How Tourism Brands Can Appear in AI Travel Recommendations

Ask ChatGPT to recommend a place to visit in Bali. Or a quiet beach town in Portugal. Or a cultural tour in Kyoto. Watch what it cites. It is not pulling from TripAdvisor scores. It is not checking your Instagram follower count. It is not looking at how many Google reviews you have.

It is reading structured data. It is looking for institutional mentions. It is checking whether your destination, hotel, or experience exists as a verifiable entity across independent sources.

This is the gap that will separate tourism brands over the next five years. Not who has the best photos. Not who has the biggest ad budget. Who is machine-readable.

The mismatch: what tourism brands optimize vs. what AI actually reads

I have worked with enough organizations to see the same pattern. The tourism board spends $200,000 on Instagram campaigns. The resort hires influencers. The DMO (Destination Marketing Organization) publishes glossy PDFs that no machine can parse. Then they wonder why ChatGPT recommends their competitor when someone asks "best eco-resort in Southeast Asia."

The answer is almost always the same. Their competitor has better structured data and more institutional mentions. That is it.

Here is the gap, shown clearly.

This is not speculation. Brands leveraging structured data see up to a 25% increase in click-through rates from search, and that is before AI recommendations even enter the picture ^[1]. The tourism brands that show up in AI answers are the ones that made themselves parseable.

How AI travel agents select what to recommend

When you ask Perplexity "best cultural experiences in Yogyakarta" or ChatGPT "quiet beach hotels in Algarve," the system is not doing what Google Search does. It is not ranking pages by backlinks and keyword density. It is constructing an answer by selecting entities it can verify.

The process works roughly like this:

1. Entity recognition. The AI identifies candidate entities from its training data, knowledge bases (Wikidata, Wikipedia), and real-time search results.
2. Attribute extraction. For each candidate, it looks for structured attributes: location, type, amenities, price range, ratings, operating hours. Structured data (JSON-LD, microdata) gets extracted first because it is unambiguous.
3. Source corroboration. The AI cross-references the entity across multiple sources. Does the hotel appear on government tourism listings? Is it mentioned in institutional publications? Does the information on the hotel's website match what Viator and Google Business Profile say?
4. Confidence scoring. Entities with more corroborating structured sources get higher confidence. Entities that exist only on their own website with no external verification get low confidence.
5. Response generation. The AI selects the entities it can most confidently recommend. It cites the sources it trusts most.

Notice what is missing from that process. Follower counts. Ad spend. Influencer mentions. Review volume on a single platform. These things might influence human travelers scrolling social media. They do not influence the machine that is increasingly answering the question before the human ever opens Instagram.

This is the same principle that applies to GEO versus traditional SEO. The inputs are fundamentally different. And tourism is one of the industries where the gap is widest because so much marketing budget goes to channels that AI cannot read.

The chart above is a synthesis of observed AI citation patterns across travel queries. Structured data and institutional mentions dominate. Social media metrics barely register. Paid advertising does not register at all.

Place schema: the declaration AI travel agents need

If you run a tourism business and you have no Place, TouristAttraction, or LodgingBusiness schema on your website, you are invisible to machines. Full stop.

This is not about technical SEO housekeeping. As I wrote in the essay on schema as a strategic decision, structured data is how you declare your identity to machines. For tourism, the relevant schema types are some of the richest in the entire Schema.org vocabulary. They have fields for geographic coordinates, amenities, opening hours, price ranges, contained-in relationships, and more.

Here is what a proper TouristAttraction schema looks like for a real destination:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "TouristAttraction",
  "name": "Taman Sari Water Castle",
  "description": "Historic royal bathing complex built in the 18th century by Sultan Hamengkubuwono I of Yogyakarta. Features pools, meditation chambers, and underground tunnels.",
  "url": "https://example-tourism.com/taman-sari/",
  "image": [
    "https://example-tourism.com/images/taman-sari-pools.webp",
    "https://example-tourism.com/images/taman-sari-aerial.webp"
  ],
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": -7.8100,
    "longitude": 110.3593
  },
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "Jl. Taman, Patehan",
    "addressLocality": "Yogyakarta",
    "addressRegion": "DIY",
    "postalCode": "55133",
    "addressCountry": "ID"
  },
  "containedInPlace": {
    "@type": "City",
    "name": "Yogyakarta",
    "sameAs": "https://www.wikidata.org/wiki/Q5765"
  },
  "touristType": ["Cultural Tourism", "Heritage Tourism"],
  "openingHoursSpecification": {
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Monday","Tuesday","Wednesday","Thursday","Friday","Saturday","Sunday"],
    "opens": "09:00",
    "closes": "15:00"
  },
  "isAccessibleForFree": false,
  "publicAccess": true,
  "amenityFeature": [
    {"@type": "LocationFeatureSpecification", "name": "Guided Tours", "value": true},
    {"@type": "LocationFeatureSpecification", "name": "Parking", "value": true},
    {"@type": "LocationFeatureSpecification", "name": "Wheelchair Access", "value": false}
  ],
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.3",
    "reviewCount": "12480",
    "bestRating": "5"
  },
  "sameAs": [
    "https://www.wikidata.org/wiki/Q3366543",
    "https://en.wikipedia.org/wiki/Taman_Sari_(Yogyakarta)"
  ]
}
</script>

Look at what this declaration does. It tells a machine: this is a tourist attraction. It is located at these exact coordinates, inside Yogyakarta, which is a city the machine already knows about via Wikidata. It is open these hours. It costs money to enter. It has guided tours and parking but no wheelchair access. It has a 4.3 rating across 12,480 reviews. And most importantly, it links to its Wikidata and Wikipedia entries, establishing it as a verified entity in the global knowledge graph.

Compare this to a tourism website that just has a paragraph saying "Visit the beautiful Taman Sari Water Castle, a historic site in Yogyakarta." The prose is nice. The machine cannot do anything reliable with it.

The critical fields most tourism sites miss

I see the same omissions repeatedly:

• geo (coordinates). Without latitude and longitude, the AI cannot reliably place your attraction on a map or recommend it for location-specific queries.
• containedInPlace. This tells the machine your attraction is inside a city, region, or country. It answers "things to do in [location]" queries.
• sameAs (Wikidata/Wikipedia links). This is the bridge between your website and the global knowledge graph. Without it, your schema is a local declaration with no external validation.
• amenityFeature. Travelers ask AI specific questions: "eco-resorts with yoga in Bali," "museums with wheelchair access in Amsterdam." If your amenities are not in structured data, you cannot match these queries.
• openingHoursSpecification. AI agents need to know if a place is open. If you do not declare hours, the AI either guesses or skips you.

Every missing field is a question the AI cannot answer about your destination. And when the AI cannot answer, it recommends a competitor whose data is more complete.

Institutional mentions: the other half of the equation

Schema gets you declared. Institutional mentions get you corroborated. You need both.

As I covered in the essay on brand mentions without links, unlinked mentions in authoritative publications carry enormous weight in AI training data. For tourism, the authoritative sources are specific:

• Government tourism board websites. If your national or regional tourism board lists your destination, that mention carries more AI weight than a thousand blog backlinks.
• UNESCO and heritage organization listings. A UNESCO World Heritage designation is one of the highest-trust signals a tourism entity can have.
• Academic and cultural publications. Mentions in archaeological journals, cultural studies papers, or museum catalogs establish depth of entity.
• Major OTA listings with structured data. Viator, GetYourGuide, and similar platforms use structured data extensively. When your attraction appears there with consistent information, AI models see corroboration ^[2].
• Travel guides from established publishers. Lonely Planet, Rough Guides, Frommer's. These sources are heavily represented in AI training data.

The pattern is the same across every industry I work in. The entity that exists across multiple independent, authoritative sources is the entity that AI recommends. A resort that appears on its own website, the regional tourism board, Wikidata, three OTA platforms, and a Lonely Planet guide is an entity with high confidence. A resort that appears only on its own website and Instagram is a weak signal.

The DMO opportunity

Destination Marketing Organizations are sitting on one of the most valuable assets in AI-era tourism: institutional authority. A DMO website is, by definition, an authoritative source for tourism information about its region.

Most DMOs waste this authority on glossy PDFs and campaign microsites that machines cannot parse. The DMOs that will win are the ones that:

1. Publish structured data listings for every attraction, hotel, and experience in their region.
2. Use TouristAttraction, LodgingBusiness, and Event schema consistently across all listings.
3. Maintain a Wikidata presence for their destination with proper claims and references.
4. Ensure their CRM data flows to their CMS with proper schema markup ^[3].

When a DMO does this, every tourism business in their region benefits. The DMO becomes the authoritative node that connects individual businesses to the destination entity. AI agents trust the DMO listing. That trust flows to the businesses listed there.

Semantic chunking: write for extraction, not for reading

Here is a concept that matters enormously for tourism content. AI-generated summaries now appear for roughly 30% of travel-related queries ^[4]. When the AI builds that summary, it might pull just one paragraph or one sentence from your page.

If that extracted sentence does not make sense on its own, you have lost the opportunity.

This is called semantic chunking. Every section of your content should be independently comprehensible. Not because humans read that way (they don't). But because machines extract that way.

Bad example: "As mentioned above, this area is perfect for the kind of traveler who wants something different." Extracted alone, this sentence is useless. What area? What kind of traveler? Different from what?

Good example: "Nusa Penida's south coast offers 200-meter limestone cliffs, minimal tourist infrastructure, and three dive sites rated among Bali's top ten for manta ray sightings." Extracted alone, this sentence is a complete, citable recommendation.

Every page on your tourism website should be reviewed with this question: if the AI pulls just this one paragraph, does it stand alone as a useful answer?

The audit: five things to check this week

If you run a tourism business, hotel, tour operator, or DMO, here are the five things you should check immediately.

1. Run your URL through Google's Rich Results Test. If no TouristAttraction, LodgingBusiness, Hotel, or Place schema appears, you have zero structured declarations. Fix this first.
2. Search your destination name on Wikidata. If no entry exists, create one. If an entry exists but has sparse claims, add references. This is your node in the global knowledge graph.
3. Check your national/regional tourism board website. Is your business listed? Is the listing current? Is the information consistent with your own website? If not, contact them. This is free institutional authority.
4. Ask ChatGPT and Perplexity about your destination. Do they mention you? Do they get the facts right? If they do not mention you, you have an entity visibility problem. If they mention you but get facts wrong, you have an entity consistency problem.
5. Compare your schema to your competitors. Pick the top three destinations or hotels that appear in AI recommendations for your target queries. Check their structured data. I guarantee they have richer schema than you do.

These five checks will tell you exactly where you stand. Most tourism brands are at step zero: no structured data, no knowledge base presence, and no idea that this matters.

What about reviews and social proof?

Reviews still matter. Just not the way most people think.

A high TripAdvisor score does not directly influence AI recommendations. But aggregateRating data embedded in your schema does. When your JSON-LD declares a 4.5 rating across 8,000 reviews, that structured rating becomes an attribute the AI can use in its confidence scoring.

The distinction is where the data lives. A rating displayed as text on a third-party review platform is just text the AI might read. A rating declared in your structured data is a machine-readable fact the AI can compare and rank.

Social proof on Instagram? Essentially invisible to AI recommendation engines. AI training data includes Wikipedia, published books, academic papers, government websites, and structured web data. It does not include your Instagram story highlights.

This does not mean you should stop doing social media marketing. People still use Instagram to plan trips. But you should stop confusing social media presence with AI visibility. They are completely different channels with completely different mechanics.

The $200,000 reallocation

Here is the uncomfortable math. A typical mid-size tourism brand spends roughly $200,000 per year on digital marketing. The breakdown usually looks like this: 40% social media ads, 25% influencer partnerships, 20% Google Ads, 10% content creation, 5% "technical SEO."

That 5% technical SEO line item is where schema implementation gets buried. If it gets done at all.

I am not suggesting you move the entire budget to structured data. That would be absurd. But I am suggesting that the organizations spending $0 on entity infrastructure and $80,000 on Instagram ads are making a strategic error that will compound over the next three to five years.

As AI search grows (and it is growing fast, with travel being one of the highest-adoption categories), the tourism brands that invested in machine-readability will get recommended. The ones that invested only in human-visible marketing will not.

A reasonable reallocation: move 15-20% of your digital budget toward structured data implementation, knowledge base presence, and institutional mention strategy. That is not a revolution. That is insurance against the single biggest shift in how travelers discover destinations since the invention of Google.