Three in four travelers — 75% — say good hospitality depends on people, whatever the technology. That figure anchors new research from hotel software vendor Mews, which surveyed 3,250 travelers across eight European and US markets in July 2026, and it frames an uncomfortable question for operators running AI projects: does the time saved by automation ever reach the guest?

The data does not show guests rejecting AI. It shows them sorting tasks into two piles.

• 89% want either a mix of humans and AI, or AI kept backstage on repetitive work.
• 59% want AI to absorb repetitive tasks so staff can build trust with guests.
• Humans were preferred at all nine touchpoints tested.

The preference for humans is strongest at revenue-critical service moments: 72% want a person for bar and restaurant service, 66% for the front desk welcome, 58% for check-in and keys. It is weakest where the transaction is purely functional: booking a taxi (44% human, 28% AI) and room service (50% human, 23% AI).

What does the data mean for staffing decisions?

One figure should concern every owner: 67% of travelers would worry about staff losing their jobs to AI, and 51% disagree with the statement that they would not care if a hotel cut staff to adopt it. Replacing positions with software is no longer a purely internal HR decision. Guests read it as a brand signal, and a thinner, more transactional stay confirms their fear.

Where do the saved hours actually go?

The common pattern, according to the analysis: a chatbot takes the parking questions, the upsell email sends itself, the night audit report builds itself, and the team recovers roughly 90 minutes a day. Then the 90 minutes disappears — absorbed by an unfilled position, another dashboard, next month's rota, or the general blur of a shift.

The vendor reports success. The GM sees a lower payroll line. The guest notices nothing. Time saved is an input, not an outcome. If the recovered hour never reaches a guest, the hotel has simply automated its way to a cheaper version of the same stay.

How can operators measure reinvestment?

The piece proposes a single metric, the Reinvestment Rate: hours redeployed into named guest moments, divided by hours saved by AI. Near zero means the AI is a cost play. Climbing means the hotel is building service that competitors' chatbots cannot copy.

A worked example for a 40-room independent shows the mechanics. An AI assistant handles pre-arrival questions — parking, breakfast hours, early check-in, restaurant bookings — returning about 90 minutes a day to reception. The GM then assigns that time deliberately:

• Two staff at the desk during the peak arrival window, with no screens to clear.
• A two-minute "how was your night?" check for every guest the morning after, catching the noisy room or broken shower before it reaches a review.
• A front desk handover to the bar team of one personal detail per guest — the touchpoint where 72% of guests want a human anyway.

The measures to watch, the analysis argues, are staff names mentioned in reviews, service issues fixed before checkout, and direct rebooking — not hours saved.

Automate the taxi, protect the bar

The practical read of the Mews numbers is a map. Automate where guests are relaxed: transport, routine requests, FAQs, admin, reporting. Protect and invest where they want a person: the welcome, the keys, the table, the bar, service recovery.

Sequencing matters. If the PMS, guest data and handovers are messy, the hours AI frees get consumed by workarounds before they reach a guest. The recommended order is fixed: fix the foundations, then automate, then reinvest.

A caveat on the source

Mews commissioned the research and sells hotel software. The release specifies sample size, markets and fieldwork date but not full weighting or the questionnaire. The numbers should be treated as a strong directional signal rather than settled fact — though the direction, the author notes, matches what guests and hotel teams report week to week.

The immediate next step for operators is practical: run the count this week, track the hours AI tools actually save, and name the guest moment each hour went into. Until a hotel can point to that moment, it has not improved the guest experience — it has made it cheaper.