What to do this week

Run the OTA citation test. Ask ChatGPT and Perplexity to recommend hotels in your market. Note which sources are cited in the response. If your property appears only through OTA links, that's your baseline.

Ensure your website is a credible source. Check that room types, amenities, location details, and policies are in the raw HTML, not hidden behind JavaScript. Add Hotel schema markup if you haven't already.

Ask your tech partners about MCP. Whether you use Cendyn, Mews, Cloudbeds, or another platform, find out if they're building MCP integrations. This is the infrastructure that will determine whether AI books through you or through an OTA.

Boost Background

Every hotel operator knows the math. A $300 room booked through Booking.com or Expedia costs $45 to $90 in commission. Multiply that across thousands of annual bookings and the number is staggering. What most operators don't realize is that AI search is quietly reinforcing that dependency. When a traveler asks ChatGPT for hotel recommendations, OTA listings account for 55.3% of the sources cited in the response. Your property's own website? Just 13.6%.

The platforms taking 15 to 30% of your revenue are also the ones feeding AI the information it uses to recommend you. That's a structural problem. And it's one that premium hospitality brands are uniquely positioned to solve.

How OTAs captured AI visibility

This wasn't a deliberate strategy by Booking.com or Expedia. It's a byproduct of how AI models find and trust information. OTAs have massive, well-structured, crawlable databases with consistent formatting, real-time pricing, standardized descriptions, and millions of reviews. That's exactly the kind of content AI systems prioritize.

Cloudbeds' study of 145 hotels across six destinations confirmed the pattern. Booking.com was the most frequently cited source, followed by Expedia and TripAdvisor. Meanwhile, independent properties were recommended at a 4.4-percentage-point lower rate than branded hotels. The structural advantage isn't the OTA's brand, it's their data infrastructure. Their pages are clean HTML. Their schema markup is comprehensive. Their content updates in real time. Most hotel websites can't match any of that, and AI models notice.

HotelRank's analysis of over 10,000 prompts across GPT-5 models found the pattern intensifies by price segment. Budget and mid-range properties saw OTA links in up to 50% of AI responses, while luxury properties saw significantly fewer. The higher the hotel's star rating and price point, the more likely AI was to surface direct links. Premium properties already have an edge in AI, but only if they build the digital infrastructure to capitalize on it.

The commission math AI is changing

Consider what's at stake. Independent hotels currently send roughly 61% of their online bookings through OTAs, according to Phocuswright data cited in a WebProNews analysis. At average commission rates of 15 to 25%, that's a massive revenue transfer. Direct bookings consistently yield 20 to 30% higher profit per reservation after factoring in commission savings and lower acquisition costs.

AI search is creating a new path to direct visibility, but only for properties that prepare. When ChatGPT's browsing tool or Perplexity's citation engine visits your website, they look for clear content in the HTML source, structured data about your rooms, and consistent information matching third-party platforms. If your website delivers that, you become a source AI can cite directly. If it doesn't, the OTA listing becomes the default.

The properties we work with that have invested in server-side rendered content, Hotel and LocalBusiness schema, and FAQ pages covering the questions travelers actually ask are starting to see a shift. Their websites appear as cited sources alongside OTAs rather than being invisible.

MCP: the emerging direct channel

The biggest structural change is still unfolding. Model Context Protocol, or MCP, is an open standard that allows AI tools like ChatGPT, Claude, and Gemini to connect directly with hotel systems and pull live availability, rates, and inventory. Cendyn launched AI Connect in late 2025 in partnership with DirectBooker, pushing hotel rates directly into AI platforms through MCP. Other providers like The Hotels Network are building similar solutions.

The implication: instead of AI pulling your rates from Booking.com's listing, MCP allows it to pull directly from your booking engine. The traveler sees your direct rate inside the AI conversation. No intermediary. No commission.

This is early. Agentic booking is still maturing. But hotels building MCP-ready infrastructure now will have months of lead time when adoption arrives. In AI, early positioning compounds.