100 of the world's largest travel brands, the top 20 by revenue in each of five categories, scored against our AI visibility checker. AI assistants are becoming the front door of travel discovery. This is a benchmark of the field.
Travel discovery is moving inside AI assistants faster than any behavioural shift we've measured in a decade. Most travel brands know it. Few have done anything about it. The audit found the same gaps across every category we tested.
Search has always been the front door of travel marketing. That door is being replaced. What was once a ranked list of options has become a single synthesised answer, and the rules for being inside that answer are fundamentally different.
Travellers scrolled a results page and clicked one of ten links. Generic destination pages were good enough; depth, structure and entity coverage didn't matter much because the click decided the winner, not the content itself.
An AI assistant gives one answer. Your brand is named in it, or it isn't. Winning in this new layer (increasingly called GEO, or Generative Engine Optimization) means content engineered for factual density, entity coverage and machine-readable structure: the signals AI models actually use to choose a source.
We selected one representative content page per brand: the kind of URL a traveller would actually research, not the homepage. Every page was scored against the same framework, which combines three layers of signal into a single 100-point score.
The framework is identical to the public Obvlo AI Visibility Checker, so any travel business can reproduce a brand's score in seconds. The brand-level dataset is held privately; this report shares only category-level distributions.
llms.txt, FAQ blocks, citation-friendly formatting, semantic HTML: the signals AI crawlers specifically read.Page is reachable, returns 200, not soft-blocked by robots.txt or JS gating.
Primary content present in initial HTML, not injected client-side.
Valid schema.org markup (TouristDestination, Hotel, Place, Article…).
llms.txt presenceValid /llms.txt pointing AI crawlers to canonical content.
Distinct named entities (places, neighbourhoods, dishes, events) present and consistent.
Specific, verifiable facts per 500 words above category baseline.
Author, publication date, last-updated visible in HTML and schema.
Content updated within the last 180 days or marked seasonally current.
Headings, lists, definitions and Q&A blocks AI models extract cleanly.
Inbound mentions from third-party sources AI crawlers already index.
Want to see how a single page on your site scores against this framework? The same checks, instant first results.
Run the free check →Each dot is one travel brand's representative page. The horizontal axis is its AI visibility score out of 100; the colour is its category. Names are withheld; distributions are not.
Click a category to isolate it. The story is in the shape: where each category clusters, how wide each spread is, and how few brands cross the 60-point threshold where AI citation becomes consistent.
Vertical position carries no meaning; dots are jittered for visibility only. Hover any dot for its category and score.
Aggregate scores tell you the size of the problem. Sub-scores tell you the shape. And the shape is identical across every category we tested: solid infrastructure, patchy content, almost no AI readiness. The travel industry has built websites for the search engine of 2018, not the one that's already replacing it.
No category averages above 60. But the ranking is not what most travel executives expect. The smallest category by revenue (tour operators, median $500M) outscores the largest (airlines, mean revenue $20B+) by 25 points. Every category has at least one outlier in the 70s. The problem is solvable inside the category structure, not because of it.
The David category. Median revenue here is a fraction of any other category, but seven of twenty cross the AI-ready threshold, and the only brand to max AI readiness sits here.
Second at everything. OTAs come second on every sub-score: 56% on infrastructure, 25% on AI readiness, 44% on page analysis. Competent at every layer, leading on none. The category that should, on paper, be most AI-ready isn't.
The paradox. Their job is to be discoverable. The DMO responsible for one of the world's three largest tourism markets (by receipts) scored 11/100. Public mandate, public money, private visibility.
12× intra-category gap. One global chain scored 74; another in the same revenue bracket scored 6. The difference is which one decided destination content was a brand asset and which decided it was overhead.
The outliers don't cluster. Three carriers cleared the AI-ready threshold with scores of 64, 78, and 81, one from each major region. A fourth sits just below at 59. The other 16 carriers, including several of the world's largest by revenue, scored a median of 13.
Where would your brand sit on this table? Same checks, your URL.
Score your site →Of the 100 brands we audited, 85 had at least one significant failure. Some of those failures are surprisingly basic. Most have a fix measured in hours, not quarters.
% of brands with this check flagged as a top issue (n = 85 of 100 with issues recorded).
Every category has its outliers: the brands scoring 70+ when their peers score 20. The top-decile pages share remarkably consistent traits, regardless of who runs them. None are about creative brilliance. All are about structural discipline.
FAQ schema is the most-missed check in the dataset at 65% of brands. It's the lowest-effort layer-2 signal AI crawlers specifically read; AI assistants extract FAQ blocks almost verbatim into conversational answers. Cheapest fix on the list, most missed.
The brands at 60+ write pages that answer the traveller's question ("what is this place really like?") before the brand asks for the booking. The brands at sub-30 use destination pages as funnels straight to a booking widget.
Not just Article. TouristDestination, Place,
FoodEstablishment, Event, TouristTrip: properly nested, with
geo coordinates where relevant. The top decile averages 6+ schema types per page; the median brand averages 1.
Top pages show dateModified within the last 90 days. Among at-risk brands, the average page has not been touched in 18+ months, and the timestamp is buried in CSS, invisible to AI.
The brands that have crossed the AI-ready threshold are disproportionately the smaller ones in their category. Big organisations have more content but more friction; smaller ones ship the markup in a week, not a quarter.
AI models follow the citation graph. Pages that cite credible third-party sources (council records, official tourism boards, press) accumulate citations themselves: the same flywheel that built Wikipedia, now operating in real time.
If you're below 60, you don't need a rebrand. Four of the five moves below are configuration changes, none requiring a CMS migration; the fifth is the longer content play that separates the AI-ready leaders. (If your score sits in single digits with all of it on infrastructure, our crawler couldn't reach the page. See the methodology note above.)
The single most-missed check in the report. 65% of audited brands lacked it. AI assistants extract FAQ blocks almost verbatim into conversational answers, and the markup is 15 lines of JSON-LD per page.
37% of brands have inadequate schema.org markup on the pages we tested. Use
TouristDestination or Article as the root, nest Place,
Event and FoodEstablishment entities. This alone closes most of the
AI-readiness gap.
llms.txt file at your root.Lowest-effort, highest-signal change in the report. One file, ~30 minutes of work, no engineering ticket. Point AI crawlers at the content you actually want cited, not your booking funnel. The fact that almost nobody has done this is the opportunity.
In HTML, in schema, and ideally in your llms.txt sitemap. 19% of brands
had content freshness flagged. AI models discount stale content sharply; updating the date alone
isn't enough, but the absence of one is fatal.
The pattern across every AI-ready brand in this audit: their winning pages answer the traveller's question about the destination before the brand asks for the booking. The four fixes above prepare your existing content for AI extraction. This is the work that decides whether AI models want to cite you at all. GEO stops being a technical exercise here and becomes a content strategy: original, factually dense, destination-led pages your competitors aren't writing. It's the work Obvlo does for travel brands.
Every score in this report came from the same free tool we offer the public. No sign-up, no credit card. Paste a URL and see how your destination content scores against the brands in this benchmark.