Custom GPT: Hotel Revenue Analyst with Embedded Rate Logic

What You'll Build

A Custom GPT configured as your hotel's Revenue Analyst — pre-loaded with your rate logic, comp set positioning, and stakeholder communication templates. Unlike a standard ChatGPT conversation, this GPT opens already knowing your property and immediately produces hotel-specific outputs: comp set narratives, displacement calculations, owner report drafts, and rate justification language — without any context-setting on your part.

Why This Matters for Revenue Managers

Revenue managers who share their property context with every new AI session are wasting the most valuable part of their day. A Custom GPT acts like a junior analyst who has already been briefed — you give the data, they produce the output.

What You'll Need

• ChatGPT Plus subscription ({{tool:ChatGPT.plan}} — {{tool:ChatGPT.price}}/month)
• Your property fact sheet (or notes from your last ownership presentation)
• Sample comp set data from Lighthouse or STR (even a few weeks is enough to calibrate)
• One example of a good owner report narrative you've written — the GPT will learn your voice from it

Architecture Overview

The Custom GPT has three configurable layers: (1) a system prompt with property context and standing instructions, (2) uploaded knowledge files containing your rate rules, comp set data, and template examples, and (3) conversation starters that surface the most common tasks with one click. The result is a tool that behaves less like a generic AI and more like a trained assistant who knows your hotel.

Build Guide

Phase 1: Prepare Your Knowledge Files

Before building the GPT, prepare two documents to upload as knowledge files. These will be stored inside the GPT and referenced on every response.

File 1: Property & Market Context (plain text or Word doc)

Write 400-600 words covering:

• Property name, brand, room count, segment, location
• Primary guest segments and their characteristics
• Comp set (name, brand, room count, relative positioning — higher/lower/same segment)
• Local market demand drivers and seasonality
• Current RevPAR and MPI performance vs. comp set
• Pricing philosophy: rate floors, MPI targets, discount rules, restriction triggers

File 2: Rate Strategy Rules & Templates

Create a document with two sections:

Section A — Rate Decision Rules: Write out your decision logic as explicit IF/THEN statements. Example:

Section B — Communication Templates: Paste in 2-3 examples of your actual stakeholder communications — a GM rate justification email, an owner report paragraph, a sales team comp rate brief. The GPT will calibrate its tone and format to match your existing style.

Phase 2: Build the Custom GPT

1. In ChatGPT, click your profile icon → My GPTs → Create a GPT
2. In the Create tab, describe what you want:

   Prompt

   "You are a hotel revenue management analyst for [Hotel Name]. You know our property, comp set, and pricing philosophy. You help me draft stakeholder reports, analyze competitive data, build rate justifications, and produce displacement calculations. You always give specific recommendations, not generic frameworks. When I share data, lead with the action I should take."

3. Switch to the Configure tab for detailed setup:

   • Name: "[Hotel Name] Revenue Analyst"
   • Description: "Hotel-specific revenue management assistant for [Hotel Name] — rate strategy, reporting, and stakeholder communication"
   • Instructions: Paste your property context here (the same content as File 1, but written as direct instructions: "You know the following about our hotel: ...")
   • Knowledge: Upload both files you prepared in Phase 1
4. Add Conversation Starters (click "+ Add starter"):

   • "Draft a weekly GM performance summary — here's the data:"
   • "Analyze this comp set pricing and give me a rate recommendation:"
   • "Write an owner report narrative for this month's results:"
   • "Run a group displacement check — here are the group details:"
5. Set Capabilities: Enable Web Browsing (for market research requests) and Code Interpreter (for Excel formula generation)
6. Click Save and set visibility to Only me

Phase 3: Test with Real Scenarios

Run these three tests to validate performance:

Test 1 — Context retention: Ask "What's our rate floor on weekends?" without providing any data. It should answer correctly from your uploaded context.

Test 2 — Competitive analysis: Paste 5 lines of comp set rates for a specific date. Ask for a rate recommendation. The output should reference your specific comp set properties by name and tie the recommendation to your MPI target.

Test 3 — Report drafting: Provide a bullet list of last week's metrics (RevPAR, occupancy, ADR, comp set index). Ask for a GM summary. The output should match the tone and format of your uploaded template examples.

If any test produces generic output, strengthen the relevant section in your instructions.

Real-World Walkthrough: Group Displacement in 4 Minutes

A sales manager messages you: "Got a request for 65 rooms, 4 nights, rate $159, dates March 14-17. They want an answer by noon."

You open your Revenue Analyst GPT and type:

The GPT runs the displacement math, formats the result, and outputs:

• Net group revenue vs. displaced transient comparison
• Break-even transient occupancy threshold
• Recommended counter-offer rate if displacement is negative
• A 3-sentence explanation to share with the sales manager

Total time: 4 minutes including typing, versus 25-45 minutes building the model manually in Excel.

Customization Options

• Multiple room type tiers: Add rate rules per room type (standard vs. suite) for more granular displacement analysis
• Event calendar: Upload your annual events calendar as a knowledge file — the GPT can factor known events into any date-specific recommendation
• Seasonal updates: Update the knowledge files quarterly as market conditions shift

Maintenance

• Monthly: Review outputs for accuracy drift — if recommendations stop feeling right, the market context may need updating
• Quarterly: Refresh comp set data and pricing rules in knowledge files
• Annually: Upload new STR annual summary, update budget assumptions

What This Won't Do

• Pull live data from your PMS or rate shopping tools — you paste data in; it analyzes it
• Execute rate changes — it recommends; you act
• Replace a full RMS — it handles the human communication layer, not automated pricing algorithms