OpenAI's ChatGPT Data Shows Who Hotels Could Reach With Ads

Information-seeking now accounts for about 20% of all ChatGPT messages, up from 16.7% in July 2024, according to OpenAI's published Signals data. That single figure is the most concrete thing hoteliers can extract from the dataset, because it contains no travel section at all. No destinations. No hotel queries. No category for trips.

Anyone quoting a percentage of ChatGPT usage that is travel in the coming weeks is guessing. Maybe an educated guess, but a guess.

What does the data actually show?

The dataset reveals intent: what people try to do when they open the app. The relevant group for hotels is information-seekers, the users most likely to be planning something, a trip included. Three quarters of that activity has nothing to do with work.

The analysis asked three questions, the same ones an operator would ask before opening a shop on a new street:

• Who walks by?
• What are they looking for?
• Can a sign be placed in front of them?

The answers vary sharply by geography. In Q3, Spain (26.3%) and France (26.1%) led the large markets in share of information-seeking messages. The UK followed at 23.0%, Germany at 22.9% and the US at 21.0%. India and Brazil sat near 16%.

Age matters more than geography. Information-seeking runs at 18.1% of messages among 18-to-24-year-olds, against 25.2% among 55-to-64-year-olds. The older users ask more.

Which users can even see ads?

Here the economics get murky. Ads appear only on two plans: Free and Go, the $8 tier. Plus and Pro subscribers see none. OpenAI published no breakdown of how many users sit on each plan, by country or otherwise. The only signal is behavioral: Free is the most personal-use plan, about 75% non-work, while Pro skews 68% work.

Consider a concrete case. A Paris hotel whose guests are mostly American faces a favorable age profile: 46% of US ChatGPT messages come from users aged 35 and over, roughly the group that can afford a high-end hotel in a major European city. About 21% of US messages are information-seeking.

Then the data goes quiet. How many of those users are on a free account? Unknown. The open question is whether a traveler with money for a luxury Paris stay uses the free version, or pays $20 a month and never sees an ad at all. Plenty of people do not pay for software they use casually, so the free user with money may be more common than assumed. The dataset cannot settle it.

Should hotels buy ChatGPT ads?

The data will not answer that. Reasoning alone cannot settle it either. The recommendation is a small test: one or two source markets, a leisure segment, measured with the usual caveats. It is early, so operators should expect to learn rather than expect miracles.

Before spending anything, operators can work the public dataset themselves:

• Take the top three or four source markets.
• Check the age mix and the information-seeking share in each.
• Compare that profile against actual guests.

A hotel in Seville drawing German guests in their fifties sees a very different picture from a Lisbon hostel full of 25-year-olds. The data does not give all the answers, but it gives something, and that is more than operators had before.