Causal-Inference Study: AI Autopilot Pricing Lifts Hotel Revenue Per Square Meter by 13%

Most AI revenue-management case studies compare a property’s performance before and after adoption. A new study of more than 6,000 hotels examining Mews RMS’s “Autopilot” AI dynamic-pricing feature does something rarer: it uses causal inference to match adopting properties against similar non-adopting ones in the same markets. The result — a 13% lift in revenue per square meter over 18 months for hotels that ran Autopilot nine or more months — carries statistical odds the study describes as “seven in ten thousand” of occurring by chance.

The mechanism is pricing frequency, not smarter pricing in the abstract: manual price changes averaged roughly 120 a month before Autopilot, jumping to more than 1,700 a month under full automation — a 14x increase, with top-adopting properties exceeding 4,700 monthly price changes, far beyond what a human revenue manager could sustain. Notably, the system lifted both average daily rate and occupancy simultaneously, the outcome revenue managers typically struggle hardest to achieve by hand. Adoption is still partial: only 55% of Mews RMS customers currently run full Autopilot, and the algorithm needs roughly nine months of operation before its impact becomes measurable.

The figure lines up with what Mews found in its own March 2026 survey of 500+ hoteliers: hotels using its revenue management system saw a 13.7% revenue uplift per square meter in that dataset too — two different measurement approaches landing on nearly the same number. That survey also found fragmented technology stacks and data-accuracy concerns as the leading barriers to hotels moving past pilot mode, which is the more common failure point than the pricing algorithm itself.

For revenue leaders evaluating AI pricing tools, the takeaway is less about which vendor and more about commitment: the gains showed up only after nine months of full deployment, not from a pilot running alongside the old manual process.