IHG Hotels & Resorts is teaching large language models to parse 6,000 properties across 19 brands while maintaining that artificial intelligence will never control the guest relationship. Chief Commercial & Marketing Officer Heather Balsley previewed the strategy ahead of Skift Global Forum, where she speaks September 18th.

The investment centers on structured data feeds that make IHG inventory legible to AI travel planners—OpenAI's ChatGPT, Google's Search Generative Experience, Perplexity, and the half-dozen models hospitality operators now track. IHG is formatting room attributes, amenity details, location context, and real-time availability into machine-parseable schema. The work runs parallel to traditional SEO but requires different taxonomies. A model needs to understand "poolside cabana with refrigerated minibar" as distinct structured data, not unstructured marketing copy. Balsley's team is building that layer across properties from Holiday Inn Express to Six Senses, a range that spans $80 roadside rooms to $2,400 overwater villas.

The operational tension is deliberate. IHG is spending to ensure models surface its properties when a user asks "find me a pet-friendly hotel near Heathrow with EV charging," but Balsley argues the transaction and relationship remain with IHG, not the AI layer. She positions models as discovery infrastructure, not distribution partners. The distinction matters because it determines where customer data lives, who controls the upsell conversation, and whether IHG pays a commission or simply benefits from increased qualified traffic to its own booking engine.

This maps onto a broader reallocation inside hotel groups. Marriott spent $50 million on similar legibility work in late 2025. Hilton added 22 engineers to its data-feeds team in Q1 2026. The spend is defensive and speculative—defensive because invisibility to AI models risks losing 18-to-34-year-old bookers who now start travel research in ChatGPT, speculative because nobody knows whether models will send traffic to brand sites or capture the booking themselves. Balsley is betting IHG can remain the merchant of record if it moves early on structured data and maintains product quality the model cannot replicate. The guest wants a concierge who remembers their name, not a chatbot that confirms their reservation.

Operators and agency strategists should watch three follow-on moves. First, whether IHG launches a model-specific API by Q4 2026 that lets AI platforms pull live inventory without scraping. Second, whether loyalty-program integration follows—can a model book a room using your IHG Rewards points, and does IHG allow that. Third, how IHG measures incrementality. If 30% of model-referred bookings would have happened anyway through Google or Booking.com, the ROI math changes. Balsley has not published those numbers.

IHG operates in 100 countries. The legibility work scales unevenly because schema standards for hotels remain fragmented and some markets lack reliable broadband to support real-time API calls from rural properties.