TL;DR
LuxDirect analysed responses across six AI systems in April 2026: ChatGPT, Claude, Gemini, Grok, Perplexity, and Google AI Mode. Conversational visibility testing was conducted across five platforms, with Google AI Mode analysed separately for citation and OTA routing behaviour. Across 9,600 conversational AI responses from 15 Manchester independent luxury hotels, two properties captured 25.9% of all AI mentions. Ten of 15 properties had zero unprompted competitive visibility across all five conversational platforms. AI recommended direct booking in the majority of booking intent interactions but provided no URL in 88% of responses. Google AI Mode cited online travel agency domains at a ratio of 2.8 to 1 over hotel direct booking domains. Fourteen of 15 properties had no Hotel schema markup. Manchester's AI visibility market remains less consolidated than London's. The window for independent properties to establish stronger AI visibility before the market consolidates remains open.
In April 2026, LuxDirect completed CS9-Manchester across six AI systems: ChatGPT, Claude, Gemini, Grok, Perplexity, and Google AI Mode. The study ran 2,475 structured prompts across five conversational AI platforms (ChatGPT, Claude, Gemini, Grok, and Perplexity), generating approximately 9,600 responses across repeated query runs and platform iterations. Google AI Mode was analysed separately via the DataForSEO API for citation and OTA routing behaviour, which uses a distinct methodology from conversational platform testing. The study measures how AI systems surface, describe, and recommend hotels in conversational responses. Responses were scored only when a hotel was explicitly named or unambiguously referenced. Generic references such as "a luxury hotel in central Manchester" were excluded unless attributable to a specific property. Findings are based on structured query testing across six AI systems via API. This is the third in LuxDirect's series of structured AI visibility studies covering London, Cambridge, and Manchester.
Visibility is concentrated and the gap is structural
Across 9,600 conversational AI responses, platforms generated 1,525 hotel mentions. The distribution is not close to even. The top two properties captured 25.9% of all AI mentions between them. The top five captured 56.1%. The bottom five collectively held 13.7%. The concentration pattern replicates LuxDirect's London finding (57.2% top five share) and its Cambridge finding. Observed patterns across LuxDirect's London, Cambridge, and Manchester studies suggest visibility concentration increases as AI systems accumulate more reference material around frequently cited properties.
Full visibility rankings: all 15 properties
Hotel identities are anonymised. Full named reports are available to clients.
| Rank | Property | Type | Mentions | ASOV | Mention rate |
|---|---|---|---|---|---|
| 1 | Property A | Independent | 200 | 13.1% | 31% |
| 2 | Property B | Independent | 195 | 12.8% | 30% |
| 3 | Property C | Independent | 158 | 10.4% | 25% |
| 4 | Property D | Independent | 156 | 10.2% | 24% |
| 5 | Property E | Independent | 146 | 9.6% | 23% |
| 6 | Property F | Independent | 103 | 6.8% | 16% |
| 7 | Property G | Independent | 100 | 6.6% | 16% |
| 8 | Property H | Independent | 94 | 6.2% | 15% |
| 9 | Property I | Independent | 92 | 6.0% | 14% |
| 10 | Property J | Independent | 72 | 4.7% | 11% |
| 11 | Property K | Independent | 60 | 3.9% | 9% |
| 12 | Property L | Branded | 46 | 3.0% | 7% |
| 13 | Property M | Independent | 42 | 2.8% | 7% |
| 14 | Property N | Independent | 35 | 2.3% | 5% |
| 15 | Property O | Independent | 26 | 1.7% | 4% |
Top 5 ASOV share: 56.1% · Bottom 5 ASOV share: 13.7%
ASOV: AI Share of Voice, calculated as each property's total mentions as a proportion of all 1,525 hotel mentions recorded across the conversational platform study. Mention rate: percentage of conversational AI responses in which each property appeared across all query categories.
67% of properties have zero unprompted competitive visibility
When conversational AI platforms are asked about a specific property by name, the average hotel in the study appears in 60% of responses. When platforms are asked to recommend the best luxury hotels in Manchester without a named prompt, that figure falls to 7%. For 10 of 15 properties, the unprompted competitive mention rate is exactly zero across all five platforms.
AI knowledge of the property exists. AI advocacy does not.
This distinction matters commercially. A hotel that AI recognises but never recommends occupies a structurally weak position that product quality alone does not appear sufficient to correct.
Platform behaviour is divergent
Each conversational platform exhibits distinct recommendation patterns. The mean mention rate varies from 20.7% on Claude to 13.3% on ChatGPT. Several properties score near zero on one platform while maintaining measurable presence on others.
| Property | ChatGPT | Claude | Gemini | Grok | Perplexity | Total |
|---|---|---|---|---|---|---|
| Property A | 41% | 47% | 24% | 16% | 27% | 31% |
| Property B | 31% | 55% | 16% | 20% | 30% | 30% |
| Property C | 23% | 33% | 16% | 32% | 20% | 25% |
| Property D | 14% | 41% | 19% | 14% | 34% | 24% |
| Property E | 18% | 26% | 16% | 16% | 38% | 23% |
| Property K | 0% | 9% | 12% | 13% | 13% | 9% |
| Property L | 4% | 8% | 10% | 12% | 2% | 7% |
| Property N | 1% | 5% | 13% | 3% | 5% | 5% |
| Property O | 1% | 3% | 7% | 9% | 0% | 4% |
Selected properties shown. 0% means no mentions across the full study on that platform.
Property K registered 0% on ChatGPT while scoring 12–13% on Gemini and Grok. Property L registered 2% on Perplexity while scoring 12% on Grok. The platform a traveller opens determines whether a hotel exists in their consideration set at all. A hotel monitoring a single platform receives a fundamentally misleading picture of where it actually stands.
Booking intent is where commission leakage happens
For branded and discovery queries, conversational AI platforms mention properties and reference hotel context directly. The risk shifts sharply when a guest signals booking intent. Across booking intent queries, AI failed to provide a direct hotel booking path in approximately 77% of interactions market-wide. No property achieved direct booking channel inclusion in more than one in three booking intent interactions.
| Property | Direct booking channel inclusion | Interactions without direct path |
|---|---|---|
| Property A | 33.3% | 66.7% |
| Property E | 33.3% | 66.7% |
| Property B | 30.0% | 70.0% |
| Property D | 30.0% | 70.0% |
| Property C | 28.3% | 71.7% |
| Properties F, G, H, I | 25.0% | 75.0% |
| Property J | 23.3% | 76.7% |
| Properties K, L | 18.3% | 81.7% |
| Properties N, O | 15.0% | 85.0% |
Google AI Mode: citation behaviour analysed separately
Unlike the five conversational platforms, Google AI Mode was analysed for citation and OTA routing behaviour rather than conversational recommendation visibility. These are distinct measurements requiring distinct methodology. Google AI Mode responses were collected via the DataForSEO API across 23 query types for all 15 hotels. Analysis extracted 481 citation URLs from response text.
| Metric | Value |
|---|---|
| Total citation URLs extracted | 481 |
| Online travel agency domain citations | 70 (14.6%) |
| Direct booking domain citations | 25 (5.2%) |
| OTA to direct citation ratio | 2.8 to 1 |
| Properties with zero direct citations | 3 of 15 |
The property with the highest ASOV in the conversational platform study received 17 online travel agency citations and zero direct booking citations across its Google AI Mode responses. Observed patterns suggest high conversational AI visibility does not reduce online travel agency citation exposure on Google AI Mode. The most visible properties tend to generate more citation material overall, including more online travel agency listing pages, which AI systems appear to treat as authoritative reference material alongside direct hotel sources.
14 of 15 properties have no Hotel schema markup
A schema markup audit of all 15 hotel homepages found that 14 of 15 properties have no Hotel or LodgingBusiness structured data. Zero properties have aggregate rating schema. Zero have checkin and checkout time fields.
| Check | Result |
|---|---|
| Any JSON-LD schema detected | 6 of 15 (40%) |
| Hotel or LodgingBusiness type present | 1 of 15 (7%) |
| Street address field present | 3 of 15 (20%) |
| Aggregate rating schema | 0 of 15 (0%) |
| Checkin and checkout time fields | 0 of 15 (0%) |
| Booking engine detectable in page source | 2 of 15 (13%) |
AI systems rely disproportionately on third-party sources when structured property data is absent. Observed patterns from LuxDirect's three UK market studies suggest schema presence does not directly predict high AI mention rates: editorial coverage and direct search authority appear to be stronger determinants. However, properties combining low editorial coverage with absent structured data consistently appear at the bottom of the visibility distribution. Schema appears to function as a floor-raiser rather than a ceiling-breaker.
Summary
CS9-Manchester studied six AI systems across 15 Manchester independent luxury hotels. Conversational visibility testing across five platforms generated approximately 9,600 responses; Google AI Mode was analysed separately for citation and OTA routing behaviour. Two properties captured 25.9% of all conversational AI mentions. Ten of 15 properties had zero unprompted competitive visibility. AI recommended direct booking in text but provided no URL in 88% of booking intent responses. Google AI Mode cited online travel agency domains at 2.8 times the rate of direct hotel domains. Fourteen of 15 properties had no Hotel schema markup. Manchester's AI visibility market remains less consolidated than London's. Independent properties that address their AI signals now will be in a materially stronger position than those that engage reactively.
Frequently asked questions
How concentrated is AI hotel visibility in Manchester?
Significantly, though Manchester's AI visibility market remains less consolidated than London's (57.2% top five ASOV share) or Cambridge's (87% top four share within that study's sample). In the Manchester conversational platform study, the top five properties captured 56.1% of all AI mentions and the bottom five held 13.7%. Ten of 15 properties had zero unprompted competitive visibility across all five platforms, meaning AI never included them when asked to suggest the best luxury hotels in Manchester without a named prompt.
Why does platform divergence matter commercially?
The platform a traveller opens determines whether a hotel exists in their consideration set at all. In the Manchester study, one property registered 0% on ChatGPT while scoring 12–13% on Gemini and Grok. Another registered 2% on Perplexity while scoring 12% on Grok. A property monitoring a single platform receives a fundamentally misleading picture of where it actually stands. A hotel invisible on one platform but visible on others has a platform-specific gap that appears specifically addressable without a full content overhaul.
When does online travel agency leakage occur in AI interactions?
At the moment of booking intent. For branded and discovery queries, conversational AI platforms mention properties and reference hotel context directly. When a guest signals transactional intent, AI failed to provide a direct hotel booking path in approximately 77% of booking intent interactions. In 88% of those responses, AI provided no URL at all.
What is the most immediately actionable finding from the Manchester study?
Schema markup implementation. Fourteen of 15 properties have no Hotel or LodgingBusiness structured data on their website. Zero have aggregate rating schema. Zero have checkin and checkout time fields. Implementing these takes a developer less than one hour and removes a recognition barrier currently affecting 14 of 15 properties. Schema alone will not produce top-tier mention rates, but its absence combined with low editorial coverage consistently appears at the bottom of the visibility distribution across LuxDirect's dataset.
What percentage of Manchester properties have zero competitive AI visibility?
67%. Ten of 15 properties had zero unprompted competitive visibility across all five conversational platforms. When AI is asked to recommend the best luxury hotels in Manchester, these properties do not appear in any response on any platform, regardless of product quality or guest satisfaction scores.
How does Manchester compare to London?
Although the top five concentration figures appear similar between Manchester (56.1%) and London (57.2%), London exhibits sharper long-tail collapse, with lower-tier properties appearing less consistently across platforms and query types. Manchester retains a broader mid-market visibility layer at this stage. Observed patterns across LuxDirect's studies suggest this layer compresses as AI training data accumulates around top-tier anchor properties. Independent properties that establish AI signals now are likely to hold a structural advantage as the market consolidates.
Why was Google AI Mode analysed separately from the other platforms?
Google AI Mode operates differently from conversational AI platforms such as ChatGPT, Claude, Gemini, Grok, and Perplexity. For those five platforms, LuxDirect measures conversational recommendation visibility: how frequently a hotel appears in AI-generated responses to discovery, comparison, and booking-intent queries. For Google AI Mode, the relevant measurement is citation behaviour: which domains AI embeds as reference sources, and whether those citations route guests to hotel direct booking pages or online travel agency platforms. These are distinct phenomena requiring distinct methodology. Combining them into a single metric would obscure the difference between how AI recommends hotels and how AI routes bookings.
How LuxDirect Works
LuxDirect sits between your hotel and the AI discovery layer. We align what AI remembers about you with the hotel you actually run.
Every week, we systematically monitor how six leading AI platforms recommend your hotel across high intent guest searches. We show you where AI is diverting guests to OTAs, where competitors are outperforming you, and where your positioning is weak or underrepresented. Then we resolve the structural issues driving it and strengthen your direct booking position within the AI layer, systematically reducing dependency on OTA routing.
You do not need an internal technical team. LuxDirect operates as a visibility concierge. You approve. We execute.
Starting at £199 per month. The average luxury OTA commission runs between 18% and 22%. If LuxDirect recovers just one booking per month from AI mediated OTA routing back to your direct site, the service has paid for itself several times over.
See pricing → Book a demo →