ChatGPT ads cost: what a hotel pays by currency, by click and by booking

A campaign on a daily budget costs a hotel at least EUR 15, USD 25, GBP 15 or CHF 20 a day, about €450 or $750 for 30 days, and the price of each click depends on the market. You pay per click or per 1,000 impressions, never for a booking, even on a campaign that optimises for bookings.

3 prices are easy to mix up. The minimum daily budget is what a campaign must be allowed to spend. The maximum bid is the most you will pay for one click, set at ad-group level. The price you actually pay comes out of a relevance-weighted, second-price auction, so it is often below your bid and depends on who else wants that conversation.

Early prices no longer apply. In February, trade reports put the entry at about $200,000 and $60 per 1,000 views. Those figures still rank on Google, and a hotel reading them will think the channel is closed. Self-service buying with click pricing opened in the US on 5 May and reached Europe on 31 August.

The minimum daily budget depends on the billing currency of your account, and OpenAI’s table has 37 rows. These are the ones most hotels will meet, with what the minimum comes to over 30 days.

A daily budget is an average over a 7-day budget week, Sunday to Saturday in your account’s time zone. A single day can spend up to twice the amount, and the week never more than 7 times it. A campaign that uses its week early can stop while still showing as active. A campaign-total budget is the other option, with a total you set yourself; OpenAI’s table lists minimums for daily budgets only. The Slovak hotel below paid for its 14 days with one, roughly €100 at its published click price. If you use a fixed bid, the daily budget must also be at least your highest bid, which matters when a bid is high.

The currency is fixed when the account is created, along with the country and the time zone. A Swiss hotel that opens in CHF stays in CHF.

A click has cost hotels anywhere from about €0.11 to $3.50, and the honest answer is that nobody knows your price until you run it. OpenAI suggests starting with a maximum bid of $3 to $5; that is a starting bid, and European campaigns have cleared well below it. Only the first 3 rows below are hotels.

Why the spread? Each market fills at its own pace, each test had a different bid and a different goal, and the Slovak hotel lowered its cap step by step. A hotel agency working on US campaigns says a bid under $2.99 was unlikely to serve in its campaigns; the Slovak cap of €0.10 did serve. OpenAI’s low-bid warning is guidance and does not stop an ad running.

A month at the minimum buys roughly 175 to 460 clicks in euros and about 230 in dollars, at the prices published so far. That is enough to learn whether the channel behaves for your hotel, and too little to prove anything about bookings.

Remember that the budget is per campaign. Hotel Klimczok ran 2 campaigns at the Polish minimum of 65 zł, one for clicks and one for conversions, and each got about 10,200 impressions. If you want to compare 2 ideas fairly, you pay the minimum twice.

The cost of a booking is the price of a click divided by the share of clicks that book. A table is more use than any benchmark here, because no public figure says, at a scale anyone can plan on, how often a ChatGPT click books a hotel.

The last column is there for a reason. The Slovak hotel’s Google Ads turned 1.3 visits in 100 into a purchase, which is 1 in 77, while Meta managed 0.2 and the 780 recorded ChatGPT ad visits 0. One hotel in 2 weeks proves nothing about yours, and it warns you not to assume Google’s conversion.

Then compare the result with what the booking is worth to you. Say a €600 stay would otherwise come through a booking site at an assumed 18% commission. That commission is €108, so a direct booking is worth €108 more to you. At 1 booking in 100 clicks, you break even at a click of €1.08. A booking that would have come direct anyway is worth nothing extra, and platform reports cannot tell you which is which.

Most hotels should start with Clicks, or with Conversions if the booking event already works, because Views buys exposure you cannot read. The objective cannot be changed once the campaign exists, and neither can the billing option on a conversion campaign.

Evidence on which wins is thin. OpenAI’s own pack reports a median 44% lower cost per action than click bidding, on 27 advertisers in July, and labels it observational. Hotel Klimczok’s 2 campaigns went the other way in their first week: conversions were about 10% cheaper per click, yet the click campaign started 29 checkouts against 12. Run both only if you can afford both minimums.

The other costs are time and pages, and in a first test they can easily exceed the media. The media can be small; the work around it cannot.

OpenAI is currently offering new advertisers $500 of credit after $500 of spend, until 31 December 2026. Read the terms for your country before counting on it.

The cost of a test is small next to its lessons, provided you stop on time. Set the end date and the highest cost per booking you accept before you start, as the guide to ChatGPT ads for hotels explains. The first step, what to do when the money starts moving, is in how to advertise on ChatGPT. If your page is the weak link, start from how to get your hotel on ChatGPT and the study of 3,288 answers behind it.