Heather Balsley, Chief Commercial & Marketing Officer at IHG Hotels & Resorts, is reconfiguring how 6,000-plus properties appear to large language models before guest-facing AI search becomes the default booking path. The preview announcement for her Skift Global Forum session indicates IHG is treating AI legibility as infrastructure work, not a marketing experiment.
The work involves making property data—room attributes, location context, amenity sets—structured for model ingestion rather than human browsing. IHG operates 19 brands across 100 countries, which means Balsley is essentially rebuilding metadata architecture for an estate that ranges from Holiday Inn Express to Regent. The scale implies this is a multi-quarter engineering effort, not a prompt optimization exercise. Her public framing suggests IHG is moving faster than most operators who still treat AI as a channel-marketing question rather than a foundational distribution shift.
The strategic tension is legibility versus control. Making hotels comprehensible to AI models means surrendering some narrative authority to the training corpus and inference layer. If a model decides InterContinental Bangkok is "best for business travelers near embassies," that classification propagates across every assistant using that training set. Balsley's simultaneous argument—that human relationships will remain central to guest capture—suggests IHG is treating AI as a discovery layer, not a decision layer. The bet is that models surface properties but relationships close bookings, which works only if IHG maintains direct guest touchpoints through loyalty infrastructure and post-stay engagement.
This matters because AI-mediated search is already redistributing booking volumes in ways legacy analytics cannot track. A guest asking ChatGPT or Perplexity for "best hotel in Singapore for a family reunion" does not generate a Google search event or a metasearch click. The query is invisible to traditional attribution models. If IHG properties do not appear in those inference results, the company loses not just the booking but the awareness event. Balsley's approach suggests IHG is treating this as a zero-sum visibility fight where operators who move early capture lasting representational advantages in model weights.
The timing is worth noting. Skift Global Forum runs in late September, which places this session weeks before the November earnings cycle for most public hospitality companies. If IHG has material traction data—conversion rates from AI-referred bookings, changes in direct-to-brand traffic—Balsley's session becomes a preview of fourth-quarter earnings narratives. Operators and allocators should watch for IHG disclosures around "non-traditional digital channels" or "AI-assisted bookings" in the next earnings call. Any quantification of AI-referred room nights would be the first public datapoint at scale.
The broader implication is that hotel distribution is bifurcating into legacy search and model-mediated discovery, and operators must now maintain parallel visibility strategies. IHG's move suggests the winning approach is aggressive model legibility paired with reinforced direct-relationship infrastructure. The companies that lose are those treating AI as a future problem or delegating it to a digital-marketing subcategory.
Balsley's public positioning—making hotels legible to AI while asserting models will never own the guest—is the clearest articulation yet of the dual-track strategy required. The first track is engineering work: structured data, API access, training-set representation. The second is relationship work: loyalty depth, post-stay engagement, direct-channel incentives. IHG is betting both tracks are necessary and neither is sufficient alone. The Skift session will clarify whether the company has operational evidence to support that framework or is still building the measurement infrastructure to test it.