A young couple in London is planning a long weekend in Ibiza. They do not open Google. They open Claude, or Perplexity, or ChatGPT, and they ask a single question: "Where should we stay for a boutique experience near a good beach club, with direct booking?"
The answer comes back as three or four names. A short paragraph on each. A recommendation. Often the same names will be repeated if they ask the same question on a different AI tool ten minutes later. They click through. They book. The whole process takes under fifteen minutes, end to end.
This is now happening at scale. And most operators on this island are not in the conversation.
The shift, in numbers.
Phocuswright's March 2026 research, which the firm called "the fastest behavioural shift in travel in a decade", found that 56 percent of US leisure travellers used AI to plan at least one trip in the past 12 months, up from 24 percent in 2024 and 43 percent in late 2025. The growth rate is the headline. Behaviour that did not exist three years ago is now mainstream.
The age skew is sharper still. Among travellers aged 18 to 35, the figure is 71 percent. eDreams Odigeo's 9,000-person global study, also from 2025, recorded a 183 percent year-on-year increase in AI travel platform use among the same age group. These are the guests who, in the next five to ten years, will represent the dominant booking demographic for boutique hospitality.
And it is not only the young. The same eDreams study found that 48 percent of consumers aged 55 and over used AI for travel for the first time in the past 12 months. The behaviour is broadening, not narrowing.
Cloudbeds' June 2025 study, the first of its kind on AI hotel recommendations, looked at 810 prompts across Claude, Perplexity, and ChatGPT in six destinations. OTAs accounted for 55.3 percent of all sources AI engines cited. But the more revealing finding sat one layer down: 72.4 percent of AI-recommended hotels were branded properties or affiliated with large groups. Independents had a structural visibility disadvantage in AI search results from day one.
A separate 2026 index, HotelWorld AI, found that approximately 84 percent of global hotels do not appear in AI search recommendations at all. The category of guest using AI to discover where to stay is growing every quarter. The category of operator visible in that space is small, and largely accidental.
Why this is different from SEO.
Operators have heard the phrase "new search optimisation thing" too many times. SEO went through its evolutions. Google's algorithm shifted year after year. The fatigue is real, and reasonable.
But AI search is structurally different from search-engine search, and the difference matters operationally.
A search engine ranks pages. The page that ranks highest gets the click. AI engines do not rank. They generate an answer, citing sources. The reader does not see ten options to choose from. They see a recommendation. The job is no longer to rank above competitors. The job is to be one of the three names in the recommendation.
This changes what the work actually looks like. SEO rewards keyword optimisation, link-building, technical performance. AI search rewards authority signals, structured data, consistent cross-platform citations, and the quality of content available about your business. AI engines build their answers from sources they trust. Operators with strong, citable digital presence get recommended. Operators without it do not.
The new infrastructure for AI visibility (technical foundations like LLMs.txt and schema markup, authority-building content, citation across platforms AI engines already trust) compounds. Once you are established as a trusted reference in your category, the cost of staying there is low. The cost of breaking in for the first time, after the chains have spent five years building authority, is much higher.
SEO rewards ranking. AI search rewards being the answer. The work to be the answer is different work.
The independent's disadvantage. And the reversal.
The Cloudbeds AI study found that 72.4 percent of AI-recommended hotels were branded or chain properties. That number is the structural disadvantage written out. AI engines are surfacing what their training data over-represents, and chains over-represent themselves because they publish more, get covered more, and produce more of the third-party content AI engines learn from.
For an independent boutique hotel on this island, that is the bad news. The default state is invisible.
The good news, and the reason this matters now rather than later, is that the disadvantage is reversible. AI engines weight authority and citability over budget. The chains had a head start, but they did not buy that head start. They earned it through volume of content, depth of citation, and digital infrastructure. An independent operator who builds the same kind of authority, deliberately, can rank alongside or above chains in AI search in a way that has never been possible on traditional booking aggregators.
This is why the chains are moving so visibly. Intercontinental Hotels Group announced in February 2026 a new AI-ready content platform specifically designed to restructure its hotel data into machine-readable formats for AI search visibility. Marriott's agentic mesh. Hilton's 41 use cases. They are not building this infrastructure because AI search is small. They are building it because they understand the window is open, and they want to be on the right side of it before it closes.
The same window is open for independents. It does not stay open forever.
What it actually takes.
The work to become visible in AI search is structured and learnable. Five stages. Understanding where you stand today across AI engines. Building the technical foundations (schema, structured data, LLMs.txt). Establishing authority through expert content and credibility signals. Becoming citable through strategic content placement on the platforms AI engines already trust. Holding the position through ongoing monitoring as AI engines and search behaviour evolve.
None of it is overnight. Meaningful citation growth typically takes four to six months, and the compounding advantage builds for years after that. The operators who start in 2026 will be visible in 2030. The operators who wait will be competing against four years of compounded authority that was built quietly while they were not paying attention.
The methodology is laid out in detail on the AI Search Optimization service page. The point of this essay is not the methodology. The point is the timing.
A young couple in London is planning their long weekend in Ibiza. They will not see most of the operators on this island. They will see the three or four that AI tools recommend.
The guests who book with you next year are already learning where you stand.
