Oh, the places they will go… Revolutionizing hotel search with collaborative AI

‍

In just the past few years, advances in AI and increases in computing power have created incredible opportunities for the hospitality industry. What could only previously be imagined can now be made operational. The question is will the industry embrace the new technologies to boost loyalty and increase bookings or continue to cede control to OTAs and the large tech companies?

Over the past 25 years working in travel technology, I’ve done business with almost every major hospitality company in the world, and in that time one conversation has consistently dominated – the difficulty in managing the legacy tech stack. Modernization of the core systems was a necessary evil as these systems were brittle, didn’t integrate well with other systems, created data silos, and most importantly stifled innovation.

As the focus about modernizing the internal technical plumbing raged on, the world outside the property walls changed dramatically. Companies like Amazon, Netflix, and many others, embraced the newest technologies to create seamless, delightful experiences for their customers. The key was using predictive analytics about preferences based on past browsing behavior and purchase history to present personalized recommendations, packaged with extreme digital convenience. Though we’ve tried to keep pace, by and large our industry hasn’t kept up.

Why? Because most people travel much less frequently than they watch movies or shop for everyday items. And their travel needs and wants change when they are traveling for business or pleasure or going to Aspen in the summer versus the winter. Therefore, building up a quality guest profile with the necessary data to predict their preferences is very difficult. Fortunately, recent changes to search driven by the introduction of Generative AI, Large Language Models (LLMs) and the natural language interface, as well as the rise of the experiential traveler are offering us another chance to move the needle.

Google, Apple and major OTAs, like Kayak and Expedia, have already started experimenting with natural language search. This means tens of thousands of users are being trained every day to ask for exactly what they want, in any way they want with the expectation of getting back exactly the result they’re looking for. At the same time, research shows that 60% of travelers don’t have a destination in mind when they start planning their next trip, while 98% of them are booking trips based on the promise of new experiences.

Not meeting these new user expectations could put the direct booking value proposition in jeopardy. Delivering, on the other hand, is likely to increase margins while also opening up new revenue streams. Unfortunately, traditional room search, the “Four Boxes,” of destination, start and end date and number of guests can’t get the job done.

Instead, imagine a potential guest coming to your site and typing into your new search box, “Where can I go to see Taylor Swift for under $500 per ticket, stay at a hotel with an on-site spa close to a Michelin starred restaurant for less than 150K loyalty points?” Or, “where can I stay that is warm in January, on the beach, a direct flight from New York and has a great kid’s club?”

What does it take to deliver a fast, accurate, highly personalized recommendation for the perfect property to meet their very specific desires and requirements? Are LLMs the solution? Well, yes and no. LLMs are simply advanced AI models using the past behavior of text based documents to predict what word should come next in a sentence or paragraph when generating an answer. While seemingly authoritative the answers given by LLMs have many well documented issues.

However, what they are extremely good at is understanding the intent, the needs and desires inherent in the query. Once you understand what potential guests are looking for, you then need live data about the world around each property, what’s currently available, what’s best, ratings, reviews, images, flight inventory, geographical information, climate information, information about animals and plants that certain people and communities are looking for. In other words, a multi-modal AI approach including advanced planning and optimization tools.

A new search box like this would also allow you to develop detailed guest profiles, contextual to their very specific requests. No longer would you rely on prediction to improve your personalized communications. You would simply use the information gathered while answering personal questions. Unlocking experiential search based on user intent also unlocks the kind of engagement customers are looking for, which in turn, boosts customer loyalty, and creates more revenue through direct bookings and opens new revenue opportunities. Bonus: it also sets up a level of hospitality that will differentiate your brand. If you embrace it.