Hotels are starting to worry that personal AI assistants will book a room, keep watching the price, cancel, and rebook somewhere cheaper. Here’s my view: that isn’t an AI problem. It’s a pricing problem that hotels built themselves, and AI is just doing what guests always wanted to do but couldn’t be bothered to.
If your revenue strategy only works because most people are too busy to check the price again, you never had a strategy. You had inertia. Agents are about to take that inertia away.
What’s actually happening
PhocusWire’s Matthew Parsons has a useful piece on the rise of consumer AI assistants such as Booking Holdings’ Vayo, OpenAI’s dots, Meta’s Muse and Instinct, which has just raised $1 billion. Vayo can already pull travel confirmations from a user’s inbox and watch bookings for price drops. Its CEO says hotel price-drop capture and transactional booking are on the way. Instinct is being used to check restaurant sites every five seconds, and more than half of its transactions are travel-related.
Rick Posey from Cendyn calls it a set of “landmines”. His worry is that agents told to keep shopping will push up look-to-book ratios and make pace look healthier than it really is, because bookings sitting on free cancellation can disappear the moment a cheaper rate turns up, even at a rival hotel.
To be fair, the same article admits there’s little evidence of damage yet. AI referrals are still around 1% of bookings for most hotel groups. But the direction is clear, and the people quoted think it’ll all land at once.
Hotels taught guests to re-shop
Re-shopping isn’t new. Corporate travel tools like FairFly (now Oversee) and TripBam built whole businesses on it. Google, Skyscanner and Kayak have offered price tracking for years. Savvy leisure travellers have done it by hand forever: book something flexible, check again next week, swap if it’s cheaper.
Why? Because hotels made it worth doing. Rates move daily. Free cancellation is standard on most rates. The same room on the same night can be cheaper tomorrow than it was today. When the price is that slippery, of course people want a robot to keep an eye on it.
I’ve argued before that hotels don’t have an AI pricing trust problem, they have a pricing problem. Re-shopping agents are the bill for that problem arriving.
The wrong fix: lock everything down
The obvious reaction is to push non-refundable rates harder, tighten cancellation windows and try to spot and block agent traffic. Eviivo’s Gwénaël Merlin suggests just that sort of balancing act.
I think that’s a mistake, and the data says so. Phocuswright research quoted in the piece found 58% of travellers would pay a higher rate for free cancellation. Guests value flexibility. Taking it away to stop software from doing its job is punishing the customer for having a smart assistant. It’s the same instinct that made hotels fight the OTAs, then review sites, then metasearch. It never worked.
And blocking agents is a losing battle anyway. Booking.com is pouring money into Vayo, as I wrote when it was buying the trip layer, and Meta is building travel into Muse, which I covered in Meta wants a cut of your travel bookings. These assistants will be where a lot of travellers live. You don’t get to opt out.
The right fix: give the agent a reason to stop looking
The best line in the article comes from Jason Cincotta at Kismet: give a known guest a logged-in rate through their AI, and “it gives the agent a reason to stop looking”. He goes further: the agent knows the guest wants late check-out, a cot, connecting rooms or dinner on night one. Package that, quote it, and there’s nothing like-for-like to re-shop against.
That’s the real point. Re-shopping only works on commodities. If your offer is the same room at the same rate as on five other sites, an agent will hunt it down to the last pound. If your offer is built around a specific guest, the agent can’t find a cheaper copy, because there isn’t one.
So the work isn’t in revenue management alone. It’s in marketing and distribution:
- Make your direct offer readable by machines. Agents can only pick what they can see. That’s why I keep saying AI isn’t snubbing independent hotels, it just hasn’t heard of them.
- Price the package, not just the room. Bundles tied to a person are much harder to undercut than a bare rate.
- Score agents like you score OTAs. Safa Rahal suggests tracking assistants on cancellations and confirmed bookings. Agents that book cleanly earn better terms. Agents that churn don’t.
- Stop reading pace on faith. If more bookings start coming from assistants, split them out and watch their cancellation curve separately.
The 1% excuse won’t last
The easy response is “it’s only 1%, we’ll deal with it later”. That’s what plenty of hotels said about ChatGPT referrals a year ago. As I wrote then, 1% is the wrong number to watch. What matters is who controls the decision, and personal assistants are being built to control it on the guest’s side.
Hotels that treat agents as an enemy will get flat, commodity bookings that cancel at the first price dip. Hotels that treat them as a guest’s representative, and give them something worth keeping, will get bookings that stick.
So, is AI re-shopping a landmine for hotels, or just a mirror held up to how they’ve priced for the last decade? I know where I stand. Tell me where you do.
Want more takes like this on AI, search and travel marketing? Subscribe to my LinkedIn newsletter.