Major visitor attraction · GEO research
Finding AI-search opportunity when the website could not be crawled
An early-stage GEO investigation that exposed a fundamental technical barrier and showed how third-party sources were already shaping recommendations.
The engagement in context
A major visitor attraction, supported through an agency partner. The investigation connected AI-search recommendations with technical accessibility and evidence beyond the organisation’s own website.
A famous attraction was missing from important recommendations
A major European visitor attraction wanted to understand why it was not being recommended consistently beside museums, cultural landmarks and family activities in its city. This was early in the development of GEO, before reliable measurement conventions or mature tools were widely available.
We needed to build the research method while using it: define useful prompts, capture outputs consistently and compare the attraction with the alternatives people were actually being shown.
The first task was deciding what a useful answer looked like
The prompt set covered discovery, comparison and itinerary planning rather than repeatedly asking for the attraction by name. That exposed the situations in which it entered the consideration set, the alternatives that displaced it and the attributes AI systems used when explaining their recommendations.
At the time, dedicated GEO platforms were still immature. The method combined controlled manual testing, recorded answers, competitor comparison and source review. It was deliberately simple enough to repeat and explicit enough for another analyst to understand how each conclusion had been reached.
The first breakthrough came from ordinary technical SEO
The investigation found that the attraction’s age gate prevented Googlebot from reaching the main site. Thin generated pages, weak structured data and slow performance added further friction. Before discussing sophisticated AI optimisation, the owned website needed to become available as evidence.
That diagnosis reframed the project. AI-search visibility was not a separate layer to bolt on later; it depended on crawlability, accessible information and a site that clearly described the entity and the experience.
The technical findings created an immediate order of operations
There was little value in debating subtle prompt optimisation while the primary website was unavailable to a major search crawler. The first recommendation was to make the core experience crawlable and indexable, then reduce low-value generated pages, clarify the site structure and add structured information that accurately matched visible content.
This sequencing helped separate eligibility from influence. Technical work could make the owned site available as evidence. Content and authority work could then improve what the wider web understood and repeated about the attraction.
The brand was still visible without help from its own website
Despite the technical problems, the attraction did appear in some AI answers. Reviews, travel platforms, community discussions, Wikipedia and Wikidata provided enough external evidence for systems to understand and occasionally recommend it.
This became the most useful strategic lesson: an organisation can earn some generative visibility without a functioning owned site, but it cannot control accuracy, depth or conversion nearly as well. GEO therefore had to consider the complete evidence environment, not just on-page copy.
A practical programme replaced a vague ambition
The final recommendations joined technical remediation, entity clarity, structured information, useful content and third-party authority into one roadmap. The client left with a measurable baseline and a clear explanation of why conventional search fundamentals still mattered in AI discovery.
The work also gave the wider team a way to evaluate later changes. They could revisit the same journeys, compare which sources appeared and identify whether the attraction was entering more of the relevant recommendation sets for the right reasons.
The learning changed how I think about GEO
Owned websites are important, but they are only one part of the evidence graph. Reviews, reference sources, local listings, publications and community discussion can establish an entity even when its own site is weak. That explains both the resilience and the lack of control we observed.
A good GEO investigation therefore starts with the answer and works backwards through the evidence. It checks technical eligibility, but it also asks who else is describing the organisation, which claims are repeated and where a person could verify them.
