Methodology
The index answers one question: when travelers ask AI assistants where to go, which destinations do the assistants name? This page documents the instrument, the formulas and the limits of that measurement.
AI assistants have become a primary travel advisor. Ahead of any booking, travelers ask them open questions: where to go on safari, where to honeymoon, where to travel in Asia. The assistants answer with destination names, and those answers shape where people actually go.
The AI Travel Visibility Index measures those answers systematically. It tracks how often each destination is named by each assistant, segment by segment, month after month. It measures visibility in AI answers: it is not a quality judgment on any destination.
The index is published per segment: 6 travel intents (how travelers ask by trip type) and 3 regions (how travelers ask within a continent). Each segment has its own tracker and its own ranking; the panel is shared.
Every segment is measured on the same fixed panel of 88 destinations: countries, regions and islands that recur in travel planning worldwide. The panel is identical across segments and frozen over time, so a movement in the index reflects a change in AI answers, never a change in what we measure.
The 88 destinations of the panel:
Each segment is measured once a week. A survey samples 6 answers from each AI assistant, each answering a destination-agnostic traveler question in English, through the assistant's consumer product. Questions never name a destination: the assistants volunteer them.
Every answer is scanned for the panel's destinations. A destination counts as cited when it is named as a recommendation in the answer. The analysis is deterministic: the same answers always produce the same scores.
The published monthly value is the average of the month's weekly readings. Weekly readings are measurement internals and are not published; monthly smoothing absorbs week-to-week noise in assistant answers.
The index currently measures 4: ChatGPT, Claude, Gemini, Perplexity. Each assistant is asked the same questions the same week. Assistants are measured through their consumer products, as travelers use them.
For a destination d and a month, each assistant e has a citation rate: the share of that assistant's answers during the month that name d. The index averages those rates across assistants and scales to 100.
index(d) = 100 × mean over assistants e of rate(d, e)
where rate(d, e) is the share of assistant e's answers during the month that name destination d.
The number of assistants whose monthly citation rate for the destination is above zero. A 4/4 consensus means every measured assistant named the destination at least once that month.
The share of all destination mentions in the month's answers that belong to this destination, averaged across assistants. Where the index says how often a destination appears, voice share says how much of the conversation it takes.
The change of the index versus the previous published month, in points. A destination cited this month but not the previous month is marked as new.
Rankings show the destinations cited by at least one assistant, ordered by index. Ties are resolved deterministically: consensus first, then voice share, then alphabetical order.
A destination's absence from a ranking reflects the assistants' answers to that segment's questions, not an editorial judgment. The index celebrates the destinations AI assistants cite; it does not grade the ones they do not.
Rankings are not for sale. No destination, tourism board or business can pay to enter, improve or influence a ranking. Measurement is operated by Epovest, an independent instrument that tracks what AI assistants say about any entity; Charts Travel publishes the travel index it measures.
You may reuse these figures with attribution: "Charts Travel, AI Travel Visibility Index, July 2026", with a link to the ranking you cite.