US July RevPAR Up 8.2% on a World Cup Bump, Half of All AI Pricing Advice Gets Overridden, Retention Is Up but Commitment Is Not
Thursday's monthly benchmark is strong and partly borrowed. New York's World Cup Final did much of the heavy lifting at the top of the table, which makes it a number to enjoy now and to budget around next summer. The more interesting convergence happened elsewhere: four separate pieces published within a few hours of each other arrived at the same conclusion about AI, and none of it was about the models. July RevPAR Rose 8.2%, and New York Did a Lot of the Lifting CoStar's July data puts US occupancy at 69.7%, up 2.3%, ADR at $171.74, up 5.7%, and RevPAR at $119.77, up 8.2% year on year. New York City, which hosted the World Cup Final, posted the largest gains in the Top 25 on both rate and yield: ADR up 24.0% to $351.18 and RevPAR up 27.1% to $305.74. Detroit was the only Top 25 market with a double-digit occupancy increase, up 10.7% to 70.7%, and 22 of the 25 improved RevPAR. Part two of the H1 global review, published this morning, gives the shape underneath the month. Global occupancy moved within roughly one to two percent of last year while ADR carried RevPAR, which makes the half pricing-driven rather than demand-driven. Europe is close to its practical occupancy ceiling and now competes on revenue management rather than volume. North America holds rate without losing demand. Asia Pacific grows on both, led by Japan, Vietnam and Thailand. The behavioural findings matter more than the regional ones: booking windows keep shortening, matching Monday's Expedia data on EMEA, and average length of stay is falling, with urban and business trips moving from five or six nights toward two or three. Set that against Tuesday's note from STR that World Cup host markets will create difficult year-over-year comparisons next summer, and the New York line above is exactly the number that needs handling carefully in a 2027 budget. Read the July data → Four Pieces, One Conclusion: the AI Bottleneck Is Human LodgIQ put a number on it. More than half of the pricing recommendations an unexplained system produces are overridden across the industry, and the argument is that this is not carelessness. A recommendation without reasoning attached is not something a revenue manager can defend to a general manager or an owner, so it sits in a queue instead of changing a rate. The same piece cites lodging technology research in which 42% of hoteliers say their own employees experience workplace technology as friction rather than help, usually because training was rushed and interfaces were left too complex to learn on the job. The proposed fix is sequencing: let the system suggest while a person decides, then let it act inside guardrails once the logic has proven itself, then monitor at a strategic level. Skip the stages and adoption stalls regardless of how good the underlying model is. A former front desk agent and housekeeping supervisor made the labour version of the same argument. US hotels are on track to
Share