Artificial intelligence is showing up everywhere in hospitality. From marketing copy to guest communication, tools like ChatGPT are helping hotel teams move faster and reduce busywork. It is natural for revenue leaders to wonder whether these tools can also support pricing decisions.
NB: This is an article from TakeUp
Subscribe to our weekly newsletter and stay up to date
After all, ChatGPT can explain revenue management concepts, summarize reports, and even suggest pricing strategies when given historical data. On the surface, that can feel powerful.
The problem is not that ChatGPT is “bad at revenue management.” The problem is that it does not actually do revenue management at all.
Revenue management is a closed loop decision system built on data, feedback, and learning. ChatGPT is an open loop language model designed to generate text that sounds reasonable. Those are fundamentally different things, and confusing them creates real risk when pricing decisions are on the line.
Revenue Management Is a Closed Loop. ChatGPT Is Not.
At its core, revenue management follows a simple but demanding cycle:
- Observe demand signals
- Set a price
- Observe how guests respond
- Update beliefs about demand and price sensitivity
- Repeat continuously
This loop is what allows revenue managers and purpose built revenue systems to learn what actually works. Every price change produces an outcome. That outcome refines future decisions.
ChatGPT does not operate in this loop.
It never observes the results of the prices it suggests. It never sees pickup accelerate or stall. It never learns whether a rate increase caused bookings to shift dates, channels, or competitors. Its recommendations are not informed by outcomes, only by patterns in language and common best practices.
As a result, ChatGPT can talk about revenue management, but it cannot measure demand response to price. And that measurement is the heart of the discipline.
1. Data Without Context Still Produces Confident but Wrong Answers
Uploading spreadsheets into ChatGPT can feel like giving it everything it needs. In reality, it is still missing most of what matters.
The model has no built in understanding of what your data represents. It does not know whether dates reflect booking dates or stay dates. It does not know which columns reflect net room revenue versus total revenue. It does not know whether an outlier week was caused by a citywide event, a snowstorm, or ten rooms being taken out of service.
