Three-quarters of travelers — 75% — say good hospitality depends on people regardless of the technology in place, according to a July 2026 Mews survey of 3,250 travelers across eight markets in Europe and the US. The finding puts direct pressure on hotel operators to account for where the hours saved by AI tools actually go: into guest-facing service, or simply off the payroll line.

The survey does not show guests rejecting technology. It shows them drawing a precise line between friction removal and attention. Among the respondents:

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

The strength of that human preference varies sharply by touchpoint, and that variance is the practical map for operators. At the top end, 72% want a human for bar and restaurant service, 66% for the front desk welcome, and 58% for check-in and keys. At the weak end, only 44% prefer a human for booking a taxi (28% prefer AI), and 50% for room service (23% prefer AI).

What happens to the saved hour?

The author of the analysis identifies a structural pattern in hotel operations after an AI deployment. A chatbot absorbs parking questions, upsell emails send themselves, and the night audit report builds itself. The team recovers, say, 90 minutes a day. Then the 90 minutes disappears — absorbed by an unfilled open position, another dashboard, a cut in 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, the analysis argues. If the recovered hour never reaches a guest, the operator has simply automated its way to a cheaper version of the same stay.

Does the reinvestment math matter for the brand?

One data point in the Mews research carries brand-level consequences: 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 the technology.

That makes headcount decisions visible to guests. A thinner, more transactional stay confirms their concern; a stay where staff demonstrably have more time for guests refutes it.

The proposed measurement is a simple ratio: the Reinvestment Rate — hours redeployed into named guest moments divided by hours saved by AI. Near zero means the AI deployment is a cost play. A climbing rate means the property is building service value competitors cannot copy with a chatbot.

What does reinvestment look like in practice?

The analysis sketches an illustrative 40-room independent hotel. An AI assistant handles pre-arrival questions — parking, breakfast hours, early check-in, restaurant bookings — returning roughly 90 minutes a day to reception. The GM assigns those minutes to three named moments:

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

The metrics to track are staff names mentioned in reviews, service issues fixed before checkout, and direct rebooking — not hours saved.

Which touchpoints should operators automate first?

The Mews data provides a sequencing rule. 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, and service recovery.

Order of operations matters too. If the PMS, guest data, and handover processes are messy, the hours AI frees get consumed by workarounds before they reach a guest. The prescription is fixed foundations first, then automation, then reinvestment.

One caveat the source itself flags: Mews sells hotel software and commissioned the research. The release specifies sample size, markets, and fieldwork date but not full weighting or the questionnaire. The numbers indicate a strong direction rather than a law.

Operators evaluating the next AI contract should ask one question first: show me the hour. Until a property can point to the guest moment where saved time landed, it has not improved the guest experience — it has made it cheaper.