February 2026Google AI Mode

Google Keeps the Click.OTAs Get Cited, Not Clicked.

We queried Google AI Mode 4,000 times to find out where hotel traffic actually goes. 79% of hotel clicks stay within Google. OTAs get 3.6%.

4,000
Queries
84,329
References
6,130
Hotel Clicks
1,146
Unique Hotels

Executive Summary

Google keeps the click. Hotels still win.

79% of hotel clicks stay within Google (via Google Business Profiles). Only 3.6% go to OTAs. Hotels get 16.3% direct β€” but since most GBP listings link back to the hotel, the effective direct access rate is ~93%.

Yet OTAs dominate reference citations β€” the small source links AI Mode uses. Intermediaries account for 46.6% of all citations. This means AI Mode learns from OTAs but sends traffic to Google and hotels.

Local Pack appears in only 6.5% of queries but when it does, 81% of its links go directly to hotel websites β€” making it the most hotel-friendly surface.

Section 1

Where Do Hotel Clicks Go?

Of 6,130 hotel mentions users can click in AI Mode responses:

~93%
Effective Direct Access Rate

Hotel website direct (16.3%) + GBP listings that link to the hotel (~97% of 79.1%) = ~93% of hotel clicks ultimately reach the hotel.

79.1%
via GBP
16.3%
Direct
3.6%
OTA

Hotel Click Destinations

Hotel Click Destinations

DestinationCountShare
Google (GBP)4,85179.1%
Hotel website (direct)1,00216.3%
OTA2213.6%
Meta/review site420.7%
Hotel chain site140.2%

OTAs get almost nothing from hotel clicks β€” only 3.6%. Google keeps 79% within its own ecosystem (GBP), and hotels get 16.3% direct. The OTA dominance is only in the reference citations, not in the actual hotel cards users click.

Two Surfaces, Two Behaviors

AI Mode displays hotels in two distinct surfaces with very different link behaviors.

Link Type by Surface

AI Overview81% of mentions

4,968 hotel mentions

GBP4,812
Direct61
Meta42
OTA39
Chain14

GBP dominates β€” Google acts as intermediary between user and hotel

Local Pack19% of mentions

1,162 hotel mentions

Direct (hotel website)941
OTA182
Google39

81% supplier rate β€” most hotel-friendly surface

Reference Citations

Separate from hotel clicks. These are the inline source links AI Mode uses to support its answers. OTAs dominate here β€” but users don't click these the same way.

46.6%
Intermediaries
TripAdvisor, OTAs, meta
21.2%
Suppliers
Hotel websites
19.9%
Google Properties
Maps, Travel, etc.
Section 2

Local Pack Appearance

How often do map results appear in AI Mode responses?

6.5%
of queries triggered a local pack

By Query Type

Local Pack Appearance by Query Type

Local Pack Rate by Query Type

Query TypeLocal Pack Rate
Boutique10%
Generic8.5%
Persona6.1%
Star Filter6%
Neighborhood1.6%

Neighborhood queries have the lowest local pack rate (1.6%) despite being the most location-specific. Boutique queries trigger it most often (10.0%).

By City

Local Pack Appearance by City

Local Pack Rate by City

CityLocal Pack Rate
London14.2%
Lisbon9.2%
Paris8.4%
Barcelona5.8%
Berlin5.4%
New York5.2%
Section 3

Reference Citations

84,329 inline citation references β€” the small source links AI Mode uses to support its answers.

References by Type

Reference Citations by Source Type

Reference Citations by Source Type

Source TypeCountShare
Metasearch/Reviews24,65129.2%
Independent Hotels17,11120.3%
Google Properties16,77719.9%
OTAs14,62717.3%
Editorial/Media9,21410.9%
UGC (Reddit, etc.)1,2131.4%

Top 10 Domains

Top 10 Referenced Domains

Top 10 Referenced Domains

RankDomainTypeCitationsShareRuns In
#1google.comGoogle16,77019.9%1,863
#2tripadvisor.comMeta13,89116.5%3,422
#3booking.comOTA3,2693.9%1,792
#4agoda.comOTA2,7663.3%1,629
#5cntraveler.comEditorial2,4632.9%1,750
#6expedia.comOTA2,2992.7%1,604

TripAdvisor appears in 3,422 of 4,000 runs (86%) β€” the most consistently cited source. Google cites itself in 19.9% of all references.

Section 4

Hotel Concentration

How concentrated are AI Mode's hotel recommendations?

1,146
Unique Hotels
45.2
Overall HHI
Low concentration
7.6%
Top 5 Share
of all mentions
51%
Appear Only Once
of unique hotels

Concentration by City

Concentration (HHI) by City

HHI Concentration by City

CityHHI
Bordeaux810.6
Vienna505.7
Lisbon403.5
Barcelona353.9
Paris335.5
Berlin300.4

HHI (Herfindahl-Hirschman Index): <1500 = low, 1500–2500 = moderate, >2500 = high.

Concentration by Query Group

Concentration by Query Group

GroupTiersUnique HotelsMentionsHHITop 5 %Once %
Generic13831,589121.315.4%56.4%
Filtered2, 3, 48193,49651.18.1%52.4%
Persona53001,045108.214.6%53%

Why hotels concentrate differently than other categories

SparkToro's research on AI consistency found high inconsistency for brands and products β€” ask the same question twice and you get different answers. Hotels are the opposite.

The reason is structural: hotels are geographically bound with finite supply. There are only so many good hotels in Bordeaux. The AI has no choice but to concentrate because the real-world inventory is limited.

Bordeaux illustrates this well: with an HHI of 810.6 (highest in the study), the top 5 hotels account for a large share of all mentions. But this isn't AI bias β€” it's geography. A smaller city with fewer hotels will naturally show higher concentration than London (HHI 156.3, lowest), which has hundreds of options.

This makes hotel AI visibility fundamentally different from product or brand visibility: your competitive set is small and fixed, and the AI knows it.

Methodology

Study Design

Data Collection

  • 40 prompts (8 cities x 5 query tiers)
  • 25 runs per prompt per country via SERPAPI
  • 4 country proxies: US, DE, FR, ES
  • Engine: google_ai_mode (no_cache=True)
  • Date: Feb 2–3, 2026
  • Total: 4,000 queries

What We Measured

  • Hotel click destinations (GBP, direct, OTA)
  • AI Overview vs Local Pack surface
  • Reference citations (URL, domain, type)
  • Local pack appearance rate
  • Concentration metrics (HHI)
  • Cross-tab outlier detection

Data Quality

100%
Variance Validation
160/160 groups show variance
8.4%
Zero-Reference Runs
337 of 4,000 runs
119
Filtered Hotel Names
Empty names removed

Query Tiers

TierExample
1. GenericBest hotels in Paris
2. BoutiqueBest boutique hotels in Paris
3. Star FilterBest 3-star hotels in Paris
4. NeighborhoodBest hotels in Paris Marais
5. PersonaBest boutique hotels in Paris for a couple

Cities

BarcelonaBerlinBordeauxLisbonLondonNew YorkParisVienna

Data Source

Data collected via SERPAPI google_ai_mode engine on Feb 2–3, 2026.

Cache disabled (no_cache=True) to ensure fresh responses for each query.

SERPAPI is the same API used by ChatGPT for web search.

See How Your Hotel Ranks in AI Search

Get a personalized AI visibility report for your property.