Why it matters
A growing share of visitors plan their trip by asking ChatGPT, not by scrolling through search results. The AI gives them a short list. The businesses on that list get the visit; the rest never come up.
25%
of American travellers used generative AI to plan their summer trips in 2026, up from 15% a year earlier.
49%
of those AI users asked it to find restaurants. 54% used it to research where to stay.
1%
lower: how AI-referred visitors to US travel sites converted compared with everyone else in July 2026. A year earlier it was 47% lower.
67%
longer on site for visitors who arrive from an AI answer, and 42% less likely to leave straight away.
Do they buy?
A year ago, people who clicked through from an AI answer to a travel website were much less likely to book than everyone else. By July 2026 they converted almost exactly as well, and they stayed longer.
| Other visitors | 100 (baseline) |
|---|---|
| AI visitors, July 2025 | 53 (47% lower) |
| AI visitors, July 2026 | 99 (1% lower) |
Index derived from Adobe Digital Insights, AI traffic to US travel sites (July 2026 vs July 2025, published Aug 2026).
Compared with visitors from Google, it depends on the study, the sector and the year. We show all three rather than pick the most flattering one.
US travel websites
AI visitors converted 1% below other visitors in July 2026, up from 47% below in July 2025.
Travel sites in the US; compares AI with all other traffic.
Adobe Digital Insights, AI traffic to US travel sites · July 2026 vs July 2025, published Aug 2026
95 websites, mixed sectors
AI referrals converted at 7.18% vs 6.08% for Google organic, about 1.2× better. For consumer-facing (B2C) sites: about 1.5×.
Business-to-business product sites were the exception: AI converted at 0.6× organic.
SEO Works, AI referral traffic benchmark · 95 websites, Sep 2025 – Aug 2026
973 online shops
Google organic converted about 13% better than ChatGPT referrals, though ChatGPT’s rate rose over the year.
Online retail, an earlier period (2024–25), and the only peer-reviewed study here.
Kaiser & Schulze, Marketing Science 45(4), 2026 · 973 e-commerce sites, Aug 2024 – Jul 2025, peer-reviewed
The honest summary: Google still sends far more visitors. In the 95-site study, organic search delivered 39 times more traffic than AI assistants. But AI is the fastest-growing way tourists find businesses, and the people it sends are about as likely to book.
The obvious part
Google shows ten results and the tourist picks. ChatGPT names a handful of businesses, usually four or five, and explains why. If you are not one of them, you are not in the conversation at all.
Visitors plan from home in the US or the Netherlands, weeks ahead. By the time they walk past your door, the dive shop, the dinner and the car rental are often already chosen.
It builds its answer from your website, Google listing, Tripadvisor and island guides. When those are wrong or out of date, AI repeats the mistake. In our testing, several Bonaire businesses had ChatGPT business cards linking to the same unrelated website.
Bonaire has a few dozen dive shops, and a similar handful of operators in most other categories, all competing for the same visitors. Every question you are missing from is a booking that goes to someone down the road.
1 · Free
Find your business in the rankings: which questions you show up for, and who gets named when you don’t.
See the rankings →2 · Free
Claim your profile to see what AI says about you word for word, and the problems we found in how it lists you.
Claim your profile →3 · With Regulars
Our partner Regulars fixes your listings and website so AI gets your facts right. Your score is measured the same way as everyone else’s.
Visit Regulars →Find out in your free report: every question you show up for, what AI says about you, and anything it gets wrong.
These studies mostly measure US travellers and US websites, so read them as a direction rather than a Bonaire figure. As our own monthly scans build up, we will add island-specific data here. For how we measure the rankings themselves, see How we measure.