AI Knows Your Pillow Preference.

It Still Can't Tell You Why the View Was Breathtaking

Two conversations last month — a hotel technology vendor briefing where I was shown an AI personalisation roadmap, and a client programme review where we were assessing preferred hotel performance. The pitch and the problem were running in parallel without touching each other.

The pitch: AI will deliver luxury-calibre service at every price tier, personalising every stay to the individual traveller. The problem: one of the preferred hotels in the programme — chosen for its character and the fact that employees actually liked staying there — had just signed with the same AI personalisation vendor I’d seen in the briefing.

I started asking what exactly is being personalised when every hotel uses the same platform. Then I started worrying about what travel managers are selecting for in their sourcing reviews without realising it is changing.

What surprised me most in those briefings: nobody in the room was asking what falls outside the measurement. AI personalisation in hospitality works on what it can measure: floor preference, room service timing, loyalty tier. It cannot rank the view that made a guest rebook. That signal lives in reviews — “breathtaking,” “a sense of arrival you can’t manufacture” — not in the booking infrastructure where AI agents shortlist from. The luxury signal never reaches the algorithm.

The deeper problem: most hotels are sourcing AI from the same short vendor list. A boutique design hotel and a business property can encode different brand values into the same platform — but they are running the same underlying model. Distinctiveness as an input does not guarantee distinctiveness as an output.

The question worth adding to your next preferred hotel sourcing review: will the properties you selected for their character still feel the same after their AI stacks are fully deployed?

Read the full analysis → rajeevgoswami.me/ai-luxury-hotel-personalisation

In the full post:

• Why the 34% ADR premium boutique hotels command is exactly what vendor-concentrated AI will quietly erode — and why McKinsey’s 3–10% revenue lift projection may be the industry’s most expensive distraction

• How the GDS (the global booking infrastructure your OBT uses) was built to rank functions, not experiences — and why the luxury signal never reaches the shortlist

• Why neither side of the transaction — travellers extending business trips into bleisure, or hotels protecting a rate premium — can afford for AI to flatten the difference