How ChatGPT recommends hotels: what measuring a whole destination shows

Travellers don't ask AI for "hotels". They ask for a type of stay. When you measure every hotel in a destination, one traveller category at a time, it becomes clear that there is no single answer to "does AI recommend us?"

AI See You is an independent AI recommendation research company for travel. We measure which hotels ChatGPT, Claude, Gemini, Perplexity, Copilot, Google AI Mode and Google AI Overviews recommend across entire destinations. In September 2026 that covered 77 destinations in four countries, 8,944 accommodation properties and 318,641 AI recommendations analysed (source: /press, September 2026 coverage).

This article uses our public September 2026 rankings for Sydney to show what that measurement reveals. Sydney: 201 hotels ranked, 5,946 AI recommendations analysed, 25 recommendation categories tracked and nine core categories published (source: /recommendations/australia/sydney).

1. AI recommends for a traveller, not for a city

The same city produces different winners depending on who is travelling. Across Sydney's nine published categories in September 2026, seven different hotels were the most recommended:

What this means: a single question such as "best hotels in Sydney" tells a hotel very little. The commercially useful question is which recommendation markets your hotel competes in, and who AI puts ahead of you in each.

2. The competitive set changes with the traveller

The number of hotels AI ranks in each category varies widely, from 16 in Airport & Transit and 18 in Luxury to 60 in Family and 69 in Accessible. A hotel's competitors in a "romantic stay" answer aren't the same as its competitors in a "family holiday" answer.

This is why AI See You derives a competitive set from measurement, from the properties AI consistently recommends against a hotel for a given traveller, rather than from a static list chosen once (source: /methodology, "Competitive set").

3. Some hotels compete in many markets at once

Counting appearances in Sydney's nine published top-10 lists for September 2026:

• Four Seasons Hotel Sydney appears in six: #1 Family, #1 Accessible, #2 Business, #4 Luxury, #7 Romantic, #10 Wellness.
• Capella Sydney appears in five: #2 Luxury, #2 Wellness, #3 Business, #3 Romantic, #6 Boutique.
• The Langham Sydney leads two categories (Romantic and Wellness), ranks #3 in Accessible and #5 in Luxury.

Counted from the public Sydney ranking page.

Breadth and leadership are different strengths. A hotel can lead one category, place mid-list in several, and be absent from others where travellers might reasonably expect it.

4. Mentions aren't recommendations

A hotel appearing in an AI response isn't automatically recommended. AI responses include genuine recommendations, but also negative inclusions ("avoid…") and contextual references. AI See You classifies these before anything is counted (source: /methodology, "Recommendations, not mentions").

Counting every mention would overstate some hotels and understate others. Every recorded recommendation in our research traces back to its captured AI response.

5. AI cites sources, and they can be measured

Where AI platforms supply citations, we record them. For Sydney in September 2026 we captured 11,472 citations from 619 distinct recorded source domains across the seven surfaces (source: /recommendations/australia/sydney, "Sources AI cites").

Which publications and websites AI cites alongside a hotel, and alongside the hotels ranked above it, is part of the evidence behind each recommendation.

6. One prompt is not a measurement

AI recommendations can change. Our intelligence isn't based on a handful of prompts run on one afternoon. Every destination is measured monthly through the AI See You Pulse, a certified measurement that must pass certification before publication. Those readings roll up into quarterly intelligence reports, so hotels see the patterns that persist, what changed and what deserves attention (sources: home page "Repeated measurement", /methodology "Pulse").

What hotels can take from this

1. Know your recommendation markets. Find out which traveller categories AI places your hotel in, and which it leaves you out of.
2. Know who is ahead of you, category by category. Your AI competitors may not be the hotels you benchmark against today.
3. Look at the evidence, not just the rank. How AI describes the hotels ahead of you, and which sources it cites, is where the explanation usually starts.
4. Measure the whole market repeatedly. A single prompt or a single-property audit can't show the competitive picture.

About this research. AI See You measures every property in a destination with the same methodology, whether or not it is a customer. Hotels don't pay to enter our rankings. Rankings reflect captured AI recommendations during the stated measurement period and are not ratings, reviews or editorial selections by AI See You. Read the methodology →

See where your hotel stands in every category.

See the Sydney rankings →Hotel Intelligence →Talk to us →