← All press releases
Press release

AI assistants name classified hotels around two and a half times as often

Study
November 23, 2026

A new Bluerails study based on the official German Hotel Classification database analyses around 60,000 queries to ChatGPT, Claude, Gemini and Perplexity. Officially classified hotels are named in 48% of answers, unclassified neighbours in the same town in 20%. The gap grows with every star category.

Based on the official German Hotel Classification (Deutsche Hotelklassifizierung) database

“
AI assistants need reliable, verifiable information to make recommendations. That is exactly what the official hotel stars have provided for 30 years. The study shows that classified hotels are significantly more visible in this new channel than their local competitors.
Christin Neumann
Managing Director, DEHOGA Deutsche Hotelklassifizierung GmbH
“
We wanted to know whether officially classified hotels are more visible in AI search than their neighbours in the same town. The answer is clear, and it holds for each of the four AI assistants we tested. What counts next is whether AI can also book the hotel directly. Hoteliers should therefore make their stars, prices and availability readable for machines too.
Jens Mannanal
CEO and Co-Founder, Bluerails
48% vs. 20%
of answers name the hotel: classified vs. unclassified properties in the same town
60% vs. 15%
for five-star hotels vs. their unclassified neighbours
55% vs. 30%
can be booked online on their own website
Bar chart: AI assistants name classified hotels more often than unclassified neighbours in the same town. 5 stars 60% vs. 15%, 4 stars 52% vs. 18%, 3 stars 45% vs. 21%, 2 stars 37% vs. 22%.

Key findings

  • Named around two and a half times as often. 48% of answers name the classified hotel, 20% the unclassified neighbour in the same town.
  • The gap grows with every star category. +16 percentage points at two stars, +24 at three, +34 at four and +45 at five stars.
  • The pattern holds for all four assistants. ChatGPT, Claude, Gemini and Perplexity, each analysed separately.
  • Almost twice as often bookable online. 55% of classified hotels offer online booking on their own website, compared with 30% of the comparison properties.

About the study

The study compares 1,000 pairs, each made up of an officially classified hotel and a randomly selected unclassified property in the same town (2,000 properties in total), across around 60,000 queries to ChatGPT, Claude, Gemini and Perplexity. It measures whether a hotel is named in the answer, not whether it is recommended. The study shows a clear correlation; it does not claim to prove causation.

The full study is available in German. Bluerails offers hotels services around AI visibility and has agreed an intention to cooperate with DEHOGA Deutsche Hotelklassifizierung GmbH.