full-time), Ben (kitchen,
part-time 20h), Clara (front of house, weekend casual)"].
Availability: [who can't work when].
Expected footfall by day/daypart: [rough estimate].
Hard limits to respect: no more than an average 48-hour week, at least 11
hours' rest between shifts, a 20-minute break for anyone working more than 6
hours.
Give me the result as a table with start/end times per person and day, and flag
any shift that sits close to one of these limits.

If your rota already lives in a spreadsheet, attach the XLSX or CSV instead of copying every shift into the prompt. Explain the date, employee, role, start, end, break, availability, and location columns, then ask for a proposed change list before asking for a rewritten file. That gives you an audit trail: "move Tuesday's late shift from A to B" is much easier to verify than a silently regenerated rota.

The draft still needs an owner review. A rota contains staff names, so check the tool's data handling before uploading it. The financial effect is material: at €25,000 monthly revenue, one percentage point of labour cost share equals roughly €3,000 a year.

From spreadsheet draft to connected staffing

Super44 keeps availability, time off, roles, published shifts, and POS-informed demand together when preparing the weekly rota. The agent explains assignments and compliance warnings, and the owner approves the plan before staff are notified. Changes, sick calls, and uncovered shifts can trigger help with cover. If the rota still lives in a spreadsheet, the prompt above provides a transparent first draft.

Social posts, supplier emails, and translations in minutes

Marketing is already the leading reported AI use case for full-service restaurant operators at 19%, and 81% of surveyed operators plan to increase their AI use (Toast; National Restaurant Association). Social captions, supplier emails, and menu translations all benefit from precise facts and a defined tone.

Social post for a daily special:

Write an Instagram caption for [today's special/promotion] at our [cafe/bar/
restaurant]. Tone: [e.g. "casual, minimal exclamation points, no emoji
overload"].
Give one concrete reason it's worth trying (ingredient, seasonality, limited
run), not just "come try it!" with no reason attached. Max 3 sentences plus 3
relevant hashtags.

Supplier email over a price increase:

Write a polite but firm email to our supplier [name/product]. Situation: [e.g.
"price for X rose 12% with no advance notice"].
Goal: [e.g. "ask for justification, propose a volume discount or price hold"].
Tone: businesslike, not confrontational, but clear about what we expect. Max
120 words.

Menu translation for tourist guests:

Translate this menu item into [language]: [paste the original].
Keep any dish names that are internationally recognized in their original form
and briefly explain them in parentheses. Tone: appetizing, not a literal,
technical translation.

Specific inputs reduce editing. "Price rose 12% without notice after three years as a customer" gives the model something useful; "price went up" does not.

With Super44: Connected email, accounting, POS, and review data give supplier emails real price history and social drafts real products and guest language. The owner decides what to send or publish.

What belongs in a prompt, and what doesn't

Public review text is generally low-risk to paste; names, contact details, and staff information need more care. The UK's Information Commissioner's Office expects organizations to assess data protection before putting personal data into a generative AI tool. Its 2023 review of Snap's "My AI" chatbot prompted Stephen Almond, the ICO's Executive Director for Regulatory Risk, to warn the industry. For restaurant work, replace staff names with placeholders before uploading a rota and avoid adding private complaint details that are not already in the public review.

Use these prompts for occasional tasks and to inspect the assumptions behind an answer. If every Monday starts with the same export, connect the source instead. Super44 keeps the restaurant context current and can surface the next issue without waiting for another upload.