AI Is Rewriting How the World Finds Greece — and the Industry Conversation Hasn't Caught Up

Search for "AI and Greek tourism" and the results are dominated by one story: the Greek National Tourism Organisation's July 2026 relaunch of VisitGreece.gr, complete with an AI trip planner, a voice assistant built with ElevenLabs, and a €7.22 million digital transformation budget. It is a genuinely significant investment, and this site has covered it in detail elsewhere. But it is also, in the context of what is actually happening to travel discovery globally in 2026, a fairly narrow story — a single national tourism board building its own AI-powered front door, at the same moment that the front door itself is being rebuilt by forces entirely outside any single country's or company's control.

That larger shift is what this report covers: the accelerating move of travel discovery away from search-and-click and into AI-mediated conversation and recommendation, what the data says about who is winning and losing as that happens, and — most usefully for the tour operators, hoteliers, and destination marketers reading this — what a business actually needs to do to stay visible as the mechanics of discovery change under it.

The scale of the shift: this is not a niche trend anymore

The headline numbers, drawn from several independent 2026 datasets, describe the same underlying movement from different angles.

On the search side: Google's AI Overviews now appear on an estimated 25.11% of all searches, per a 21.9-million-search study published in Q1 2026. Pew Research separately found that when an AI summary appears above the results, users click through to a traditional result in just 8% of visits, versus 15% when no summary appears — and broader analysis suggests roughly 83% of searches that trigger a full AI Overview end without any click at all. The zero-click rate specifically has climbed from 22.8% in mid-2024 to 26.7% by late 2025.

On the generative-AI-platform side: Similarweb's 2026 Generative AI Landscape report found that AI website visits worldwide reached 9.5 billion a month between June 2025 and May 2026, up 70% year-on-year — and that the market has fragmented rapidly in the process. A year earlier, ChatGPT held roughly 76% share of generative AI web traffic; by mid-2026 that had slid to around 53%, with Gemini climbing past a quarter of all traffic and Claude becoming the fastest-growing platform in the category, now approaching a tenth of the market. For any business thinking about AI visibility, the practical implication is immediate: optimizing for one AI platform no longer covers most of the addressable audience.

On the travel-specific side, the two most directly relevant figures: Phocuswright's 2026 research found that one in ten US internet users now starts their travel discovery journey inside a generative AI tool rather than a search engine, and Deloitte's 2026 travel outlook found that generative AI use specifically for trip planning roughly tripled between 2023 and 2025.

None of these figures describes some future possibility. They describe the current state of the market, as of mid-2026.

Why travel is ground zero — and why citation isn't the same as traffic

Here is the detail that should get more attention in travel-industry planning than it currently does: Travel & Hospitality has the highest AI citation rate of any industry Similarweb tracks. By May 2026, roughly 23% of relevant ChatGPT prompts included a citation to a travel source — well ahead of the next-highest category, automotive, at around 20%, and dramatically ahead of professional services, which sits under 4%. In plain terms, when someone asks an AI assistant a travel question, it is unusually likely — relative to almost any other kind of query — to name and link a specific source.

That sounds like good news for travel content publishers, and in one sense it is: it means AI systems consider travel questions to be the kind of thing worth sourcing rather than answering purely from internal knowledge. But Similarweb's own analysis flags the catch directly: there is "a clear disconnect between the pages ChatGPT cites and those receiving referral traffic." Being named inside an AI-generated answer does not reliably send the visit that a search-result listing used to guarantee. A traveler can read a full, satisfying answer inside the chat window — sourced, in part, from your content — and never click through to see it for themselves. This is the zero-click dynamic again, but sharpened: even winning the citation battle doesn't automatically win the traffic.

There is a genuine silver lining inside this same dynamic, though. AI referral traffic — the visits that do arrive from an AI platform — is growing fast and, by some measures, converts unusually well. Similarweb documented AI referral traffic up 237% year-on-year specifically for the travel category (behind only marketplaces at 312% and news at 278%), and separate industry analysis has put AI-referred conversion rates at up to several times the rate of traditional organic search traffic, on the logic that a traveler who has already had a detailed AI conversation about their trip and then clicks through anyway is unusually motivated — closer to bottom-of-funnel intent than a cold search click. The net effect for travel businesses is a smaller but higher-intent traffic stream from AI sources, layered on top of a shrinking traditional organic stream — a genuine reallocation, not simply a loss.

Who's actually winning the AI recommendation battle right now

The structural question underneath all of this is: when an AI system recommends a hotel, a tour, or a destination, where is it actually pulling that recommendation from?

Increasingly, the answer favors platforms with large, standardized, machine-readable inventories over individual business websites. Booking.com and Expedia already expose structured, consistent data across millions of properties worldwide — meaning, as one 2026 industry analysis put it, "every property on those platforms inherits a machine-readable data layer whether it invests in one or not." Independent hotels, smaller tour operators, and destination-specific sites that rely on their own websites have to build that structured layer themselves, or accept that AI systems will describe them using whatever secondhand information — OTA listings, review aggregators, directory entries — happens to already be indexed about them elsewhere.

This dynamic is accelerating as the major platforms race to close the loop between AI recommendation and actual booking:

- Google demonstrated an agentic "weekend planner" agent at Google I/O on May 19, 2026 — a tool that pulls from review sites, activity platforms, hotel availability, and real-time pricing to build a complete personalized itinerary with direct booking links, entirely inside Google Search. Google's head of Search called the underlying interface redesign "the biggest upgrade to the search box in 25 years," and AI Mode itself crossed one billion monthly users within its first year. Separately, as of June 2026, Google has been spotted testing direct hotel-booking links inside AI Mode responses that route travelers to official hotel websites rather than through Booking.com or Expedia — a meaningful signal for an OTA industry that has spent two decades collecting 15–25% commissions as the default intermediary between hotels and travelers.
- OpenAI launched ChatGPT apps with Expedia and Booking.com as initial partners, and has leaned further into routing actual purchases to those retailers rather than completing bookings inside ChatGPT itself — Expedia's own AI assistant, "Romie," and its tools letting partner AI agents query its inventory directly, are a clear signal that major OTAs intend to remain the booking infrastructure underneath these AI conversations rather than being disintermediated by them.
- Perplexity partnered with SelfBook in spring 2026 to add native hotel booking directly inside its own interface.
- Sabre, PayPal, and MindTrip announced a partnership in February 2026 to build what they describe as travel's first end-to-end agentic booking pipeline — a traveler describes a trip in natural language, MindTrip queries Sabre's Mosaic APIs (covering more than 420 airlines and 2 million hotel properties), and PayPal's infrastructure handles payment, all inside a single conversational interface, targeting a Q2 2026 launch.

The common thread across every one of these initiatives is the same: AI agents evaluate businesses based on structured, machine-readable data, not on how a website looks to a human visitor. A hotel or tour operator whose only online presence is a conventionally designed website — built for a human to browse — is, from an AI agent's perspective, largely unreadable, regardless of how good that website actually is.

The Greek-specific gap: real momentum, narrow public conversation

Set against this global picture, the current state of the Greek tourism industry's public conversation about AI is worth naming directly, because the gap is instructive.

Greece's most visible 2026 AI-in-tourism story is the VisitGreece.gr relaunch: an AI Trip Planner, an AI chatbot spanning both conversational "Chat Mode" and a certified-information "InfoDesk Mode," the redesigned mAI Greece assistant with ElevenLabs voice interaction, and a broader digital ecosystem including a digital repository and interactive tourism museum. Earlier in 2026, Prime Minister Kyriakos Mitsotakis and Tourism Minister Olga Kefalogianni met directly with ElevenLabs' CEO to discuss expanding AI voice technology across Greece's tourism platforms, and a separate international forum in London — "Smart and Regenerative Tourism & Hospitality Competitiveness in the AI Era" — positioned Greece as pursuing AI-driven destination management, including AI-powered heat maps and visitor-flow systems at archaeological sites.

This is real, well-funded, high-level activity — but it is fundamentally a story about one national platform improving how it serves travelers directly. It says comparatively little about the question actually occupying global travel trade press throughout 2026: whether the thousands of individual Greek hotels, tour operators, transfer companies, and destination-specific websites that make up the actual fabric of Greek inbound tourism are structured in a way that lets ChatGPT, Gemini, or Google's AI Mode find, trust, and recommend them at all.

The underlying demand-side shift, meanwhile, is demonstrably happening in the European market Greece depends on. Data presented at a SETE-hosted conference in December 2025 — drawing on the European Travel Commission's Tourism Economics/Vitreous World Travel Trends Survey — showed that use of AI chatbots or virtual assistants in planning or researching a trip rose from 10% in Q3 2024 to 18% in Q3 2025, an eight-percentage-point jump in a single year, among the tools travelers reported using. That single, industry-sourced data point confirms the global disruption story is not an abstraction for Greece's own source markets — it is measurably underway among the same European travelers currently booking Greek holidays.

Risk and opportunity for independent Greek operators

The structural risk for smaller, independent Greek tourism businesses is real and worth stating plainly: AI agents currently favor platforms with large, pre-existing, standardized data — meaning Booking.com, Expedia, and TripAdvisor listings may increasingly mediate how a Greek boutique hotel or a family-run tour operator gets described to an AI-using traveler, even when that business has invested heavily in its own website and brand. A business with no structured data of its own risks being described by whatever secondhand information already exists about it elsewhere, rather than by its own words.

But the same body of 2026 research points to a genuine, specific opportunity running the other direction. Industry analysis of GEO consistently finds that AI systems reward specific, well-documented, authoritative content over generic, high-budget SEO content — citation quality over citation quantity, in the words of one 2026 hospitality-marketing analysis: "five well-placed mentions in named publications will outperform fifty mentions in low-authority outlets." AI models appear to read reviews for qualitative signal — a property consistently described as "perfect for families" or a guide consistently praised for authentic, first-hand destination knowledge gets associated with those traits and surfaced accordingly, independent of overall search-marketing budget.

And there is a genuinely encouraging demand-side signal specific to this dynamic: the same SETE-hosted European conference data found that growing AI tool use in trip planning is already correlating with more off-the-beaten-track and off-season destination discovery — with roughly half of European travelers surveyed now actively seeking less popular or "off the beaten track" destinations, a share notably higher among Austrian, Belgian, and Dutch travelers specifically. For a Greek tourism ecosystem heavily weighted toward a handful of famous islands, and increasingly concerned with the overtourism and seasonal-labour pressures that concentration creates, an AI-mediated discovery layer that is structurally better at surfacing niche, well-documented, less obvious options than a conventional SEO-dominated search results page is a genuine strategic opening — provided the content describing those niche options is actually structured to be found.

What Greek tourism businesses should actually do

The practical guidance emerging from 2026's GEO literature is more concrete than "invest in AI" — it is a specific, executable set of content and data practices:

Implement relevant schema markup. Hotel, TravelAgency, and TouristDestination schema types for core business information; Offer schema for specific packages, prices, and availability windows; FAQPage schema for genuine question-and-answer content; Review schema to surface ratings in a machine-readable form; and, for any business publishing original data or research — arrivals figures, seasonal pricing trends, visitor surveys — Dataset schema specifically, since AI systems are shown to pull statistics directly from clearly marked-up datasets when answering research-style queries.

Structure content to answer questions directly, not just to rank. The GEO literature consistently distinguishes optimizing for the click (traditional SEO) from optimizing for the citation (GEO): a page can rank highly in conventional search and never appear in an AI-generated summary, while a lower-ranking page with clearly extractable, directly-answering content can be cited repeatedly. FAQ-format content blocks — question in, answer out — map directly onto the format AI systems are already producing.

Prioritize citation quality over citation volume. A handful of mentions in named, authoritative regional and national travel publications outweighs a much larger number of mentions in low-authority directories or link-farms, because AI systems weight source authority explicitly rather than simply counting mentions.

Maintain review velocity, not just review volume. Fresh reviews — ideally several per month rather than a large but static historical total — appear to carry more weight with AI systems assessing a property's current character and reputation, and responding publicly to reviews signals active, credible management.

Publish specific, original, well-sourced content. Generic destination marketing copy is exactly the kind of content AI systems can already synthesize without needing to cite a source. Specific, first-hand, data-backed content — the kind this site's own insights section is built around — is comparatively difficult for a generic AI system to replicate from scratch, which is precisely what makes it citable rather than redundant.

What to watch next

Several open threads from 2026 will determine how quickly this shift compounds through the rest of the year and into 2027. Google has confirmed agentic flight and hotel booking is coming to AI Mode but, as of mid-2026, had not set a public launch date; watch whether that capability ships broadly before the 2027 summer booking season, since a live agentic-booking feature inside the world's dominant search engine would represent a much sharper inflection point than anything documented so far. The Sabre/PayPal/MindTrip consortium's targeted Q2 2026 launch is worth checking against actual delivery. And Similarweb's tracked shift away from ChatGPT's near-total AI-search dominance toward a genuinely multi-platform landscape (Gemini, Claude, Perplexity, Meta AI) means any Greek tourism business's GEO strategy now needs to account for several distinct AI platforms rather than one — a materially harder task than it would have been reading the same data twelve months earlier.

For travelers actually planning a Greek trip using AI tools today, our companion guide to using AI travel planning for a Greece holiday covers the practical, traveler-facing side of this same shift.