CHILLNN's AIO Lab Conducts Large-Scale Study on Features of AI-Recommended Accommodations
CHILLNN's AIO Lab has conducted Japan's first large-scale survey investigating which accommodations are recommended by AI search. The results revealed that high OTA rankings do not necessarily equate to AI recommendations, emphasizing the importance of presence on social media, reviews, and third-party media.
📋 Article Processing Timeline
- 📰 Published: May 21, 2026 at 02:10
- 🔍 Collected: May 20, 2026 at 17:32
- 🤖 AI Analyzed: May 20, 2026 at 18:08 (36 min after Collected)
Background and Issues
The methods used by travelers to find hotels have evolved with technological advancements, and we are currently at a new turning point: "AI Search." However, there has been no systematic research in the Japanese market investigating which hotels AI recommends. This study represents the first large-scale survey of accommodation facilities in Japan.
Overview of the Initiative
Targeting six cities—Asakusa, Hakone, Kanazawa, Kyoto, Nagoya, and Naha—the study involved inputting 540 queries into three AI models: ChatGPT, Gemini, and Google AI Overview to investigate which hotels were recommended. Key findings are as follows:
- High OTA Ranking ≠ AI Recommendation: Approximately 50% (363 facilities) of hotels within the top 30 on OTAs were never recommended by AI. Conversely, there were 13 facilities that were consistently recommended by AI despite not having high OTA rankings.
- Diverse Online Presence and High-Quality Reviews are Key: Hotels with stable AI recommendations had broad exposure across their own websites, Instagram, major domestic and international OTAs, and editorial media, and had achieved high ratings of 85 or above (converted score) on major OTAs.
- Prominence of 'Third-Party Content' Citations: Editorial media, UGC, and review sites accounted for approximately 36% of the URLs cited by AI, whereas citations of hotel official websites remained at only 0.24%.
Future Outlook
While this study used general-purpose prompts, it is expected that travelers will increasingly use more personalized information and niche prompts when searching for hotels. Following this study, we are proceeding with niche query research that specifies more concrete conditions (e.g., travelers in their 20s, anniversary trips) and aim to clarify the conditions under which unique hotels are more likely to be recommended by AI. We will also continue to issue recommendations for practical measures to enhance the AI visibility of hotels.
The methods used by travelers to find hotels have evolved with technological advancements, and we are currently at a new turning point: "AI Search." However, there has been no systematic research in the Japanese market investigating which hotels AI recommends. This study represents the first large-scale survey of accommodation facilities in Japan.
Overview of the Initiative
Targeting six cities—Asakusa, Hakone, Kanazawa, Kyoto, Nagoya, and Naha—the study involved inputting 540 queries into three AI models: ChatGPT, Gemini, and Google AI Overview to investigate which hotels were recommended. Key findings are as follows:
- High OTA Ranking ≠ AI Recommendation: Approximately 50% (363 facilities) of hotels within the top 30 on OTAs were never recommended by AI. Conversely, there were 13 facilities that were consistently recommended by AI despite not having high OTA rankings.
- Diverse Online Presence and High-Quality Reviews are Key: Hotels with stable AI recommendations had broad exposure across their own websites, Instagram, major domestic and international OTAs, and editorial media, and had achieved high ratings of 85 or above (converted score) on major OTAs.
- Prominence of 'Third-Party Content' Citations: Editorial media, UGC, and review sites accounted for approximately 36% of the URLs cited by AI, whereas citations of hotel official websites remained at only 0.24%.
Future Outlook
While this study used general-purpose prompts, it is expected that travelers will increasingly use more personalized information and niche prompts when searching for hotels. Following this study, we are proceeding with niche query research that specifies more concrete conditions (e.g., travelers in their 20s, anniversary trips) and aim to clarify the conditions under which unique hotels are more likely to be recommended by AI. We will also continue to issue recommendations for practical measures to enhance the AI visibility of hotels.
FAQ
Why are top-ranked OTA hotels sometimes not recommended by AI?
AI evaluates not just OTA rankings but also broader indicators like social media, reviews, and coverage in third-party media, so hotels lacking a diverse digital footprint may not be recommended.
What can hotels do right now to be recommended by AI?
It is recommended to enhance official website content, increase visibility on social platforms like Instagram, and strive for high ratings on review sites and media outlets.
What is the purpose of this survey?
As traveler search behavior shifts toward AI, the survey aims to visualize the criteria for AI hotel recommendations and provide hotels with guidelines for the AI era.