In this article, we analyze three of the most common misconceptions surrounding GEO and large language models (LLMs), which today can directly impact a property’s ability to capture direct demand and remain competitive in the new digital landscape.
“If my hotel has great reviews, it will always be recommended by LLMs”
Large language models do not generate their responses based on a single signal, such as reviews, but rather on a combination of sources and information available online. Reviews are only one part of the broader picture.
This means that even a hotel with excellent reviews may fail to appear in generated responses if the available information is fragmented, inconsistent, or difficult for the models to interpret.
This is where a very concrete and often underestimated issue emerges: many hotels have built their reputation over time primarily on OTAs and intermediary platforms (Booking, Expedia, TripAdvisor), while their official websites often remain less comprehensive, outdated, or lacking in structured content.
The result is counterintuitive but critical: if the richest, most consistent, and most “AI-readable” source is not the hotel’s official website but the OTA listing, then OTAs themselves will be favored in the responses generated by LLMs.
In other words, the hotel’s reputation does not disappear - but it ends up strengthening the visibility of the platforms hosting it. And dependency on intermediaries, rather than decreasing, is replicated within the AI ecosystem itself.
“Traditional SEO is enough to remain visible in AI systems”
A common mistake is to assume that traditional SEO is also sufficient for visibility within systems powered by generative artificial intelligence. In reality, these are two environments that only partially share the same operating logic.
To understand the difference, it is useful to start from how large language models work. LLMs are based on machine learning systems and neural networks trained on enormous amounts of data. Their behavior is not deterministic: they do not apply fixed rules, but generate responses based on statistical probabilities. In other words, they do not simply retrieve information like a search engine; they construct the most plausible answer according to the context and the user’s query.
This fundamentally changes the way content is interpreted.
Even the structure of a website takes on a different role. A website is built using HTML, CSS, and JavaScript: HTML defines the informational structure, CSS handles the visual appearance, and JavaScript manages interactions. However, in most cases, LLMs do not “browse” a website the way a human user would. Instead, they extract and process textual content or structured data.
As a result, what matters is not only how a page looks, but above all how clear, coherent, and semantically understandable the information is.
SEO was originally designed to make content readable for traditional search engines, meaning systems that generate a SERP (Search Engine Results Page) - the page of results we see on Google after a search. It is a discipline based on standardized elements such as semantic structure, meta tags, and structured data that enrich information (prices, reviews, property type). It is an effective system, but one designed for a stable and deterministic ranking model.
GEO, on the other hand, introduces a different logic. In generative systems, there is no fixed SERP identical for every user: responses are dynamically constructed each time by drawing from multiple sources and synthesizing them. In this process, factors such as content clarity, informational density, consistency across sources, and the ability to respond directly to a specific search intent become critical.
For this reason, strong SEO rankings do not automatically guarantee visibility in AI systems. A piece of content may be perfectly indexed by traditional search engines while still not being sufficiently structured or “interpretable” to be used as a source in AI-generated responses.
The implication is clear: SEO remains a necessary foundation, but it is no longer sufficient on its own to ensure visibility within new AI-based search systems.
The good news is that this does not mean websites need to be rebuilt from scratch. AI assistants do not primarily rely on the visual interface of a website, but rather on the informational and semantic structure of its content and data. For this reason, it is not the “showcase” layer aimed at human users that needs to be radically transformed, but the informational layer interpreted by machines. The website therefore continues to fulfill its commercial and conversion role for human visitors, while at the same time becoming understandable and usable for AI agents as well.
“Only young people use AI-based systems”
Another widespread belief is that artificial intelligence is mainly used by younger generations, particularly Gen Z and Millennials. However, recent data does not fully support this assumption.
According to the 2025 Hospitality and Travel Report by Adyen, the use of artificial intelligence in travel decision-making is now widespread across generations and, in some cases, shows trends that are the opposite of what many would expect.
The report highlights that around 50% of Gen X and Baby Boomer users already use AI tools to decide where to stay. Not only that: this demographic also represents the fastest-growing segment in AI adoption, with a +98% increase in 2025.
This significantly changes the perspective on AI adoption within the travel industry. Usage is no longer limited to younger or digitally native audiences, but is rapidly expanding among older demographics as well, particularly among users seeking to simplify decision-making and reduce the time required to choose.
In this sense, artificial intelligence is no longer a niche or generational behavior, but rather a new and increasingly universal way of accessing information - one that now involves users with very different profiles, needs, and levels of technological familiarity.
The implication is clear: AI adoption in travel does not follow a generational logic, but one driven by utility and immediacy. And precisely for this reason, it is growing fastest among demographics traditionally considered less digital.
As a result, one of the most deeply rooted assumptions in the hospitality industry also falls apart: the idea that “my customer is older, therefore AI is irrelevant to my business.” In reality, the typical guest profile for independent properties - boutique hotels, agriturismos, and upper mid-range B&Bs - often falls precisely within the 45–65 age group, which is now among the most active in adopting these tools.
Ignoring this shift therefore does not mean falling behind a future trend, but rather falling behind a behavioral change that is already happening within real consumer demand.
In conclusion
GEO and LLMs are not replacing existing channels in the hospitality industry, but they are redefining how information is collected, filtered, and synthesized during the early stages of the decision-making process. The misconceptions analyzed here show how oversimplified interpretations can lead to incomplete - and potentially harmful - assessments of a hotel’s actual digital visibility.
However, there is one element that goes beyond visibility itself and concerns the balance of the entire ecosystem: the role of OTAs within AI-generated responses. Booking.com and Expedia are already investing strategically to strengthen their presence in generative systems through highly structured content, established authority, and digital assets that individual properties often struggle to replicate.
The result is a potentially counterintuitive scenario: greater exposure within AI systems does not automatically translate into more direct bookings, but may instead further reinforce the role of intermediaries, which risk becoming the natural destination even within these new discovery channels.
For this reason, the challenge is no longer simply about being present within new search systems, but about actively controlling how a property is interpreted, selected, and “translated” by AI. Without an adequate technological strategy, this new flow of demand may once again be captured by intermediaries.
At the same time, however, a structurally different opportunity is emerging for the first time: the hotel’s official website can potentially return to the center of the decision-making process, because it represents the direct source of the service and, as such, can be recognized by AI systems as a primary and highly reliable source of information.
In this context, competition is no longer based solely on visibility, but on the authority with which a property is able to position itself as the original source of its own offering within the new generation of search systems.