by Nikki Irving-Miralrio, Morpho Creative Consulting
Issue 1 · October 15, 2026Three stories before winter
The Rate Sheet, the Robot,and the Pumpkin
The slow week is not a rate problem.
There is a pumpkin on my front step going soft on one side.
Which means we have reached that very specific point in the hospitality calendar when owners start staring at November occupancy, AI starts appearing in every industry headline, and somebody opens a blank document called BLACK FRIDAY IDEAS and types “30% off.”
So this week, three stories. One about the slow weeks. One about the machine that may or may not know your property exists. And one about why your Black Friday plan should probably sell January instead of discounting it.
Story one · The stay
The slow weekis not a rate problem
I was sitting on my bed with a laptop on my knees, looking at somebody else’s occupancy report. October was the column that looked like a flat tire.
And I knew the next sentence before it arrived, because it is almost always the same sentence: “So what do we drop the rate to?”
Read those numbers together. The short stay still dominates. The long stay is what is growing. And most small properties are doing very little to catch it.
Try it · The turn math
Fill six soft nights. How many room turns does it cost you?
Each turn is housekeeping, linens, check-in, check-out, messages, questions. A six-night guest gives you six nights of revenue with one room turn.
So does the channel. OTA bookings canceled at 21.8%. Direct bookings canceled at 10.6%. More than double. The average cancellation arrived 39 days before the stay, so some of the discounted OTA inventory that looks full in September is fiction.
Compare · OTA vs direct
Different datasets. Same direction. The guest who books directly is more likely to spend more, stay connected to the property, and come back through a door you own.
I am not telling you to fire the OTAs. You cannot, and you should not. I am telling you to stop using the most expensive, most cancellable, least loyal channel in your business as the automatic answer to your slowest weeks. Then blaming the weather.
What I would do between now and January
0 of 5 done
The discount is the lazy answer. It is also the answer that teaches the market what you are worth.
The agent is not stealing your booking.It never saw you
I keep running the same experiment. A property’s website open in one window, a clean AI session open in the other, and a set of ordinary guest questions. Not trick questions. The kind a person actually types. Somewhere quiet for a long weekend. Somewhere I can bring the dog.
Replay · The dog question
> ▍
What that property’s site was actually telling machines:
A competitor with a clear page about being dog friendly got the mention instead.
A lot of the properties I look at have the same problem. One of them had the structured data. It had the llms.txt file. Somebody had done the technical work and billed for it.
The fields were empty.
Name, address, phone, logo, hours. That was all of it. No amenities. No pet policy. Nothing about the thing that would have won the dog question.
The machine was not biased against that property. It had nothing to read. Having the file is not the same as having the facts in it. The containers were there. The contents were not.
The industry is telling owners a frightening story right now: AI agents are going to take over the booking. That is not where the immediate problem is. Skift reported:
<1%
of Booking Holdings room nights originate from AI channels
0
for Marriott, per a Bernstein estimate
0
AI usage by the booking and checkout stage
And Skift reported on 3 April 2026 that only 2 percent of U.S. consumers say they are willing to let an autonomous agent do the booking, while 80 percent of travel executives plan to deploy agentic AI at scale.
Read that last line twice. They are building a thing that 2 percent of people want.
Almost nobody is completing the transaction inside the machine. But the machine is increasingly involved before the transaction, when a traveler is deciding what belongs on the shortlist. Phocuswright found 56 percent of U.S. travelers had used AI to plan a trip in the twelve months to March 2026. More than half your market now asks a machine for the shortlist before they ever type a hotel name. And that is where independent properties are getting lost.
The Barcelona study
Stiplo analyzed 27,360 AI travel answers across 659 Barcelona properties. Independent hotels were eight times more likely than chains to be unknown to the AI system. Once the system knew the independent property, it recommended it at roughly the same rate as a chain.
That is not a preference problem. It is not a chain conspiracy. It is a records problem.
The same research found Google review volume was the factor most strongly associated with discoverability. Not a magic file. Not a clever paragraph full of robot words. Evidence, consistency, and enough public information to establish that the property is real and known for something.
Now the part that should annoy you.
On 27 August 2026 Google began turning on hotel booking inside AI Mode, where a guest chats, picks a room, reads the cancellation policy, and pays with Google Pay without ever leaving the conversation. Here is the entire launch partner list.
Jochen Koedijk, Expedia Group’s marketing chief, told Skift on 22 May 2026 that “agents are becoming a new audience, in addition to businesses and consumers.” The enterprise crowd heard it first, and they are wiring themselves into the pipe while you are rewriting your homepage headline.
I sell AI work, and I do not love all of this. In the same Skift coverage, Skyscanner’s chief executive cited a global survey in which 51% of travelers said they distrust the reliability and accuracy of AI travel information. They are right to be skeptical. But “I do not trust it” and “it is helping build my shortlist” can both be true. Only one of those is yours to fix.
I have seen focused subject pages associated with stronger discovery across a small number of properties. I treat that as a useful pattern, not a guaranteed ranking factor. Anyone promising you a guaranteed AI ranking is selling you something.
The machine is going to describe your property to a stranger whether you participate or not.
Where the lift came from: only 4.6% more visitors, but 56.6% higher booking-engine conversion. They did not buy a bigger crowd. They did a better job with the people already standing at the door.
Before you pour money into ads, fix the offer page. Test the booking engine on a phone. Make sure the email list is clean. Remove the little bits of friction you have stopped noticing because you already know how your own site works. Your guest does not.
Everybody crowded around the thing growing 17% and left the thing growing 307% sitting on the table.
Think about the person shopping on the last Friday in November. They may not know when they are traveling. They are shopping for a present.
A voucher is a present. A discounted Tuesday in February is a scheduling problem.
And the cash arrives in November, when you need it.
The six-week plan
Tap a step to open it.
Holiday travelers are watching price. Deloitte’s 2025 survey found average travel budgets were down 18%, and 63% of travelers ranked price among their top three lodging criteria. Price-aware does not automatically mean “cut the rate.”
Give value you can take back next year. A third night. Late checkout. Breakfast. Firewood. The dog stays free. A credit at the restaurant down the street that will trade you for it.
Protect the rate. All of it protects your rate, and your rate is the one thing you cannot un-teach your market.
And put a real cancellation term on anything deep. If you sell 40 percent off as fully flexible in November, you did not sell a room. You sold an option, and you find out in January.
Measure what tells you it worked
0 of 4 tracked
Because Black Friday is not really a test of whether people like a deal. We settled that a long time ago. It is a test of whether your own front door works.