Research In ProgressJanuary 2026

How Consistent Are AIHotel Recommendations?

Replicating SparkToro's consistency research for hotel queries. They found AI produces identical brand lists less than 1% of the time. Our hypothesis: hotels are MORE consistent due to geographic constraints.

Data Collection Progress0.0%

0 / 4,000 runs completed

40
Prompts
100
Runs Each
8
Cities
4
Proxy Locations

The Research Question

In January 2026, SparkToro published research showing that AI tools produce consistent brand recommendation lists less than 1% of the time. This has major implications for marketers tracking "AI visibility."

Our hypothesis: Hotel recommendations should be MORE consistent than general brand recommendations because hotels are geographically constrained. There are only so many hotels in Parisβ€”unlike infinite "project management tools" or "CRM software" options. Moreover, ChatGPT grounds itself in Google Maps (through scraping). Like Shopping, there is evidence they are using the place_id for entity reconciliation.

<1%
SparkToro's list match rate for brands
?%
Our expected rate for hotels
Hypothesis: 5-25%

Cities Tested

Paris
France
megaMarais
New York
USA
megaSoHo
London
UK
megaShoreditch
Barcelona
Spain
largeGothic Quarter
Berlin
Germany
largeMitte
Lisbon
Portugal
mediumAlfama
Vienna
Austria
mediumInnere Stadt
Bordeaux
France
smallSaint-Pierre

Prompt Variations

1
"Best hotels in [city]"
SparkToro-equivalent baseline
8 cities Γ— 100 runs
2
"Best boutique hotels in [city]"
Type filter effect
8 cities Γ— 100 runs
3
"Best 3-star hotels in [city]"
Star rating filter effect
8 cities Γ— 100 runs
4
"Best hotels in [city] [neighborhood]"
Geographic constraint effect
8 cities Γ— 100 runs
5
"Best boutique hotels in [city] for a couple"
Combined constraints effect
8 cities Γ— 100 runs

Want to be notified when results are ready?

This research will be published as a follow-up to our AI Hotel Landscape 2026 report.