Why do different AI platforms recommend such different travel brands for the same trip?
👉 Recommend this DiscussionRecommend this?
★
★
★
★
★
Reactions
Why do different AI platforms recommend such different travel brands for the same trip?
Why do different AI platforms recommend such different travel brands for the same trip?
was posted by Jadyn Worrell
in
Discussion
,
Travel Tech
,
AI
,
Marketing
,
Caribbean
.
Featured on Oct 5, 2026 (1 month ago).
This post is not rated yet.
Sharing is caring
QR Code
Comments
I’ve been looking into how AI-powered search is affecting travel discovery and noticed something interesting.
I ran the same three Barbados/Caribbean travel searches across ChatGPT, Gemini and Perplexity and I got 9 responses in total. Across those responses, 29 different travel brands were surfaced, and no single company appeared in all 9.
What stood out to me was how much the brands changed depending on both the platform and the type of question being asked. Even established travel companies and Caribbean specialists appeared in some searches and disappeared completely in others.
It made me wonder how travel businesses should actually think about AI visibility when there doesn’t seem to be one consistent “AI search result.”
For anyone working in travel marketing, distribution or AI, are you seeing similar differences between ChatGPT, Gemini and Perplexity when researching destinations or travel companies?
I’ve noticed something similar from the destination side. The results can change quite a bit depending on how the question is phrased, not just which AI platform is used. It makes me think travel brands need to focus less on “ranking” in one AI tool and more on having clear, consistent and useful information across different sources.
That’s exactly the part I found most interesting. Changing the intent behind the question sometimes changed the brands being surfaced just as much as changing the platform.
Your point from the destination side is interesting too. Have you noticed certain types of destination information or sources being picked up more consistently than others?
I’ve noticed practical information tends to come up more often, especially transport, opening hours and the better-known places. Smaller local spots can be very different depending on how you ask the question. That’s actually what made me interested in this topic.
29 different brands across just 9 responses is quite a spread! What I'd be careful about is treating AI visibility as a static ranking. For travel brands, I'd rather track a consistent set of high-intent searches over time — specific destinations, dates, budgets, and travel styles — and see which recommendations actually match the traveler's needs. Getting mentioned is one thing; being recommended to the right customer is another.
Alex, I think that's the key distinction. A brand appearing in an AI response doesn't necessarily mean it's being recommended to the right traveller.
The 29 brands across nine responses were an exploratory observation, not a ranking or evidence of commercial performance.
A more useful assessment would track a consistent set of high-intent searches over time, then evaluate whether the recommended brands actually fit the traveller's destination, budget and travel requirements.
That distinction between visibility and recommendation relevance is something we're looking to incorporate into our assessment framework. Appreciate the perspective.
Jadyn, your distinction between platform and question is useful. I would add a repeated run of the same exact question on each platform before attributing a difference to either one. In a small archived New York API example I examined, the shortlist also changed between two runs of the same provider. That cannot tell us which factor caused the change, but it makes the execution date, interface and complete outputs worth preserving. Did your nine responses include repeats, or was each platform/question combination run once? Disclosure: I founded Concord Passport.
Rui, that's an important distinction, and I agree.
Our initial exercise was exploratory rather than a controlled, repeated-run study, so the results shouldn't be interpreted as evidence of persistent differences between platforms.
Repeating identical prompts within each platform, while recording the execution date, interface and complete outputs, would make the comparison substantially more useful.
We're incorporating that distinction into how we structure and report future AI visibility assessments. Appreciate you raising it.