How ALLMO.ai Powered Nokumo’s AI Visibility Study in Travel & Hospitality
Nokumo partnered with ALLMO.ai to analyze 3,600 AI responses across four leading large language models. The clearest signal in the data: one domain, Booking.com, appears in 95.3% of every query tested and captures one in seven of all citations. For every independent-hotel URL an AI recommends, it cites roughly two OTA URLs. That number did not come from a survey or a vendor claim. It came from running 450 real hospitality prompts, in three languages, across five European markets, through ALLMO’s automated research framework.
For independent hotels, vacation rentals, and apartment owners trying to reduce OTA dependency, the real lesson is not the scale of the gap. It is that despite t.
Why Nokumo Ran the Study
Nokumo ran this study because the question had no vertical-specific answer at scale. When a traveler asks an AI engine to recommend accommodation in a specific location, how often does an independent property appear versus an OTA listing? And what drives who gets cited?
Anecdotal tests of one or two models in one language tell you almost nothing. Nokumo needed enough responses, languages, and markets to see which signals actually determine AI citation frequency in hospitality. So they approached ALLMO.ai to build the dataset.
How ALLMO contributed the critical AI Search Data part of the study
ALLMO provided the prompt research framework and the prompt monitoring that powered the entire study.
Together with Nokumo, we constructed 450 natural language prompts on hospitality and travel search, the kind real travelers type when planning a trip, not the kind a marketer writes for a keyword tool. The prompts covered destionations in five countries. Each prompt was translated into German, English, and Croatian, split across four leading AI models: ChatGPT, Gemini, Mistral, and Perplexity, across five countries.
The result was 3,600 distinct AI responses.
ALLMO’s Citation Intelligence layer then extracted every brand mention and every URL that appeared across those responses. The totals:
- 13,859 specific brand mentions identified.
- 20,370 URLs cited for grounding extracted.
Rather than leaving Nokumo to clean thousands of raw conversational responses by hand, ALLMO delivered a structured CSV export, expert commentary, and direct access to the analytical dashboard. That automation turned unstructured AI output into a clear map of market visibility.
In addition, the AI Page Indexing audit was run for 500+ hotel websites to verify if they are visible.
The Findings ALLMO’s Data Revealed
The results overturned several assumptions in the hospitality space.
The most jarring finding was the invisibility of direct accommodation websites. Only 5.7% of the independent properties tested were detected by any AI model. The other 94.3% are effectively invisible to generative search. When an AI model looks for lodging, it leans on established aggregation nodes rather than visiting individual property pages.
That gap is filled by Online Travel Agencies. Booking.com appeared in 95.3% of all queries tested and captured 1 in every 7 URLs cited across all 20,370 grounding URLs. Appearing in 95.3% of queries is not market leadership. It is near-total capture. Every independent property that appears is competing for the fraction of responses where Booking.com is absent.
Language changed the outcome too. Croatian-language queries produced up to 10 percentage points more OTA dependency than the same queries run in German, even when searching for identical Austrian hotels. LLMs interpret trust and data availability differently depending on the linguistic context of the user. For operators in Croatian-primary markets, that is a structural disadvantage no schema markup will fix.
| Metric Evaluated | Independent Hotel Sites | Booking.com / OTAs |
|---|---|---|
| Overall Query Presence | 5.7% | 95.3% |
| AI Grounding URL Share | Very Low | 14.3% (1 in 7 URLs) |
| Booking Engine AI-Visibility | 22.9% Visible | 100% Visible |
| Impact of Clean Architecture | High Priority | Pre-Optimized |
The full report covers the complete dataset, country-specific visibility breakdowns, an intent-by-provider matrix, a self-assessment scorecard, a 90-day action plan, and commentary from four industry specialists. Nokumo has made it publicly available [here](The full results are available at https://www.nokumo.net/reports/ai-visibility-2026).
Business Impact for Nokumo
Nokumo turned the research into a commercial asset.
They presented the results on the main stage at VRM Days Hamburg, a conference for vacation rental professionals, and the study generated earned media across English and German travel trade press. Today Nokumo uses the report as a core engine for customer acquisition, showing prospective hotel clients exactly why they are losing organic traction and how they can fix it.
Takeaways for Accommodation Marketers & Hotel Owners
Most of these require no budget to test.
- Track citation share, not just rankings. Booking.com captures 1 in 7 of all AI-cited URLs.
- Check whether your hotel is mentioned when users ask AI models relevant questions. Test it yourself with a few prompts on ChatGPT, or use a GEO tool like ALLMO to monitor prompts at scale.
- Check whether your booking engine is crawlable by AI.
- Prioritize URL structure and web trust signals over schema. The impact differential is 2 to 3x.
- Test your property name in at least two languages. Language shifts OTA dependency by up to 10 points.
To see your own AI visibility baseline, start for free on ALLMO.ai or book a demo. For a hospitality-specific solutions including PMS, Channel Managers and an AI-Discoverable Direct Booking Platform, speak to our friends at Nokumo.
Frequently Asked Questions
Does this research apply to my country if it is not in the five studied?
The Nokumo study's specific percentages, including the 94.3% invisibility rate, are grounded in a five-country, three-language dataset. Operators in other markets should treat the numbers as directional benchmarks rather than confirmed local facts. The methodology is replicable: run hospitality prompts across AI models in your language and measure brand mentions and cited URLs.
Does schema markup still matter at all for AI search visibility?
Schema markup helps AI models parse and understand a property's room types, amenities, and location, but the Nokumo data shows it is 2-3x less impactful than clean URL paths and reliable web trust signals when it comes to citation frequency. Schema aids comprehension; URL structure and authority drive how often a property is actually cited in AI responses.
How exactly was AI visibility measured in the Nokumo study?
ALLMO's Citation Intelligence framework classified a property as visible when its brand name or direct URL appeared in an AI response to a relevant hospitality prompt. Brand mentions and URL citations are tracked as distinct measurements. The 94.3% invisibility rate means properties appeared in neither form across the 3,600 responses analyzed across ChatGPT, Gemini, Mistral, and Perplexity.
What can an independent hotel do right now to improve AI search visibility?
Run your property name through ChatGPT, Gemini, Mistral, and Perplexity using a prompt a traveler would actually type. Do it in at least two languages. The Nokumo data shows language choice shifts OTA dependency by up to 10 percentage points. Audit whether your booking engine is crawlable. 77.1% of accommodation websites have none visible to AI crawlers. Prioritize URL structure. The impact is 2-3x greater than schema markup.
Track your AI search visibility
See where your brand appears in ChatGPT, Perplexity, and other AI search engines.