Summary
- In May 2026 alone, Google, OpenAI, Anthropic, and OTAs all pointed in the same direction: hotel bookings are moving inside AI.
- When a guest asks AI about a hotel by name, the AI answers almost every time. When they ask "recommend me something," your property shows up in the answer 10–15% of the time.
- 36% of Korean travelers already use AI to plan trips — more than the US (25%). We need AI to include our properties in its answers. Here are five ways to make that happen.
May 19, 2026: Google I/O
At Google I/O in Mountain View on May 19, 2026, Google announced it's expanding the ability to complete purchases inside Search — across more categories. Vidhya Srinivasan, VP of Ads and Commerce, spelled out what's next:
"Soon we'll start with hotel bookings and local food delivery."[1]
The company that owns Search wants to own the last mile: the booking. That same day, Google unveiled an AI that works 24 hours a day in the cloud — even while your laptop sleeps. Guests can now have AI find, compare, and book a hotel while they're asleep. Google's AI Mode crossed 1 billion monthly users in its first year[2].
This wasn't just Google. That same month, three big tech companies and several OTAs made similar announcements. Hard to write this off as one company's bet.
I started as a developer. I've worked in IT my whole career. Since 2008 — when I built an overseas Korean guesthouse booking platform — I've been building hospitality tech. One thing I've learned: the future always arrives later than you think. The internet, smartphones — I felt the shift early, but it took a year (fast) to three years (slow) to reach everyone else. This time it took two months. When my 70-year-old mother said "I asked AI," I couldn't believe my ears.
Can a guest who doesn't know our name discover our property?
Hotel marketing analyst Gil Chan's May 2026 analysis captures the shift. He queried four AI search engines about five vacation rental operators. When guests ask by name — "How's Shilla Hotel Jeju?" — AI cites that property 97% of the time. When they ask a planning question — "Recommend a family stay in Jeju" — any given property appears in the answer 10–15% of the time. The rest of the slots go to large OTAs[3].
Guests who already know us will still find us through AI. New guests who don't know us yet can only choose from the OTA list AI shows them. The funnel for new customer discovery is narrowing.
Customers are already moving fast. Cloudbeds analyzed 90 million bookings and found that US travelers who plan trips via traditional search dropped from 51% to 36% in a single year. In the same period, generative AI usage more than doubled[4]. Early movers are already seeing results. Shiji published a case study: a hotel that structured its data for AI saw direct booking revenue grow 2.1x in two months. 17% of bookings came through AI channels[5].
Korea is moving faster. A survey published in November 2025 found 36% of Korean travelers already use AI to plan trips[6]. In a Deloitte survey, only 25% of US travelers did[7]. An Open Survey poll found that the share of Koreans using ChatGPT for information went from around 40% to 55% in 2025 alone[8]. When ChatGPT's answer didn't satisfy them, 77% said they'd ask ChatGPT again — only 32% said they'd go back to traditional search[9][8].
Our potential guests are already waiting for recommendations from ChatGPT and Gemini.
Five ways to get AI to recommend your property
Gil Chan distilled the conditions for appearing in AI answers: "To gain visibility, you need the same fundamentals that built the search era: authoritative content, structured data, external citations."[3] For hotels, that's five things.
1. Define your property's identity in one sentence
AI doesn't surface vague properties. Hotels that can be described in one line — "Quiet oceanview pension," "Hanok that welcomes dogs," "Boutique hotel with a wine cellar," "Guesthouse where every room faces sunrise" — appear in AI answers. "Everything's great" reads to AI as "nothing special."
Write down the strengths scattered in your head. Pick the most distinctive one. Use the exact same sentence across every channel: "Secluded oceanview pension, pet-friendly, breakfast included."
2. Centralize your information and keep it consistent everywhere
Location, room count and types, amenities, rate policies, check-in/check-out times, pet/smoking/parking policies, English descriptions — gather them in one place. Whether that's your own website, a PMS, or a channel manager that syncs property info, you need one source of truth.
AI only cites information it can verify. A property with fragmentary blog reviews, photos on Instagram, no website, and different info on every OTA gives AI nothing to trust. So AI labels the information unreliable and doesn't cite it. Using a channel manager with true single-source-of-truth (SSOT) sync makes this easy.
3. Get guests to leave specific language in reviews
AI cites "The abalone porridge stood out — the owner told me he catches the seafood himself" more than "Breakfast was good." AI uses these details as evidence when answering. Intentionally design one memorable experience and let that language flow into reviews, social posts, and external content. Generic reviews can backfire.
This information needs to be online where AI can reach it. Handwritten praise in a lobby guestbook or a website disconnected from the web doesn't help AI.
4. Get mentioned externally
AI reads external citations as trust signals. A third party AI already trusts — a local publication, a travel blogger, news — mentioning your property is 100 times stronger than you saying you're great. Pay a bit of attention to bloggers, YouTubers, and media interviews you used to ignore. The mentions add up, and they guide AI to your property.
5. Build your own channels
Keep OTA revenue, but grow direct channels and your own content at the same time. Leaning too hard on one or two channels is like putting your entire portfolio in one stock. Right now, global AI companies and OTAs are moving in different strategic directions. No one knows who wins (see table below).
AI will give trust scores to content and websites you've built consistently over years. If you don't have a website yet, it's not too late. A website AI can read well matters more than one that looks good to humans. Good tools for that already exist.
All five of these are things successful hoteliers have always done: create differentiation, organize information accurately, design guest experiences, earn external praise. Now you just need to show that work to AI — not only to humans. That makes tech infrastructure — channel managers, AI visibility tools, booking engines — more important than ever. ONDA, the company I founded, sits right in the middle of this, helping hoteliers manage digital visibility and channel integration.
Where big tech is headed
It's not all of big tech shaking up hotel bookings — it's the three companies that make AI: Google, OpenAI, Anthropic.
| Company | Move | What this means for properties |
|---|---|---|
| Search to booking to payment in one flow | "Book this for me" and Google completes the transaction | |
| OpenAI | Deploys AI engineers inside client companies to transform how they work. The team that built Virgin Atlantic's AI concierge is leading this[10] | Once you're deep in, five years later your pricing and guest service run on OpenAI |
| Anthropic | Created a standard protocol (MCP) for connecting AI to external services | Wyndham was the first major hotel company to plug hotel data into Claude (2025) and launched a ChatGPT app in May 2026. Google AI Mode integration is in the works[11] |
Skift framed the shift this way: "People use mental shortcuts — brand familiarity, loyalty. AI agents don't need shortcuts. They evaluate every option."[12] Mid-tier chains that relied on brand recognition will feel this most. Top-tier luxury will survive on strong concepts. Small independent properties with clear concepts might actually gain.
Skift called this trend "The Claude Effect" — as AI automates booking, middleman margins (OTAs) shrink and suppliers (airlines, hotels) capture more value[13]. AI-native aggregators like DirectBooker (profiled by AltexSoft) pool hotel inventory at lower commissions than OTAs. In this model, the hotel stays the merchant of record[14]. Guest data and the booking relationship stay with the hotel.
Only 2–8% of travelers say they'd trust an AI agent to complete a booking for them (varies by survey)[14]. Trust hasn't built up yet. But that number will grow. Even before that, we need to handle the discovery phase. The urgent task is to structure information so AI can surface our property in the first place.
In the middle of industry transformation
May 2026 felt like the fastest-moving month in 18 years of watching this industry transform. When the internet first appeared, when smartphones spread, when online booking platforms took hold — every time, someone changed the rules at that moment and won. That advantage lasted. This time it's even faster. Two months from now, the world might have already pushed our property far from the opportunity. Right now, "let's wait and see" strikes me as the most complacent call you can make.
There are two paths: Make your property visible on the new roads big tech and OTAs are building. Or grow a direct channel to your guests, independent of those platforms. This time, it's not hard — let's catch both rabbits.
Guests are already asking ChatGPT. Today, ask ChatGPT to recommend a hotel in your area. If your property isn't in the answer, ask why not. That's where you start.
For a step-by-step playbook, see 4-Item Checklist to Get AI to Find Your Property. For how many channels to open, see How Many OTAs Should You Sell On to Grow Revenue?.
References
- Google, "Introducing the Universal Cart and more ways to help you shop" (May 19, 2026) · blog.google
- Google Korea Blog, "[I/O 2026] A New Era of AI Search" (May 19, 2026) · blog.google
- Hotel News Resource, "The Discovery Cliff: Why Hospitality Brands Disappear in AI Search" (May 24, 2026) · hotelnewsresource.com
- Hospitality Net, "Six Forces Reshaping Independent Hotels in 2026, Including AI Discovery, Margin Pressure, and the Connectivity Imperative" (Apr 30, 2026) · hospitalitynet.org
- Hospitality Net, "Shiji Horizon Distribution and Kismet Partner to Enable AI-Driven Hotel Discovery and Direct Bookings" (May 25, 2026) · hospitalitynet.org
- ETNews, report on 2026 Korea travel trends (Nov 25, 2025) · m.etnews.com
- Hotels, "Americans are thinking twice about summer travel this year: Deloitte survey" (May 21, 2026) · hotelsmag.com
- The Korea Herald, report on ChatGPT's growing use in Korean search (Jan 29, 2026) · koreaherald.com
- Open Survey, AI Search Trends 2026 report (Feb 2, 2026) · blog.opensurvey.co.kr
- Skift, "OpenAI Builds AI Deployment Biz Around Team Behind Virgin Atlantic Concierge" (May 11, 2026) · skift.com
- Wyndham Hotels & Resorts, "Wyndham Launches Native ChatGPT App" (May 6, 2026) · investor.wyndhamhotels.com
- Skift, "B2A? AI Agents Are Travel's New Audience — And They Don't Care About Brands" (May 22, 2026) · skift.com
- Skift, "The Claude Effect Is Coming for Travel" (Mar 2, 2026) · skift.com
- AltexSoft, "Who Will Control Lodging Distribution in the Era of Agentic AI? 7 Questions Posed to Hospitality Experts" (May 20, 2026) · altexsoft.com
Frequently Asked Questions
- When someone asks AI about a property, what does it use to decide?
- AI reads information that exists as text on the web. If your property name, address, and phone number vary across channels, if key details appear only in images and not as text, or if reviews are just star ratings without specific sentences, AI has no material to work with.
- What happens when guests ask without knowing our property name?
- The gap is wide. When someone asks by name, AI returns that property 97% of the time. When they ask for a recommendation without naming a property, the chance that any specific property gets mentioned drops to 10–15%. This post addresses the second scenario.
- What should I tackle first?
- Standardize notation and add text-based information. Step one is making your property name, address, and phone number identical across all channels—down to punctuation. Step two is writing out in full sentences on your website the information that now lives only in images. After that come concrete review collection and aligning your Google Maps and other external listings. The checklist linked at the end of the post sets out the order for each item.
