November 17, 2026
A hotel room is the only product in the shop that vanishes at midnight. So who decides what it costs? I asked hoteliers, expecting to hear about software. Mostly I heard about people.
A hotel room is a strange thing to sell. A baker left with a croissant at four o’clock can sell it at half price, or eat it. A room that stays empty tonight simply disappears at midnight, along with whatever somebody might have paid for it. Hardly any other business has stock that vanishes every day at the same hour, and yet plenty of hotels still price it as though it were tinned soup.
For a few weeks this autumn I asked hoteliers one question: who decides what a night here costs? I expected to hear about software. Mostly I heard about people: owners who wake at six, a husband and wife who settle the year’s rates over a single coffee, managers who check their rivals’ prices before they check their own email. Pricing, I came to realise, is less a strategy than a habit, and habits tend to give people away.
The first owner I visited runs 20 rooms in a converted farmhouse. Her prices live on a sheet of A4 in a plastic sleeve, in the drawer under reception, beneath the spare keys and a bag of dog biscuits. There are three numbers on it, for low, mid and high season. She and her husband agree them every January and then, with visible relief, stop thinking about them.
“We know our guests,” she told me, and she clearly did. She knew who wanted the room over the garden and who would complain about the church bells. What the sheet did not know was the calendar. With three room types on sale up to a year ahead, her hotel had well over a thousand prices to set, one for every room on every night, and the sheet covered three of them. When a wedding in the village filled a dead weekend in March, she sold her last room at the March rate and counted herself lucky. The guest, who would cheerfully have paid double, felt luckier still.
An hour down the coast, the manager of a 30-room hotel had a more modern habit, though it was a habit all the same. Every morning at seven she poured a coffee, opened a spreadsheet of last year’s bookings and then three browser tabs, one for each of the hotels she regarded as her rivals on Booking.com. If they had gone up, she went up. If they had come down, she fretted for a minute and came down too.
“Forty minutes,” she said. “On a good day.” On a bad day the boiler broke or a housekeeper called in sick, and the prices were left to fend for themselves. It did not seem to have occurred to her that her rivals were very probably doing the same thing to her. A small seaside town can hold several managers who copy one another every morning, each quietly assuming that somebody else in the circle knows what they are doing.
In the city, hotels are big enough to employ people whose only job is pricing, and I expected them to have escaped this kind of guesswork. The revenue manager I met there was in his thirties and well armed: a rate shopper, a dashboard with more colours than the lobby carpet and, lately, a new adviser. Before a bank holiday weekend he had exported his bookings and asked ChatGPT what to charge.
“It’s very sure of itself,” he said, in the voice of a man who has sat through a lot of consultants. His problem was not a shortage of information. He had more than he could use. What he lacked was hours. The spreadsheet he had pasted in was already a day old, and the booking that would have changed his mind arrived at eleven that night, when he was at home and nobody, human or otherwise, was watching.
None of this is new, of course. Selling the same seat to different people at different prices was turned into a science by American airlines in the 1980s, once deregulation forced them to compete on fares. Hotels borrowed the idea soon afterwards, with Marriott among the first of the big chains. For decades “revenue management” was something large companies did, in rooms full of analysts, while independent hotels watched from the side with their sheets of A4.
What has changed is that the tools have trickled down, and the hoteliers I met had each grabbed a different one. Several had set up rules in their channel manager: if the hotel is more than 70 per cent full, add 10 per cent. Rules are the civil service of pricing. They are consistent and easy to defend, and they will never notice a festival. The booking sites now offer pricing suggestions of their own, free of charge, which one owner regarded with suspicion. “They’re optimising for somebody,” he said. “I’ve never been convinced it’s me.”
Another manager used the pricing module built into his property management system, and explained the appeal in five words: “I have enough suppliers already.” If the hotel industry ever commissions a coat of arms, that sentence belongs on it. His module worked, roughly. It rarely got any cleverer, because for the company that built it, pricing was one feature in a long queue behind payments and housekeeping.
The most sophisticated set-up I saw belonged to the revenue director of a 200-room hotel, and it had the most human flaw. Every night her system forecast demand and recommended a price for every room for the year ahead. Every morning her team approved the next few weeks and left the rest, because there are only so many hours in a day. The other ten or eleven months were, in effect, priced by a machine that wrote memos nobody had time to read. The system offered an autopilot that would publish its prices without waiting for anyone, and she was unconvinced. “If the forecast is wrong,” she said, “it’s now wrong automatically.” She said it the way people talk about self-service tills.
The owner of a small inn inland had gone further than any of them. He had handed the decision to an AI model that does not stop at recommending a price. It sets one, sends it to his booking systems and then checks whether it was right, adjusting as bookings come in. He had chosen the lowest and highest prices he would ever accept, and the machine worked between them.
He compared the first fortnight to teaching his daughter to drive. “I checked it every hour,” he said. What settled his nerves was being able to see why each price had been set, and knowing he could take the wheel on any night he liked. By the time I visited he looked once a day, after breakfast, and spent the rest of the morning on his guests, which was presumably why he had gone into hotels in the first place. When I described his set-up to the others, the farmhouse owner looked appalled. The manager with the three browser tabs was curious, then asked who would be blamed if it went wrong. It was a fair question, and the honest answer was the same as it had always been: whoever was in charge of the price.
Which brings me to the one hotelier who has stopped being in charge of it altogether, and the only person in this piece I can describe exactly as I found him. He no longer cares about pricing, and he has arranged his business so that he does not have to. The “Book now” button on his hotel’s website does not open a booking engine of his own. It sends visitors straight to his listing on Booking.com, where the prices, the payment and the confirmation email are all handled for him. The front door of his business now opens onto a company in Amsterdam.
I could see the attraction: no software, no spreadsheets, no seven o’clock ritual. But the bill arrives in other ways. Guests who found his hotel by name, typed in his own web address and were ready to book with him now book through a platform that charges him commission on every stay. He does not get their email addresses, so if he wants them back next year he will pay to be introduced to people who have already slept in his beds. And his rooms sit in a shop window built to show those same guests his competitors, one scroll down. Of everyone I met, he seemed the most at peace. Not caring about price is still a way of setting one, though. It simply means somebody else decides, and takes a cut for the trouble.
There is one more character in this story, and it has barely walked in. Hoteliers are starting to connect AI assistants such as Claude and ChatGPT to their own data. Pointed at a spreadsheet, that is just a quicker way of guessing, as the man with the colourful dashboard discovered. Plugged into a real pricing system, it becomes something closer to a chat with a colleague: why is Saturday priced like that, how is November looking, go ahead with that change. Hotels are built on conversation, so it would be fitting if pricing ended up as one too. Others, meanwhile, hand the whole question to a consultant or a brand’s head office, which does not make the question disappear. It just moves it to a desk further away.
By the end I noticed that none of these people really disagreed about prices. They disagreed, without quite saying so, about trust: whether to trust themselves, a rule, a model or a platform with the value of a night under their roof. If you run a hotel, there is a small exercise you can do over breakfast. Multiply your room types by the nights you have on sale, and you have the number of prices your hotel really has. Then ask how many of them anyone looked at this week, and whether the way you set them is any cleverer than it was a year ago. The answers tend to be smaller than the first number.
On my way out of the farmhouse I told the owner her arithmetic: a thousand-odd prices, three of them set this year. She took the plastic sleeve out of the drawer, looked at it for a long moment, and put it back under the dog biscuits.
“January,” she said.
The owner who sends his bookings to Booking.com is a real conversation; he is not named. The other hoteliers in this article are composites, drawn from conversations with property owners and managers. Details have been changed and no individual hotel is identified.