Only 24% of hotel companies say their AI investment has paid for itself overall, according to a survey released at Destination AI 2026, which brought roughly 500 hotel brand, ownership, management and technology leaders to Washington, D.C. Not a single respondent plans to cut AI spending over the next 12 months, and 37% expect to increase it by more than 10%.

The gap between those two numbers is the story. The industry has moved from experimenting with AI to running it in live operations, and it is discovering that the technology itself may be the easy part. Change management, data quality and systems integration surfaced as barriers far more often than the limits of the AI models.

The strongest ROI claims at the conference were tied directly to the P&L, not to vague productivity. Owners cited AI-driven labor scheduling producing 200–300 basis points of margin improvement. Occupancy-aware HVAC technology cut guest-room energy consumption by as much as 45% — approximately $54,000 annually for a 250-room hotel.

Where is AI actually working?

The benefits remain concentrated at corporate. More than 80% of hotel leaders say AI has improved their own work, and 90% say it has improved time spent on routine tasks. Only 15% say AI is making a difference for frontline teams.

Corporate roles are being redefined at nearly five times the rate of frontline roles — 19% versus 4%. Saving an employee an hour does not automatically create an hour of financial value, which is why speakers pushed hotels to establish baselines and measure whether AI reduces costs, increases revenue or improves the guest experience.

Conversational AI has become the settled use case. Canary Technologies reported that adopters are automating roughly 90% of guest responses across SMS, email, voice and web chat. Canary's Harman Singh Narula put the adoption argument bluntly: "Today's AI is the worst AI you'll ever use," adding that anyone who isn't experimenting is already falling behind.

Is AI killing the hotel website?

Probably not. More than half of U.S. leisure travelers now use AI channels to help plan trips, and visitors arriving through AI channels reportedly show a 45% lower bounce rate. Hilton reported that bookings through its AI planning experiences convert a few percentage points higher.

Less than 1% of hotel website traffic came from LLMs (large language models, the systems behind tools like ChatGPT) at the time of the 2025 conference, but traffic may have been the wrong metric. The operative question is no longer how much traffic ChatGPT sends a hotel. It is whether the hotel gets recommended when a traveler asks AI where to stay.

That reframes the website as both a human conversion channel and a source of truth for machines. AI engines need accurate, consistent data on room types, amenities, restaurants, parking, pet policies and destination experiences — consistent across the website, structured data, local listings and review platforms. Traditional SEO measured rankings and clicks; AI adds a layer of being found, understood and recommended for the right traveler.

What about the data underneath?

Only 42% of hotel leaders surveyed said they were confident their data was ready for AI. Transaction codes can mean different things at different properties, information lives in disconnected systems, and AI amplifies bad data as confidently as good. Driftwood Capital's Carlos Rodriguez drew a sharp line: the value is in orchestration and unique data, not another repackaged dashboard.

Owners also said they need real-time information rather than reports that are already two weeks old, and they do not want to pay for a management company's technology stack to get it. For marketers, the implication is that paid media, analytics, CRM, revenue, reviews and attribution data cannot stay isolated if AI is expected to find patterns across the customer journey.

What happens next?

Regulation is not waiting. Federal legislation remains stalled, but hotels already face exposure around algorithmic pricing, chatbots and AI in HR, and in the EU, disclosure requirements for AI interactions and AI-generated content took effect August 2, 2026.

Technical standards remain unsettled: MCP and NLWeb are emerging as default protocols for agent readiness, while ChatGPT and Google commerce formats are still being tested. Consumers, meanwhile, are not ready to hand over the keys — most attendees had not let an AI agent complete a purchase, and trust fell further when the scenario involved two AI agents negotiating with each other.

The conversation is no longer about whether hotels will use AI. It is about where AI creates measurable value, and whether hoteliers can prove it before the next budget cycle asks them to.