What Travellers Asked AI About Luxury Travel in Q2 2026

Search data told us what people typed into a box. AI citation data tells us what they actually asked — in full sentences, with intent attached. This is a quarter of those questions, measured rather than guessed: what affluent travellers put to ChatGPT, Copilot, Perplexity and Gemini about luxury travel between April and June 2026.

How this data was gathered

Every figure below is first-party. When an AI assistant answers a travel question by citing an Uncompromised Travel page, that event — the verbatim question, the assistant's intent classification and the citation — is recorded. Between the site's launch in late March and 30 June 2026 we logged 31,085 citations across 267 pages; the grounding-query dataset behind this report covers the 17,600 citation events attributable to 576 distinct questions.

The ten most-asked questions of the quarter, verbatim:

Compensation first, inspiration later

The single largest thing affluent travellers asked AI about in Q2 was not where to go. It was how to get paid when flying fails. Questions about EU261, claims companies and disruption rights generated 5,319 citation events — 30.2% of the entire tracked dataset, and five of the ten most-asked questions of the quarter.

The phrasing is telling. People do not ask the machines "what are my rights?"; they ask "most trusted airline compensation providers" and "AirHelp EU261 claims effectiveness reputation fees success rate" — vendor due diligence, conducted at scale, before a single claim is filed. It is the behaviour of a buyer who has already decided to claim and is now shortlisting counsel. For the underlying answers: filing yourself takes about twenty minutes, our comparison of the four major claims firms covers fees and success rates, and AirHelp remains the firm we would hand a claim to first.

The safari due-diligence boom

Safari generated 3,969 citation events — 22.6% of the dataset — and almost none of them were about animals. Travellers asked the machines to benchmark operators: on pricing transparency, on customer-service records, on accommodation quality, on "consistent performance history". A five-figure safari is now bought the way a company buys software — with a vendor evaluation — and the machines are being used as the analyst.

The demand for pricing transparency in particular — the No.1 safari question, by a distance — is the industry's quiet indictment: safari pricing remains opaque enough that travellers have outsourced the comparison to machines. It is why our Luxury Safari Operator Index ranks operators on exactly that, why Kenya vs Tanzania vs South Africa exists for the first-timer question, and why BookAllSafaris — where live itineraries and dates can be compared across vetted operators — is the practical next step once the shortlist exists. With Mara-crossing season running July to October, expect this cluster to grow again in Q3.

Status, points and plastic

Add them together and the loyalty economy was the quarter's second force: status and loyalty questions drew 1,310 citation events, premium travel credit cards 1,244, and cabin-class rankings 1,371. The most-asked single question in the entire dataset — "hotel loyalty program status match guide", 494 asks — belongs here, alongside the bluntly transactional "credit cards instant hotel elite status" (162) and "affluent premium credit cards travel benefits overview" (107).

The pattern is arbitrage-minded: travellers are not asking whether loyalty is worth it, they are asking the machines to find the shortcut — which status matches are open, which cards fast-track elite tiers, and which premium card earns its fee. Cabin questions ran the same way: "airlines with first class cabins 2026" (160 asks) and "best airlines for business class travel 2026" (132) are ranking requests, answered in our first-class and business-class indices.

Private aviation: the empty-leg question

Private aviation drew 831 citation events, and the standout was pragmatic: "best practices empty leg flight business travel", asked 121 times, with our empty-leg guide cited in 41.2% of responses. The honest answer bears repeating — the discount on a repositioning flight is real, the schedule belongs to the aircraft, and it suits flexibility rather than fixed diaries. The rest of the cluster was cost literacy: hourly charter economics and membership and jet-card comparisons.

Trains, platforms and a festival a year away

Three long-tail findings worth anyone's attention. First, luxury rail (453 citation events) is priced-question territory: "twilight express mizukaze price" was asked 99 times — a sixteen-room train on the Japan Sea coast, answered in our Mizukaze cost guide — with the Venice Simplon-Orient-Express, run by Belmond, drawing the same what-does-it-actually-cost scrutiny.

Second, travellers now benchmark the booking platforms themselves: "GetYourGuide OTA company features comparison" logged 118 asks, feeding our four-way platform comparison — and for what it is worth, GetYourGuide still wins on cancellation terms for most premium experiences. Third, event queries form absurdly early: "fallas 2027" was asked 154 times in June — a Valencia festival eight months away, already being researched through machines, covered in our Fallas 2027 guide and by the city's official tourism office.

How people actually talk to the machines

The intent classifications are the most revealing table in the dataset. Research dwarfs everything; explicit trip-planning barely registers.

Note the register, too. Nobody writes "we're dreaming of a safari" to a machine. They write "top safari operators with consistent performance history" — the language of a procurement brief, not a postcard. Travellers address AI assistants the way they would address an analyst, and they expect analyst-grade answers: benchmarks, rankings, named trade-offs, current prices.

What it means

For travellers: the machines are genuinely good at the due-diligence layer, and this data suggests that is precisely how sophisticated buyers already use them. The one discipline worth adding is to open the cited source — check the methodology, check the date — before a five-figure decision rests on a paragraph of synthesis.

For the trade: your brand is now benchmarked by machines drawing on independent indices, and the questions being asked are pointed — pricing transparency above all. Operators whose rates require a phone call to discover are, query by query, being filtered out of answers they never see. The uncomfortable conclusion of Q2's data is that opacity has become a distribution problem.

We will publish the Q3 edition of this index in early October, with quarter-on-quarter movement. Journalists and researchers are welcome to use these figures with attribution; the underlying category and intent tables are available on request via the editor.