How AI Models Choose Which Destinations to Recommend: The Source Trust Hierarchy Every DMO Needs to Understand

Your Destination's Biggest Competitor Isn't Another City. It's Invisibility.

Here's the new reality: Over 40% of travelers globally now use AI-based tools for travel planning and booking. Among Millennials and Gen Z, that number climbs to 60%. These travelers aren't Googling "best places to visit in Colorado" anymore. They're asking ChatGPT, Gemini, and Perplexity to plan their trips.

And here's what most DMOs don't realize: each AI model has a completely different hierarchy of sources it trusts. ChatGPT doesn't read the web the same way Gemini does. Perplexity has different favorites than Grok. If your destination isn't showing up in the right sources for the right models, you're invisible to a growing share of your potential visitors.

A landmark January 2026 study from HotelRank analyzed 245,046 unique sources cited by four major AI models across 19,579 runs. The findings reveal exactly how AI decides what to recommend — and it's not what most destination marketers expect.

The New Discovery Funnel: AI Models Are the Gatekeepers

Traditional destination marketing assumes a funnel: awareness (social, ads, PR) → consideration (website, content) → booking (OTAs, direct). AI is collapsing this funnel into a single interaction. A traveler asks "Where should I go for a long weekend in the Pacific Northwest?" and the AI delivers a curated answer — no funnel, no multiple touchpoints, no chance to retarget.

The question every DMO needs to answer: When an AI model builds that answer, does it pull from sources that mention your destination?

This isn't hypothetical. The AI tourism market was valued at $3.37 billion in 2024 and is projected to reach $13.87 billion by 2030 (Grand View Research). The shift is happening now, and DMOs that don't understand how AI sources its recommendations will lose ground to destinations that do.

How Deep Do AI Models Actually Search?

Not all AI models are created equal when it comes to research depth. The HotelRank study measured how many unique URLs and domains each model scans per query:

• Grok: 58.5 unique URLs per run, 12.4 domains — the deepest researcher

• GPT 5.2: 27.3 URLs, 16.2 domains — doubled its search depth from GPT 5.1

• GPT 5.1: 11.8 URLs, 7.8 domains

• Gemini 2.5 Flash: 11.2 URLs, 8.8 domains

• Perplexity Sonar: 8.2 URLs, 8.0 domains — the shallowest searcher

The takeaway for DMOs: consistency across sources matters more than ever. GPT 5.2 now scans 27 sources per query. If your destination is mentioned on only 2-3 sites, you'll lose to competitors appearing on 10+. The days of relying on your DMO website alone are over.

The Source Trust Hierarchy: What Each AI Model Reads

This is the most actionable data in the entire study. Each AI model has a distinct set of sources it consults most frequently. Here's the percentage of runs where each domain was cited:

Grok's Sources

• TripAdvisor: 99.9% of runs

• Expedia: 96.4%

• Hotels.com: 81.7%

• Booking.com: 76.4%

• Facebook: 63.5%

• Reddit: 54.5%

Grok is the social proof engine. It heavily indexes TripAdvisor, Facebook groups, and Reddit. If your destination has a weak presence on these platforms, Grok won't recommend you.

Gemini's Sources

• Booking.com: 63%

• Expedia: 37.3%

• Agoda: 33.5%

• YouTube: 13.6%

• Marriott.com: 11.2%

Gemini is Google's model, and it loves Google's ecosystem. YouTube is a top-10 source for Gemini (13.6%) but barely registers for other models. DMOs investing in YouTube content get a Gemini advantage. Booking.com dominates at 63%.

ChatGPT (GPT 5.2) Sources

• Booking.com: 53.9%

• Hotels.com: 31.9%

• Marriott.com: 30.6%

• Wikipedia: 30%

• Expedia: 28.9%

• TripAdvisor: 20.5%

• The Times (UK): 19.5%

ChatGPT trusts editorial and reference sources. Wikipedia appears in 30% of GPT 5.2 runs (75% for GPT 5.1). Major news outlets like The Times rank in the top 10. This means your destination's Wikipedia page and earned media coverage directly influence ChatGPT recommendations.

Perplexity's Sources

• TripAdvisor: 95.5%

• Expedia: 68.6%

• Hotels.com: 36.8%

• Booking.com: 33.3%

• ForbesTravelGuide: 11.8%

Perplexity is the OTA-heavy model. It also has a partnership with TripAdvisor (95.5% citation rate). If your destination's hotels and attractions aren't well-represented on TripAdvisor and Expedia, Perplexity won't find you.

User-Generated Content: The Hidden Influence Layer

One of the study's most surprising findings: user-generated content platforms are major AI sources, and each model has its favorite:

• Gemini favors YouTube (14% of runs)

• GPT 5.1 relies on Reddit (15% of runs)

• Grok heavily indexes Facebook groups and Reddit (63.5% and 54.5%)

This upends traditional DMO strategy. You can spend your entire budget on your own website and social channels, but if travelers and creators aren't talking about your destination on Reddit, YouTube, and TripAdvisor, the AI models that increasingly drive discovery won't have the signal they need to recommend you.

The implication: DMOs need a UGC strategy that's optimized not just for human eyeballs, but for AI ingestion. Encourage creators to make YouTube content about your destination (Gemini will find it). Engage in Reddit travel communities (ChatGPT will surface it). Ensure your TripAdvisor listings are complete and actively reviewed (Perplexity and Grok depend on it).

Where Does AI Send Travelers? Direct vs. OTA Traffic

There's good news for destinations and properties focused on direct relationships: 75-91% of hotel links from AI models go directly to hotel websites, not OTAs. This is a massive shift from traditional search, where OTAs often dominate the top results.

• GPT 5.2: 91.1% direct, 8.9% OTA

• Gemini 2.5 Flash: 89.4% direct, 10.6% OTA

• GPT 5.1: 87.9% direct, 12.1% OTA

• Perplexity Sonar: 74.7% direct, 25.3% OTA + TripAdvisor

For DMOs, this means AI is actually better for your local businesses than Google Search was. When a traveler Googles "hotels in Bend Oregon," they see a wall of Booking.com, Expedia, and Hotels.com ads. When they ask ChatGPT the same question, they get direct links to the actual properties 91% of the time.

But here's the catch: AI can only link to your properties if it knows they exist. And it knows they exist based on the sources it trusts. Which brings us back to the source hierarchy.

Geographic Bias in AI Recommendations

The study revealed significant geographic patterns in how AI models recommend:

• North American cities see more chain hotel recommendations (Toronto 47%, New York 45%, LA 43%)

• European and Japanese cities favor independents (Tokyo 75%, London 69%, Paris 69%)

• Emerging destinations see highest OTA rates (Cairo 42%, Shanghai 44%)

This geographic bias matters for DMOs. If you're marketing a destination with many independent properties (which most DMOs are), your businesses may already have an AI advantage in European-style markets. But in North American markets, chain properties get disproportionate AI visibility simply because they have better-structured data across more sources.

The solution: help your local businesses get listed and well-reviewed across the platforms that AI trusts most. This is a concrete, actionable role for DMOs in the AI era.

The Traveler Persona Factor

AI doesn't recommend the same way for every traveler. The study found that traveler persona significantly changes what AI recommends:

• Business travelers: 45% chain recommendations, 10% OTA

• Luxury travelers: 32% chain, only 9% OTA — highest direct rate

• Families: 30% chain, 19% OTA

• Couples: 26% chain, 13% OTA — highest independent rate at 61%

For DMOs targeting couples and luxury travelers, independent properties have a strong AI advantage. For destinations focused on business or group travel, the AI landscape skews toward chains. Understanding these dynamics helps you tailor your AI visibility strategy to the travelers you actually want to attract.

A Practical AI Visibility Checklist for DMOs

Based on this research, here's what destination marketers should prioritize:

For ChatGPT Visibility

• Update and expand your destination's Wikipedia page with current, well-cited information

• Pursue earned media in major publications (The Times, Condé Nast Traveler, etc.)

• Ensure properties are listed on Booking.com with complete profiles

• Encourage Reddit discussions about your destination in travel subreddits

For Gemini Visibility

• Invest in YouTube content — destination guides, creator partnerships, video tours

• Optimize Booking.com listings (Gemini's #1 source at 63%)

• Maintain strong Google Business profiles for all local businesses

For Perplexity Visibility

• Prioritize TripAdvisor presence — it appears in 95.5% of Perplexity runs

• Ensure Expedia listings are complete (68.6% citation rate)

• Seek Forbes Travel Guide coverage

For Grok Visibility

• Activate on TripAdvisor (99.9% of Grok runs cite it)

• Build presence in Facebook travel groups

• Engage in Reddit travel communities

• Ensure Expedia and Hotels.com coverage

Why DMOs Need AI Visibility Monitoring

Here's the fundamental problem: you can't optimize what you can't measure. Most DMOs have Google Analytics. Many have social listening tools. Almost none are monitoring how AI models talk about their destination.

The Sojern State of Destination Marketing 2026 report found that 64% of DMOs are now writing content specifically formatted for AI engines — answering specific questions, publishing lists, updating listings, and producing articles aligned with real traveler queries. But writing for AI without monitoring whether AI actually picks up your content is like running ads without tracking conversions.

AI visibility monitoring means systematically tracking:

• How often each AI model recommends your destination

• What sources the AI is pulling from when mentioning you

• How your destination is positioned relative to competitors

• What sentiment and attributes AI associates with your brand

• Whether your content optimization efforts are actually moving the needle

This is an entirely new category of intelligence that didn't exist two years ago. But given that 40% of travelers are now using AI for trip planning, it's rapidly becoming as essential as SEO monitoring was in 2010.

The Bottom Line: AI Visibility Is the New SEO

The research is clear. AI models are becoming major gatekeepers of travel discovery, and they each have distinct, measurable source preferences. DMOs that understand and optimize for these preferences will capture a growing share of AI-referred travelers. Those that don't will watch their destinations fade from the AI-mediated conversation.

The action plan is straightforward:

• Audit your presence across the top sources for each major AI model

• Invest in UGC platforms (Reddit, YouTube, TripAdvisor, Facebook) — not just your own channels

• Help local businesses maintain complete, well-reviewed profiles on the platforms AI trusts

• Monitor your AI visibility across models — not just your Google rankings

• Structure your content for AI consumption: clear answers, factual data, structured formats

The destinations that win in 2026 and beyond won't just be the ones with the best Instagram campaigns. They'll be the ones that show up when AI answers the question: "Where should I go?"