Hello,
The value of data is going up, especially that it is much easier to visualize it and use it with AI. Just drop it in the right place and get an interface built for the user. I tried it a few times for random personal project, pretty neat.
PS: If you haven’t filled out my survey, I’d love to have your reader feedback.
Best, Martin
Generate 50-200 enriched MQLs
every month.
Just find out more ➛
Opinion
The UI is not the product anymore
For the last 30 years, value in technology was created by building better interfaces. The solutions existed, but better interfaces made it easier to use and thus more users, more money.
GDSs already had search, inventory, rates, bookings and payment. But you needed to be a trained travel agent who understood cryptic commands. OTAs came along and put a consumer interface on top, and stole the show. Arguably also grew the pie.
Dashboards are another example. Tableau made it dramatically easier for ordinary business users to turn databases into charts and dashboards without too much coding knowledge.
I think AI is reversing where the value sits. And we are already seeing the beginning of it.
Last month Salesforce and Anthropic announced Claudeforce. Allowing users to access Salesforce data and workflows directly through Claude. Salesforce has been building itself as a platform for years, so this isn’t completely new. But Claude is now increasingly becoming the interface while Salesforce provides the data, permissions and functionality underneath. And that’s a great thing for the users because interfaces are becoming perishable commodities.
Imagine planning a family holiday for ten people. Instead of everyone passing around links and spreadsheets, you ask your AI to build an interface for the trip. It could contain the itinerary, hotel options, flights, maps, prices and activities. Everyone gets access and can vote for hotels, select preferred activities, enter arrival details or indicate which flight they want. The organizer can see the results and eventually complete the bookings.
Once the holiday is booked, throw it away. Next year the trip is completely different with different people, so you give the AI the research and ask it to build another interface around what matters this time. The cost of designing the interface approaches zero. So the user interface becomes whats needed for the moment.
The same applies inside hotels. Imagine a 300-person wedding at a ranch in Wyoming. The organizers could generate a temporary dashboard containing guest arrivals, airport transfers, horse riding schedules, dietary requirements, room allocations, staff tasks and everything else needed for that particular event. Staff could filter it by department or task and guests could have their own relevant view.
A 300-person wedding at a luxury hotel in Paris would require something completely different. Same PMS perhaps. Same event-management system. Maybe even the same underlying data fields. But no off-the-shelf system can cater for all the particularities of every new hotel concept.
Until now, building two completely different interfaces would have been expensive and largely pointless. So software vendors had to create generic dashboards that worked approximately well for both properties. AI changes those economics. If a useful interface can be generated from a voice prompt, used for four days and discarded, there is no particular reason for those two hotels to share one.
This isn’t a SaaSpocalypse (unless your only added value was building a UI). The software underneath still needs to exist. In fact, it arguably becomes more important.
AI can generate the wedding dashboard. What it needs is solid and clean data that is reliable, guest profiles, payment status or dietary requirements. The PMS, CRM, POS, event system and everything else. That data needs to be structured and secure, and the connections need to work reliably.
For years, the added benefit of tech was being able to make it easier to use through interfaces and design. But a lot of that is going to change with natural language inputs.
If interfaces can be made on-demand by the users, much of that value moves upstream. The scarce asset becomes the clean data and reliable pipes needed to make those disposable interfaces useful.
The next generation of valuable tech companies might be the ones we see less, not more.
About me: I'm a fractional CMO for large travel technology companies, and the co-founder of Pragmatik, an agency that drives MQLs through highly relevant content and distribution. You can find out more at wearepragmatik.com. Agentic Breaks the Booking Funnel
AI agents could fundamentally change travel booking because they remove many of the emotional nudges marketers have spent years optimizing. If an agent is comparing price, policies and convenience on a traveler’s behalf, much of the traditional buy-now signals matter less. Also agents will totally skip sponsored listings. OTAs may still have an advantage because someone needs to handle fulfillment and problems, while metasearch looks much more exposed.
AGENTIC TRAVEL + SKIPPING ADS
No, AI isn’t conscious
Mustafa Suleyman pushes back against the idea that AI models should be treated as conscious entities with feelings or rights. AI isn’t a person. It isn’t conscious. It is a machine of 0s and 1s. I remember working in a print shop when I was younger. And a printer there used to consider his press alive and gave it a name. It was counter productive, because when things went wrong it was that “entity” that had a bad vibe. When it always was a mechanical issue. Giving machines an identity makes them harder to control. So don’t refer to your AI as a person. It is an thing and it is OK to control things.
AI CONSCIOUSNESS
The Great Disintermediation
As I mention in my column, AI could become another layer between media companies and their audiences, summarizing information without necessarily sending the reader back to its source. That is a much bigger problem than declining search traffic because the interface itself could become the destination. The media industry has suffered from this. Hotels a little less (a bed is a lot harder to enjoy on a screen) but this is going to change relationships again.
AI DISINTERMEDIATION
AI Hotel Bookings in China
Fliggy’s flyai integration with major Chinese smartphone brands reportedly drove an 800% increase in AI-generated orders during Spring Festival (numbers matter here as well). Fliggy wants to be available across models, operating systems and service platforms. Is there an OTA doing that here?
AGENTIC DISTRIBUTION
Guests Worth 18 Times More
Bookboost’s analysis found that guests with ten or more stays generated around 18 times the revenue of first-time guests. I had done an analysis a while ago and the percentage of return guests in a city like Paris was less than 9% in independent hotels, a group has more, a great group got up to 12%. Chains have more. A friend told me loyalty programs start making sense above 44,000 rooms. All interesting data, if you can do something for more.
GUEST LIFETIME VALUE

Wellness Becomes the Stay
Packaging wellness into three-, five- and seven-night programs rather than treating the spa as another hotel amenity is a great way to turn a hotel/resort into a destination. I wonder if some could convert to that model in the off-season.
WELLNESS TRAVEL
Airbnb Wants Hotels
Brian Chesky wants the hotel category to become a multi-billion-dollar business. They were always an OTA, but they cleverly marketed themselves against hotel chains originally. The thing with rentals is bookings are hard to scale, once you’ve sold the “room” that location is no longer available. Hotels have multiple rooms, so the same product can be sold more than once. Airbnb had to market destinations or emotions, they can’t always deliver on that. But hotels are a more scalable option for growth. And hotels could do with
AIRBNB HOTELS + AIRBNBS OTA RANK⁺
Better Designed Sustainability
Pollution and sustainability is a problem we need to engineer and design our way out of, this project in Bangkok attracted more than 951,000 visitors in its first year while also reducing heat, and pollution. It is a great example of sustainability creating something people actually want rather than simply ticking an ESG box. I don’t know a lot of roofs in cities that are big enough for this. But if it can inspire some it is worth it. As I frequently comment we can’t regulate our way to sustainability, we need to actively find smarter solutions that simultaneously increase quality of life if we want it to work.
URBAN SUSTAINABILITY + ROOF PARK
Hotel Labor Headcount vs Hours
HotelData’s numbers show hours per occupied room (HPOR) falling by more than 3% even while occupancy increased. Comparing that with guest satisfaction would be an interesting exercise (if rooms are turned over faster, did quality suffer). It’s the kind of stuff that should exist in an hotel’s data set so it can be queried and verified. A bit like I mentioned last week. PS: On the record, nobody wants to officially talk about reducing headcount in our industry, but off the record…
HOTEL LABOR + HOTEL ONTOLOGY
Retailer becoming Ad Networks
Retail media has become so attractive that almost every retailer seems tempted to create its own network. The hotel chains are doing it too, and OTAs are growing their media business as well. How is targeting, do we want ads for new fridges in our hotel rooms and booking confirmation emails? When does this become too much for the consumer?
RETAIL ADVERTISING + EVERYONE WANTS ONE
Who should buy Tripadvisor
Tripadvisor has accumulated an enormous amount of review and traveler-intent data over the years. With AI this becomes interesting. I argued that if anyone should buy Tripadvisor, Airbnb would be a good option. Accelerating their footprint into the hotel space and other travel verticals.
TRAVELER REVIEWS
Can AI Finally Connect the Trip?
Travel has talked about the connected trip for decades, but in reality tech has made it less connected. Booking is apparently working on this. I personally think this is where OTAs can still provide value. Because other than an inventory feed, the OTA interface will likely move to one of the big 5 AI chatbots.
CONNECTED TRIP + OTA FUTURE
• The History of Search Engines - Link
• Where to get inspiration for hospitality - Link⁺
• Measuring Political Bias in AI - Link
• The Hotel Yearbook Goes All-In on AI - Link
• The Gulf Game Theory - Link
Did you know: The word "interface" comes from the Latin "inter-" meaning "between" and the English word "face." It was first used in science in the 19th century to describe the surface between two substances. Later, it was adopted in technology and computing to describe points of interaction between systems or users and machines. Defined using Lomar Dictionary⁺