- Guests want back-office AI, people up front.
- Only 24 percent of leaders report AI payback.
- Owners should measure tools against a baseline.
GUESTS WANT HOTEL AI working in the background rather than running the front desk, according to Mews. That makes back-office automation a potentially lower-risk place for owners to seek returns, but few hotel leaders can yet show that their AI spending has paid off.
A Mews survey released Oct. 7 found that 89 percent of 3,250 travelers wanted either a combination of AI and human service or AI working behind the scenes on repetitive tasks. Separately, 75 percent said good hospitality depends on human staff, and 59 percent said AI should take over repetitive tasks so employees have more time for guests.
Guests drew a clear line, according to Mews, a hotel software provider. Respondents preferred human assistance at all nine service touchpoints tested. Some 72 percent preferred a person for bar and restaurant service, 66 percent for the front desk welcome and 58 percent for check-in and key collection. Even when arranging taxis, 44 percent chose a person, compared with 28 percent who chose AI.
"Guests are telling us exactly where they want AI and where they don't," said Richard Valtr, founder of Mews. "The hotels that win this next decade won't be the ones that automate the most. They'll be the ones whose systems are unified enough that AI can actually deliver a new way of operating, and where staff can get back to providing a genuine human experience."
The survey was conducted in July 2026 across the U.S., France, Germany, Italy, the Netherlands, Spain, Sweden and the U.K.
Savings versus service
Hotels are prioritizing cost savings over improvements in service.Photo credit: iStock
The State of Hotel AI survey shows what hotel companies are chasing. Of 107 hotel company leaders surveyed in August and September, 34 percent named cutting costs and improving productivity as their main AI goal for the next 12 months, the largest single response. Another 32 percent weighed several goals equally.
Only 13 percent identified new revenue generation as a priority, while 9 percent cited guest experience. No respondent planned to spend less on AI, and 37 percent expected to spend more than 10 percent more. Set beside the Mews findings, the focus on cost highlights a tension.
The largest group of hotel leaders puts lower costs first, yet Mews found that guests accept AI for repetitive work and want people where judgment and trust matter. The study also found that 67 percent of travelers would worry about staff losing jobs to AI, while 51 percent disagreed that they would not care if a hotel cut staff to adopt it.
The surveys do not show how these views affect bookings or profits, and they were not designed together. The link between them is therefore an inference. It suggests that savings from back-office work may carry less guest risk than savings from thinner guest-facing teams.
Returns have not caught up. Only 24 percent of leaders said their AI investment had paid for itself overall, and 28 percent said it was too early to tell. Although 90 percent said AI had improved the time they spend on routine tasks, the study notes that personal productivity gains do not automatically establish organizational returns.
Some 67 percent cited change management, data quality or systems integration as barriers, compared with 8 percent who cited limitations of AI models. Only 42 percent were confident their data was ready for AI.
The study also shows where change has landed so far. Leaders reported redefining corporate roles far more often than frontline roles, at 19 percent against 4 percent. These figures suggest that much of the change so far has affected office teams rather than the hotel staff guests meet, although that is an inference.
Proof comes next
Hotels must now demonstrate measurable results from their AI investments.Photo credit: iStock
The likely change is a higher standard of proof. The State of Distribution 2026 report, developed by NYU's Jonathan M. Tisch Center of Hospitality, HEDNA and RateGain, found that fewer than one in 10 hotels reported that generative AI had cut their manual work by more than 30 percent.
RateGain founder and managing director Bhanu Chopra said buying AI was easy, but getting value from it was not. With no leader in the State of Hotel AI survey planning to spend less, owners are likely to be asked what each tool saved and how staff used the time gained. That expectation is an inference.
Owners and operators can start with one back-office task. In the State of Hotel AI study, 38 percent of leaders said operational efficiency was where AI delivered its strongest results. Owners should record a baseline, such as staff hours or response times, and decide in advance how the freed time will be used.
The study also found that workflow redesign, at 32 percent, was among the leading drivers of AI success, well ahead of executive mandates at 14 percent.
Guest-facing automation needs testing first. Mews recommends automated check-in and check-out, yet 58 percent of its respondents preferred a person for check-in and key collection. Operators should test automation at one property and measure guest feedback before rolling it out across a portfolio.
Investors can ask management teams to show the baseline, the total cost of each tool and the measured results behind AI-heavy plans, because few leaders report payback so far. Developers can ask whether property management, payment, housekeeping and guest messaging systems can share information before committing to new technology.
The NYU, HEDNA and RateGain benchmark found that disconnected systems and vendor fragmentation continue to hold hotel commercial teams back.
Separately, a new Hospitality AI Use Case Catalogue shows how hotels are exploring AI across different business functions.
