Five London Hotels Capture 57% of AI Recommendations: What a New Study Reveals About AI Travel Distribution

A new form of travel distribution is shaping where luxury hotel guests go before they ever reach a booking page. AI platforms such as ChatGPT, Gemini, Perplexity, Claude, Grok and Google AI Mode now answer hotel recommendation queries for millions of travellers. For hoteliers, the question is no longer whether this channel exists. It is whether their property appears in it and where the guest is routed when it does.

LuxDirect recently completed its CS7 London study: 2,700 structured queries run across 25 luxury hotels on six AI platforms. LuxDirect has now analysed more than 12,500 AI travel queries across five UK luxury markets. The London findings challenge several assumptions about how hotel reputation, size, and brand recognition translate into AI visibility.

A Concentration Dynamic Developing Faster Than SEO

The headline finding from the study is one that should concern any independent luxury hotel not actively monitoring AI platforms: five hotels captured 57% of all AI recommendations across the 25-hotel panel. The remaining 20 hotels collectively accounted for less than half of all mentions, regardless of their real-world quality, reputation, or marketing spend.

This concentration pattern will be familiar to anyone who remembers the early years of search engine optimisation. In the mid-2000s, Google's ranking algorithm began consolidating organic traffic around a small number of well-optimised properties. Hotels that moved early captured disproportionate visibility. Those that delayed found themselves effectively invisible.

The AI concentration dynamic appears to be following a similar curve, but with two important differences. First, the consolidation is happening faster. AI platforms are already producing highly consistent recommendation patterns across queries, with a small cluster of properties receiving repeated endorsement while others are functionally absent. Second, the signals driving AI recommendations are less transparent than Google's ranking criteria were, even in their early form. Hotels cannot easily audit why they are or are not appearing.

Size and Brand Recognition Are Not the Determining Factors

One of the more striking individual findings from the CS7 study concerns the relationship between hotel size and AI visibility. Across the 25-hotel panel, a 26-room independent property appeared more frequently in AI recommendations than a 174-room chain property in the same city and the same quality tier.

This is counterintuitive by conventional distribution logic. Larger properties typically benefit from greater brand recognition, more online content, higher review volume, and larger marketing budgets. All of these factors have historically correlated with better visibility across digital channels. In AI recommendations, the correlation breaks down.

The implication is that AI platforms are drawing on a different set of signals to form their recommendations. Structured data quality, editorial citation patterns, how a hotel is described across authoritative third-party sources, and the consistency of information across the web all appear to carry more weight than raw brand or marketing scale. For independent hotels with strong editorial coverage and well-structured digital presences, this creates a genuine first-mover opportunity in AI-driven discovery.

Google AI Mode and the OTA Routing Problem

The third major finding from the CS7 study concerns not which hotels appear in AI recommendations, but where those recommendations route the guest. Across all Google AI Mode responses analysed in the study, 65.1% contained links routing travellers to OTA booking pages rather than the hotel's own website.

This is a materially different problem from the concentration finding. A hotel can appear consistently in AI recommendations and still lose the direct booking if the citation pathway routes through Booking.com or Expedia. The hotel earns the recommendation. The OTA captures the booking.

For hotels that have spent years investing in direct booking strategies, this finding is significant. Rate parity work, loyalty programme development, and direct channel marketing all operate downstream of the discovery moment. If the guest's first encounter with a hotel recommendation routes them to an OTA, the direct booking infrastructure rarely gets the opportunity to compete.

The Google AI Mode routing pattern appears to be driven by the sources Google's AI draws on when constructing responses. OTA pages tend to be highly optimised for structured data, frequently updated, and heavily cross-referenced across the web. Hotel direct websites, particularly for independent properties, often lack the same structural consistency. The result is that AI platforms default to OTA citations when they need a bookable link.

What This Means for Independent Luxury Hotels

The CS7 London findings highlight three priorities for independent luxury hotels engaging with AI as a distribution channel.

The first is measurement. Hotels cannot manage a channel they cannot see. Understanding current AI visibility across the major platforms, including where a property sits relative to comparable hotels in its market, is the necessary starting point. Several properties in the CS7 panel scored in the bottom tier of our visibility index, indicating near-total invisibility in AI-assisted search, without any apparent awareness of the gap.

The second is structured data quality. The signals AI platforms use to form recommendations are heavily influenced by how consistently and accurately a hotel is represented across its digital presence. Schema markup, structured content, and editorial accuracy on third-party platforms all contribute to the signals AI platforms read.

The third is citation pathway management. Addressing OTA routing in AI responses requires active work on direct website authority and structured data, not simply having a direct booking engine. The goal is to give AI platforms a credible, well-structured direct URL to cite rather than defaulting to an OTA.

The window for early action on AI visibility is open. The CS7 study represents one of the first structured, multi-platform analyses of AI recommendations across a defined luxury hotel market. The concentration pattern is already established, but it is not yet fixed. Independent properties that understand and address their AI signals in the next twelve months will be better positioned than those that engage with this channel reactively.

Methodology

The CS7 London study was conducted by LuxDirect across 25 luxury hotels in London on six AI platforms: ChatGPT, Claude, Gemini, Grok, Perplexity, and Google AI Mode. 2,700 structured queries were run using a standardised prompt library across discovery, evaluation, and booking intent query types. LuxDirect is an AI visibility intelligence platform focused on independent luxury hotels. luxdirect.ai