# Travel AI Monitor: English, full text

> A sourced, continuously reviewed dataset on artificial intelligence in travel and tourism. Every figure carries its publisher, field period, sample and a link to the original document.
> Dataset last reviewed end to end: 2026-10-02. https://travel.rankwit.ai

## Questions

### How many travellers use AI to plan trips?
**Short answer.** Between 18% and 79%, depending on what is being asked. The defensible summary: roughly half to two-thirds of travellers have used AI somewhere in a trip, and around one in five uses generative AI specifically to plan one.
No single number answers this, and anyone quoting one without its question wording is quoting a coincidence. The published range for 2025 and 2026 spans more than sixty percentage points, and every figure in it comes from a credible publisher with a disclosed method.
The spread resolves once you separate three questions that get asked interchangeably. 'Have you ever used AI in connection with travel' returns the highest numbers: Amadeus puts it near 79% for March 2026, Booking.com at 67% for 33 markets in 2025. 'Do you use AI to plan trips' returns the middle band: Sojern 54% for US travellers, Euromonitor 51% globally. 'Do you use generative AI to plan travel' returns the lowest: Amadeus 18% for 2025.
For a general audience, the honest formulation is that about half of travellers have used AI for trip planning and the share is rising by double digits annually. For a board paper, cite the specific instrument and its wording.
**Why credible sources differ.** Three variables drive almost all of the variance: question scope (any AI use versus generative AI planning specifically), population (US panels run above 33-market global samples), and date (the series moves several points a quarter, so early-2025 and spring-2026 readings are not comparable). None of the sources is wrong; they are measuring different things.
**Read from this market.** For a US audience the cleanest number is Phocuswright's 56%, because it comes from three consecutive waves of the same question. Quote the global figures only when the audience is global; the US runs ahead.
https://travel.rankwit.ai/questions/how-many-travellers-use-ai-to-plan-trips
### Do travellers trust AI for travel information?
**Short answer.** They use it far more than they trust it. 6% of consumers fully trust AI, 42% always fact-check what it tells them, and 25% have received travel information from it that was outdated or wrong.
Usage and trust have come apart, and the gap is not closing. In the same Booking.com survey of 37,325 people, 91% said they were excited about AI and 6% said they fully trusted it. Both numbers are real and they describe different things: appetite for the capability, and willingness to act on its output unverified.
The verification behaviour is the practical consequence. Seven in ten travellers check at least sometimes, four in ten always. That makes the AI answer the opening move in a research sequence rather than its conclusion, and it means a wrong answer about your property is likely to be found out.
Regional variation is large enough to change market strategy. 32% in North America and 29% in Europe and the Middle East (the report's own region, not Europe alone) rarely or never trust AI-generated information, against 15% in Latin America and 16% in Asia Pacific.
**Why credible sources differ.** Trust measures diverge by what they ask about. Travelport finds 42% trust travel brands to use AI responsibly, a question about institutions. Booking.com finds 6% fully trust AI, a question about the technology. Both are consistent: people trust the company more than the tool.
**Read from this market.** US scepticism is the highest in the dataset. If you are writing for an American market, lead with verification behaviour rather than enthusiasm: 42% always fact-check, and that is the audience you are actually addressing.
https://travel.rankwit.ai/questions/do-travellers-trust-ai-for-travel
### Will travellers let AI book travel for them?
**Short answer.** Mostly not yet. About 12% are comfortable with AI making independent decisions, around 6% would let it handle most of their planning, and a quarter to a third would let it complete a booking they had already specified.
Delegation is where the AI travel story currently stops. Adoption of AI for planning is approaching the majority in most markets; adoption of AI for deciding is in single digits.
The ordering of the three figures is informative. Fewest travellers will let AI decide what to do (6%). Slightly more will let it make independent decisions in general (12%). Considerably more (a quarter to a third, depending on the market) will let it execute a booking they have already defined. Travellers will delegate the transaction, not the judgement.
The 80% of frequent travellers who report comfort with AI for planning is not in tension with this. Comfort with a tool and authority over a decision are different grants, and travellers are currently giving the first and withholding the second.
**Why credible sources differ.** Amadeus (≈6%) and Booking.com (12%) differ mainly because Amadeus asks about handling 'most' of a trip while Booking.com asks about independent decisions in general. Phocuswright's quarter-to-a-third range is higher again because completing a specified booking is a narrower, lower-risk grant of authority.
**Read from this market.** The 80% comfort figure for frequent travellers is a global Travelport reading, not a US cut; no US-only version of it is published. Comfort and authority are different grants either way, and for an American audience the scepticism data is the better guide to how wide that gap runs here.
https://travel.rankwit.ai/questions/will-travellers-let-ai-book-for-them
### Is AI replacing search engines for travel research?
**Short answer.** It is displacing channels rather than replacing search outright. AI assistants are now a more trusted planning source (24%) than travel bloggers (19%) or social media influencers (14%), and 69% of Millennials use AI as often as social media for trip planning.
The clearest evidence is comparative. Asked to name trusted travel planning sources, consumers put AI assistants above both of the creator channels built to influence travel decisions: travel bloggers at 19% and social media influencers at 14%. The channel losing ground here is not search; it is the people who made a business of recommending places.
Generational data sharpens it. For 69% of Millennials and 63% of Gen Z, AI is now consulted as often as social media when planning a trip. Social platforms were the default inspiration channel for both cohorts for most of the last decade.
What AI is displacing is the research sequence, not the transaction. Travellers use AI to narrow the field, then move to a booking surface. 56% use it to compare prices but only 32% to compare flights and hotels specifically; the gap marks the boundary.
For operators the consequence is structural. A search results page shows ten options; an AI answer names three or four. Positions eleven through a hundred did not have much value before, but they existed. In an AI answer they do not.
**Why credible sources differ.** Measures of 'AI versus search' rarely compare like with like. Trust rankings, time-spent data and referral analytics each tell a different story. We rely on trusted-source rankings and behaviour-change measures because they are the ones tied to a decision.
**Read from this market.** In the US the question is really about Google, not about standalone assistants: most AI answers travellers see are AI Overviews. The 50% click-through figure is therefore the number that decides how much of your traffic is actually at risk.
https://travel.rankwit.ai/questions/is-ai-replacing-google-for-travel-search
### What do travellers actually use AI for?
**Short answer.** Researching attractions, restaurants and transport (60%), comparing prices (56%), in-trip translation (45%), building itineraries (43%) and destination research (38%). Synthesis tasks dominate; transaction tasks lag.
The task-level data is more actionable than any adoption headline, because it tells you which questions a model will be asked on your behalf.
A consistent pattern runs through every survey. AI is strongest where the job is synthesis across many scattered sources: what is worth seeing, where to eat, what a reasonable price looks like, how to sequence five days. It is weakest where the job is a transaction with money and dates attached.
In-trip use is higher than most operators expect and is frequently overlooked. 96% of travellers who use AI in travel use it while actually travelling: translation at 45%, activity suggestions at 44%, navigation at 40%. The traveller who planned without AI will still ask it where to have dinner once they have landed.
**Read from this market.** The US task breakdown is the most detailed available, and research into attractions, restaurants and transport tops it at 60%. For American destination marketing that is where content investment returns fastest.
https://travel.rankwit.ai/questions/what-do-travellers-use-ai-for
### How many hotels and tourism businesses use AI?
**Short answer.** Around 70% of tourism businesses use AI, mainly for chatbots, dynamic pricing and demand forecasting. But adoption is strongly size-dependent, and operational AI is not the same as being visible in AI answers.
UN Tourism's figure of 70% is the highest-authority adoption number available for the supply side, and it counts operational AI inside the business: service chatbots, revenue management, forecasting.
That distinction matters more than the level. A hotel can run AI pricing, AI messaging and AI forecasting and still not be named when a traveller asks an assistant where to stay in its city. Internal adoption and external visibility are independent variables, and only one of them is measured by this statistic.
The size effect runs through everything. European accommodation chains rate their current situation positively at 72% against 55% for independents; 94% of large properties feel prepared for cybersecurity threats against 60% of properties with fewer than ten staff. The capability gap is consistent and it widens as properties get smaller.
**Why credible sources differ.** UN Tourism's 70% does not disclose a sample size, which is why we treat the level as indicative. Property-level surveys from Booking.com, with disclosed field periods, consistently show lower capability among independents than the sector aggregate implies.
**Read from this market.** US hotel groups report higher operational AI adoption than the global average, but the independent segment, which is most of the country's room supply, tracks the global independent figures, not the chain ones.
https://travel.rankwit.ai/questions/how-many-hotels-use-ai
### Are destinations ready for AI search?
**Short answer.** Not yet. 22% of DMOs offer AI trip planning on their site and 26% have a documented AI strategy, while around 40% of European national tourism organisations have a dedicated AI team or monitor AI use.
Destination organisations are the clearest documented case of the industry trailing its own audience, because one publisher measured both sides in the same month.
In April 2026, 54% of travellers had used AI to plan a trip. In the same month, 22% of destination organisations offered any AI planning capability on their own website. Sixty-three percent of travellers said they would use one if it existed.
Europe's national tourism organisations sit in a similar place. Around 40% have a dedicated AI team or monitor AI use, which the European Travel Commission attributes to skills and resources rather than scepticism. Note the weaker claim: having a team is not the same as running AI in operations, and the report publishes no sub-regional breakdown.
There is one countersignal worth weighing. 64% of destination marketers now produce AI-formatted content: lists, FAQs, structured answers. The content layer is moving faster than the product layer, which is rational: being correctly described in an AI answer is cheaper and more urgent than shipping a planner.
**Read from this market.** The 22% comes from 103 DMO leaders whose countries Sojern does not break out, so treat it as a global reading rather than an American one. State tourism offices and city CVBs are the US organisations it describes, and the budget cycle, not conviction, is what holds them.
https://travel.rankwit.ai/questions/are-destinations-ready-for-ai-search
### How big is the gap between travellers and the travel industry on AI?
**Short answer.** Roughly two years, by the publisher's own framing. 54% of travellers plan with AI; 22% of destinations offer it. 72% of destination leaders say two more years of inaction risks irrelevance.
Sojern titled its 2026 study 'Travelers Are Two Years Ahead of the Travel Industry on AI', and the paired survey design is what makes the claim checkable rather than rhetorical: travellers and DMO leaders, two questionnaires, one month. The traveller half is US measured; the DMO half is not broken out by country.
The gap is 32 points on current use and 41 points on stated demand against available supply. Those are not projections; they are two measurements taken at the same time.
Awareness is not the bottleneck. 72% of DMO leaders name the risk explicitly. 26% have written a strategy. The constraint is execution capacity, which is a budget and skills problem rather than a belief problem.
The asymmetry is what makes this urgent rather than merely interesting. Traveller adoption compounds on its own; institutional adoption requires a decision, a budget cycle and a procurement process.
**Read from this market.** The traveller half of this gap is US-measured (1,006 American travellers), while the 103 DMO leaders are not broken out by country. It is still the best-evidenced version of the claim anywhere, because both halves come from one study fielded in one month; it is just not a US-versus-US reading.
https://travel.rankwit.ai/questions/how-big-is-the-ai-gap-in-travel
### Does AI change where people actually travel?
**Short answer.** Yes. 40% of travellers discovered a destination through AI and 29% changed their plans based on an AI recommendation. And about half of travellers who see an AI answer in a search engine still click through to the source.
The discovery figure and the plan-change figure are the two measures that convert AI visibility into demand, and they are both behavioural rather than attitudinal.
Four in ten travellers encountering a destination for the first time through an AI answer is the inspiration-stage number. Nearly three in ten changing an already-formed plan on an AI recommendation is the stronger finding, because it measures an outcome rather than an impression.
The counterweight to the zero-click worry is measurable: Phocuswright finds about half of travellers who meet an AI answer inside a search engine still click through to the sources. Being named in the answer is therefore not only a branding outcome; for roughly half the audience it still sends a visit.
The counterweight is concentration. An AI answer names a handful of options. Discovery expands for whoever is named and collapses for whoever is not.
**Read from this market.** 40% of US travellers discovering a destination through AI matters more in a market where domestic travel dominates: the competitive set an AI answer names is often a list of American states and cities, not countries.
https://travel.rankwit.ai/questions/does-ai-change-where-people-travel
### How accurate is AI travel information?
**Short answer.** 25% of travellers report receiving outdated or inaccurate travel information from AI, and that counts only the errors they noticed.
A quarter of travellers have been given travel information by an AI that was wrong or out of date. Because the measure is self-reported, it captures errors that were caught; the true rate is higher.
The diagnosis matters for anyone trying to fix it. Most travel inaccuracy is not the model inventing facts from nothing. It is the model answering from stale or missing information: hours that changed, a property that renovated, a route that no longer runs. Where current structured facts are not reachable, a model answers from whatever it last saw, with the same confident tone.
This is why 42% always fact-check and only 6% fully trust. Travellers have calibrated to the error rate.
The practical response is not to argue with models. It is to make current facts machine-readable, structured and reachable, which is the discipline this observatory applies to itself, and the one it recommends.
**Read from this market.** In the US the practical risk is a wrong answer about hours, closures or policies reaching a traveller who then arrives and finds otherwise. That is a review problem before it is an AI problem.
https://travel.rankwit.ai/questions/how-accurate-is-ai-travel-information
### Which generation uses AI most for travel?
**Short answer.** Millennials, narrowly. 69% use AI as often as social media for trip planning, against 63% of Gen Z.
Millennials lead, which surprises people who expect a clean age gradient. The six-point gap over Gen Z is within the uncertainty of undisclosed subsamples, so treat the two as broadly comparable rather than ranked.
The likelier explanation for Gen Z not leading is the comparator. The question asks about AI use relative to social media, and Gen Z's social media baseline is the highest of any cohort. Matching it is a higher bar.
The more useful reading is that both cohorts have reached parity between AI and social platforms for trip planning. For destination marketing budgets still weighted heavily towards social, that is the finding with a cost attached.
**Why credible sources differ.** Generational splits in this dataset come from subsets of a 1,006-person US panel with undisclosed subsample sizes. Directionally reliable, not precise, and not automatically transferable outside the United States.
**Read from this market.** The generational splits here are US subsamples. Millennials and Gen Z together are now the majority of American travel demand, which is what makes their parity between AI and social a budget question rather than a curiosity.
https://travel.rankwit.ai/questions/which-generation-uses-ai-most-for-travel
### How much is AI worth to the travel industry?
**Short answer.** The most-cited figure is Euromonitor's projection of USD 1.4 trillion of travel spending growth over five years attributed to AI-unlocked potential. It is a forecast, not a measurement.
Economic claims about AI in travel divide into two kinds, and treating them as one is the most common error in industry commentary.
Measured quantities exist and are modest. 40% of AI-adopting travel agencies report a measurable productivity impact, meaning three in five do not. 33% of US travel executives report improved personalisation. These come from surveys with disclosed samples and they describe effects already observed.
Projected quantities are much larger and much softer. The USD 1.4 trillion figure is a five-year travel spending forecast with the growth attributed to AI by analyst judgement; the decomposition is not published. We label it a forecast on every page it appears on.
The gap between near-universal adoption and minority measurable return is the honest state of the evidence in 2026. The vendor research attributes it mainly to data fragmentation rather than to model capability.
**Why credible sources differ.** Forecasts and measurements are not comparable and should never be averaged. This observatory marks every figure with its evidence type for that reason.
**Read from this market.** The measured returns in this answer come from mostly US executive samples, and they are modest. The large numbers are global forecasts. Do not mix them in an American business case.
https://travel.rankwit.ai/questions/how-much-is-ai-worth-to-travel
### Are airlines and travel agencies using AI agents?
**Short answer.** Adoption is near-universal and return is not. 80% of travel agencies have already implemented AI and the remaining 20% plan to, and 52% of executives across industries report agents running in production, but only 40% of the AI-adopting agencies report a measurable gain.
The supply side of travel distribution is ahead of the consumer-facing side, because its use case is internal productivity rather than public visibility and the return is easier to justify.
The limiting factor is documented and it is not model capability. 91% of travel agencies work across four or more booking systems and half across seven or more. An agent reasoning about availability, price and policy across ten disconnected systems will be wrong often enough to lose the trust it needs to be useful.
This is why the vendor prescriptions converge: simple granular APIs over monolithic ones, explicit domain context, error handling designed for autonomous callers, and governance built in from the start rather than retrofitted.
**Why credible sources differ.** The 52% production figure is cross-industry, from a Google Cloud study, and sits above travel-specific measures. We show it as a benchmark rather than as a travel statistic.
**Read from this market.** US agencies run the most fragmented booking stacks in the dataset, which is the specific reason agentic deployment underperforms here relative to executive intent.
https://travel.rankwit.ai/questions/are-airlines-and-agencies-using-ai-agents
### Where in the world is AI travel adoption highest?
**Short answer.** Latin America leads on enthusiasm (98% excited) and Asia Pacific follows (95%). North America and Europe are the most sceptical. Usage intensity measured by telemetry does not follow market size at all.
Attitudes split sharply by region in the largest survey available. Latin America reports 98% excitement and 89% familiarity; Asia Pacific 95% and 82%. Distrust runs the other way: 32% in North America and 29% in Europe and the Middle East rarely or never trust AI-generated information, against 15% in Latin America and 16% in Asia Pacific.
Telemetry tells a different story from surveys, which is why it is worth having both. Anthropic's Economic Index puts Australia at 6.40 times the population-proportional baseline for Claude usage, the highest of 121 countries. Usage intensity does not track market size.
Within Europe the gradient is institutional rather than attitudinal. Around 40% of national tourism organisations report a dedicated AI team or active monitoring, which the European Travel Commission attributes to skills and resources.
The operational implication is that market sequencing should follow attitudes and usage intensity rather than traffic volume. The same AI-forward positioning performs differently in São Paulo and Chicago.
**Read from this market.** Not the United States, on either measure. America is the most sceptical region in the attitudes data, and on usage intensity the telemetry leader is Australia at 6.40 times its population share. For an American audience that argues for evidence-led positioning rather than enthusiasm-led.
https://travel.rankwit.ai/questions/where-in-the-world-is-ai-travel-adoption-highest
### How do I get my hotel recommended by ChatGPT?
**Short answer.** Make current, specific, structured facts about the property reachable by machines, keep them fresh, and make sure the third-party sources models already trust describe you accurately.
Start from how the failure actually happens. 25% of travellers have received outdated or inaccurate AI travel information. In hospitality that is rarely invention; it is a model answering from stale or missing data, because nothing current and structured was reachable when it looked.
That makes freshness and structure the first-order work, ahead of anything that resembles keyword optimisation. Hours, policies, room types, amenities, location relationships and price positioning need to exist as explicit, machine-readable facts that are updated when they change.
Second, the questions being asked are specific, not generic. Travellers ask about attractions, restaurants and transport (60%), price comparisons (56%) and itineraries (43%). A property described only in brand language will not be matched against 'quiet boutique hotel near the Duomo with a late breakfast'. One described in concrete attributes will.
Third, models ground their answers in sources they already weight. 64% of destination marketers now format content for AI retrieval: lists, FAQs, structured answers. Your own site matters, and so does how the directories, review platforms and destination sites around you describe you.
Fourth, measure it. Being named is not a binary; it varies by prompt, by engine and by week. Without tracking you cannot tell a wording problem from a source problem.
**Why credible sources differ.** No published study measures citation share by tactic, and anyone claiming a ranking formula for generative engines is extrapolating. The guidance here follows from documented failure modes (staleness, missing structure, generic description), not from a disclosed ranking algorithm.
**Read from this market.** For US properties, the third-party sources models lean on are predictable: Google Business Profile, major OTA listings, and the destination organisation's own site. Fix those three before touching your own copy.
https://travel.rankwit.ai/questions/how-do-i-get-my-hotel-recommended-by-chatgpt
### Is AI use in travel still growing, or has it plateaued?
**Short answer.** Still growing. Generative AI travel planning grew 64% year on year to 18%, US travellers using AI on at least one trip went 33% to 43% to 56% across three consecutive waves, and intent runs well ahead of behaviour on both US and global measures.
Three independent measurements point the same way, which is more persuasive than any level reading. Phocuswright, asking the same question in three consecutive waves, recorded 33%, then 43%, then 56% of US travellers using AI on at least one trip. Amadeus recorded generative AI travel planning rising from 11% to 18% between 2024 and 2025, a 64% increase on consistent wording. Deloitte reports nearly a quarter of travellers using generative AI for trip planning in late 2025, three times the 2022 level.
Intent is the forward indicator and it is not softening. 75% of US travellers intend to use AI for future trips against 54% who already have. Globally, 89% want AI in future travel planning against 67% who have used it. In a saturating category trial typically runs ahead of intent; here it is the reverse.
The constraint on further growth is not interest. It is trust: 6% fully trust AI and 12% would let it decide independently. Expect the planning numbers to keep climbing and the delegation numbers to move slowly.
**Read from this market.** The US series is the one that settles this: three waves, same question, 33% to 43% to 56%. No other market has a comparable run.
https://travel.rankwit.ai/questions/is-ai-growing-or-plateauing-in-travel

## Indicators

### Traveller adoption

How many people actually use AI to plan and book travel, and how fast that share is moving.

The United States carries the longest run of comparable measurements in this field, which is why the adoption range here is anchored on US panels. Treat them as a leading indicator rather than a world average: where Phocuswright records 56% of US travellers using AI on a trip, 33-market global samples sit lower, and the gap is the lead time other markets have.

- **US travellers who used AI on at least one trip: 56%**
  - Reference period: First half of 2026 (2026-03)
  - Geography: United States
  - Evidence type: Survey
  - Source: Phocuswright, https://www.phocuswright.com/Travel-Research/Research-Updates/2026/The-fastest-shift-in-travel-behavior-just-became-the-default
  - Sample: US travellers; The AI Surge: Travel's Fastest Behavioral Shift in a Decade, published March 2026
  - For comparison: second half of 2025 43%
  - https://travel.rankwit.ai/data/us-travellers-ai-any-trip
  - Method: Phocuswright, The AI Surge: Travel's Fastest Behavioral Shift in a Decade, published March 2026. US travellers; waves 1H25, 2H25 and 1H26. Among travellers who have used AI at all, 94% have used it for travel.
  - The clearest series in this dataset: the same instrument, three consecutive waves, one direction. Phocuswright calls it the fastest behavioural shift in travel in a decade, and the wave-on-wave gains hold across every generation.
- **US travellers who have used AI to plan a trip: 54%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults who travelled in the past 12 months
  - https://travel.rankwit.ai/data/ai-trip-planning-us-travellers
  - Method: Online survey of 1,006 US adults who had travelled in the previous twelve months, fielded in April 2026 and published alongside a parallel survey of 103 DMO leaders. Self-reported past behaviour, so it captures any use of AI at any point, not habitual use.
  - This is the figure to use when someone asks whether AI trip planning has crossed into the mainstream in the United States. It has. The same study found the capability was offered by only 22% of destination organisations, which is what makes this number interesting rather than merely large.
- **Consumers who have used AI in some travel context: 67%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - https://travel.rankwit.ai/data/ai-used-in-travel-global
  - Method: Online survey of 37,325 respondents across 33 markets, fielded April–May 2025 for the Global AI Sentiment Report. Among those who had used AI in travel, 98% used it for planning or booking and 96% used it while travelling.
  - The largest disclosed sample in this dataset, and the broadest question. It covers any travel-related use (before, during or after a trip), which is why it lands above the planning-specific figures.
- **Consumers using AI for trip planning: 51%**
  - Reference period: January–February 2025 (2025-01)
  - Geography: Global
  - Evidence type: Survey
  - Source: Euromonitor International, https://www.euromonitor.com/article/ai-a-turning-point-in-global-travel
  - Sample: Voice of the Consumer: Travel Survey
  - https://travel.rankwit.ai/data/ai-trip-planning-global-consumers
  - Method: Euromonitor's Voice of the Consumer: Travel Survey, fielded January to February 2025 across its standard multi-market consumer panel. The question asks about AI broadly rather than generative AI specifically.
  - An early-2025 reading, and one of the first to put global AI trip planning above half. It sits below later US-only figures and above Amadeus's narrower generative-AI measure, which is exactly where a global average should sit.
- **Travellers using generative AI to plan travel: 18%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/documents/resources/research-report/travel-trends-2026/amadeus-travel-trends-2026-report.pdf
  - Sample: Amadeus global traveller survey
  - For comparison: 2024 11%
  - Year on year: +64%
  - https://travel.rankwit.ai/data/genai-travel-planning-share-amadeus
  - Method: Amadeus global traveller survey reported in Travel Trends 2026. Repeated question wording across 2024 and 2025 waves, which makes the year-on-year change more reliable than either level on its own.
  - The most useful growth number in the dataset, because the same publisher asked the same narrow question two years running. The level is low compared with broader AI measures precisely because the question is specific to generative tools used for planning.
- **Travellers who have used AI for trip planning or management: ~79%**
  - Reference period: March 2026 (2026-03)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/en/resources/white-paper/agentic-ai-airlines
  - Sample: Amadeus traveller research, March 2026
  - https://travel.rankwit.ai/data/ai-trip-planning-amadeus-2026
  - Method: Amadeus traveller research conducted March 2026, reported in the airlines agentic AI white paper. The same study found only around 6% comfortable letting AI handle most of their planning.
  - The upper bound of the published range, and from the same publisher that reports 18% for generative AI planning a year earlier. The difference is the question, not the population: this one counts planning and in-trip management, ever, with any AI tool.
- **US travellers who intend to use AI for future trips: 75%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/ai-trip-planning-intent-us
  - Method: Same instrument as the adoption figure: 1,006 US adults, April 2026. Stated intent is a weaker signal than reported behaviour and systematically overstates future action, which is why we show it next to the behavioural number rather than instead of it.
  - Intent runs 21 points ahead of behaviour. In a maturing category you would expect the opposite (trial above intent), so this gap suggests the series has further to run rather than levelling off.
- **US travellers using generative AI for trip planning: ~25%**
  - Reference period: Late 2025 (2025)
  - Geography: United States
  - Evidence type: Survey
  - Source: Deloitte, https://www.deloitte.com/us/en/insights/industry/transportation/travel-hospitality-industry-outlook.html
  - Sample: Synthesis of Deloitte's 2025 Summer Travel, Holiday Travel and Corporate Travel surveys, fielded March to October 2025
  - For comparison: versus 2022 3x lower
  - https://travel.rankwit.ai/data/genai-trip-planning-deloitte
  - Method: Deloitte 2026 Travel Industry Outlook, synthesising its 2025 Summer Travel, Holiday Travel and Corporate Travel surveys, fielded March to October 2025. 'Nearly a quarter' is Deloitte's own wording; we do not sharpen it.
  - A three-year trend from one publisher, which makes the direction more trustworthy than the level. Deloitte's own travel outlook is the primary document; an earlier version of this page attributed the figure to Deloitte's Summer Travel Survey, which does not in fact report on AI.
- **Millennials using AI as often as social media for trip planning: 69%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: Millennial subset of 1,006 US adults
  - For comparison: Gen Z 63%
  - https://travel.rankwit.ai/data/millennials-ai-as-often-as-social
  - Method: Sojern traveller survey, April 2026, generational subsets of 1,006 US adults. Subsample sizes are not disclosed, so the generational splits carry wider uncertainty than the headline figure.
  - The comparison is the point. For most of the last decade social platforms were the default inspiration channel for younger travellers. For a clear majority of Millennials, AI now matches it.
- **Gen Z using AI as often as social media for trip planning: 63%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: Gen Z subset of 1,006 US adults
  - https://travel.rankwit.ai/data/genz-ai-as-often-as-social
  - Method: Sojern traveller survey, April 2026, Gen Z subset of 1,006 US adults. Subsample size not disclosed.
  - Gen Z indexing below Millennials is counterintuitive and worth noting. The likely explanation is that Gen Z's social-media baseline is higher, not that their AI use is lower.
- **Frequent travellers comfortable using AI for trip planning: 80%**
  - Reference period: 2024 (2024)
  - Geography: Global
  - Evidence type: Survey
  - Source: Travelport, https://www.travelport.com/press-releases/travelport-releases-second-annual-state-of-modern-retailing-report-travels-tipping-point
  - Sample: State of Modern Retailing consumer survey
  - https://travel.rankwit.ai/data/frequent-travellers-comfort-ai-planning
  - Method: Travelport's second annual State of Modern Retailing consumer survey, published January 2025. The frequent-traveller subset is self-defined, and the question measures comfort rather than use.
  - Comfort among high-frequency travellers runs well above general-population trust measures. The same report places travel in the top three industries for consumer trust in AI, a better starting position than the sector usually credits itself with.
- **Travellers who find generative AI results helpful: 78%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Phocuswright, https://www.phocuswright.com/Travel-Research/Research-Updates/2025/From-hype-to-habit-78-percent-of-travelers-say-GenAI-improves-trip-planning
  - Sample: Phocuswright, Chat, Plan, Book: GenAI Goes Mainstream
  - https://travel.rankwit.ai/data/genai-helpful-planning-phocuswright
  - Method: Phocuswright, Chat, Plan, Book: GenAI Goes Mainstream, 2025. The figure covers both trip planning and in-destination use.
  - A satisfaction measure, and the one that best explains why adoption keeps climbing. Note the exact wording: helpful, not better than search. The two are often conflated in industry commentary.
- **Consumers who want to use AI in future travel planning: 89%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - https://travel.rankwit.ai/data/ai-future-travel-planning-intent-global
  - Method: Booking.com Global AI Sentiment Report, 37,325 respondents across 33 markets, April–May 2025. Stated preference, not observed behaviour.
  - Demand is close to universal and sits far above both current use (67%) and current trust (6%). Travellers want the capability; what they withhold is authority over decisions.
- **Claude usage index, Australia: 6.40×**
  - Reference period: June 2026 (2026-06)
  - Geography: Australia
  - Evidence type: Platform telemetry
  - Source: Anthropic, https://www.anthropic.com/economic-index
  - Sample: Privacy-preserving aggregate usage across 121 countries
  - For comparison: baseline expectation 1.00×
  - https://travel.rankwit.ai/data/claude-usage-index-australia
  - Method: Anthropic Economic Index, June 2026. Privacy-preserving aggregate statistics across 121 countries; the index is usage share divided by working-age population share. Measures one assistant, not all AI use.
  - Rare first-party telemetry rather than a survey, and a useful corrective: AI usage intensity does not track market size. For travel brands, high-index markets are where AI-mediated discovery is already material.

### Discovery and visibility

AI as a discovery channel: who gets recommended, who gets cited, and who disappears.

US travellers meet AI answers inside Google more than through standalone assistants, which is why the click-through figure matters so much here: roughly half still visit the source. For American operators the practical read is that AI visibility and web traffic are not yet a trade-off.

- **Travellers who still click through after an AI answer: ~50%**
  - Reference period: First half of 2026 (2026-03)
  - Geography: United States
  - Evidence type: Survey
  - Source: Phocuswright, https://www.phocuswright.com/Travel-Research/Research-Updates/2026/The-fastest-shift-in-travel-behavior-just-became-the-default
  - Sample: US travellers who encounter AI answers in search engines
  - https://travel.rankwit.ai/data/ai-answer-click-through
  - Method: Phocuswright, The AI Surge, March 2026. Covers travellers encountering AI answers inside search engines, not standalone assistants, where click-through behaviour differs.
  - The single most useful number here for anyone weighing how much to invest in being cited. AI answers are not purely zero-click in travel: being named in the answer still sends traffic, because roughly half the audience goes to check.
- **US travellers who used AI on at least one trip: 56%**
  - Reference period: First half of 2026 (2026-03)
  - Geography: United States
  - Evidence type: Survey
  - Source: Phocuswright, https://www.phocuswright.com/Travel-Research/Research-Updates/2026/The-fastest-shift-in-travel-behavior-just-became-the-default
  - Sample: US travellers; The AI Surge: Travel's Fastest Behavioral Shift in a Decade, published March 2026
  - For comparison: second half of 2025 43%
  - https://travel.rankwit.ai/data/us-travellers-ai-any-trip
  - Method: Phocuswright, The AI Surge: Travel's Fastest Behavioral Shift in a Decade, published March 2026. US travellers; waves 1H25, 2H25 and 1H26. Among travellers who have used AI at all, 94% have used it for travel.
  - The clearest series in this dataset: the same instrument, three consecutive waves, one direction. Phocuswright calls it the fastest behavioural shift in travel in a decade, and the wave-on-wave gains hold across every generation.
- **AI assistants as a trusted travel planning source: 24%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - For comparison: influencers 14%
  - https://travel.rankwit.ai/data/ai-assistants-more-trusted-than-influencers
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, 37,325 respondents across 33 markets. Respondents selected trusted sources from a fixed list. One caveat on the labels: the release's summary names the 19% comparator 'travel bloggers' and its body text names the same figure 'colleagues'. We follow the summary, because Booking.com's Canadian edition of the same release reports travel bloggers and colleagues as separate options with different values (15% and 13%), which makes the body wording the looser of the two.
  - A channel that did not exist for travel planning five years ago now outranks both of the creator channels built specifically for it. The displaced incumbent is not search; it is the people who made a business of recommending places. For destinations and hotels, the budget implication is direct.
- **Travellers who discovered a destination via AI: 40%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/discovered-destination-via-ai
  - Method: Sojern traveller survey, April 2026, 1,006 US adults. Self-reported attribution, which tends to understate influence travellers did not consciously register.
  - Discovery and plan-change are the two measures that convert AI visibility into demand. If a model does not name your destination, four in ten travellers never encounter it at the moment of inspiration.
- **Travellers who changed plans based on AI recommendations: 29%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/changed-plans-due-to-ai
  - Method: Sojern traveller survey, April 2026, 1,006 US adults. Self-reported behaviour change.
  - Nearly three in ten report an AI answer altering a decision they had already formed. That is a behavioural outcome, not an attitude, and it is the strongest available evidence that AI visibility has commercial consequences.
- **Largest travel companies mentioning AI in annual reports: 35%**
  - Reference period: 2024 annual reports (2024)
  - Geography: Global
  - Evidence type: Analysis
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: Skift Travel 200, the largest publicly traded travel companies
  - For comparison: 2022 annual reports 4%
  - https://travel.rankwit.ai/data/skift200-ai-mentions
  - Method: Analysis of the Skift Travel 200, the largest publicly traded travel companies, in Remapping travel with agentic AI (McKinsey with Skift, September 2025). Counts any mention of AI, so it measures salience rather than deployment.
  - An eightfold rise in two years, measured in the documents companies are legally careful about rather than in marketing material. It is the cleanest available evidence that AI moved from experiment to disclosed strategy.
- **DMOs producing AI-formatted content: 64%**
  - Reference period: 2026 (2026)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/reports/state-of-destination-marketing-2026
  - Sample: 359 destination marketers worldwide, fielded with Dynata. Quoted from the publisher's summary; the full report is behind a download form
  - https://travel.rankwit.ai/data/dmos-creating-ai-formatted-content
  - Method: Sojern State of Destination Marketing 2026, 359 destination marketers worldwide, fielded in partnership with Dynata.
  - The clearest evidence that generative engine optimisation has moved from theory to budget line. Two-thirds of destination marketers have changed how they write, not just where they publish.
- **Travellers who would use an AI trip planner on a destination website: 63%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/would-use-ai-planner-on-destination-site
  - Method: Sojern paired surveys, April 2026: 1,006 US travellers and 103 DMO leaders, fielded concurrently.
  - A 41 point gap between stated demand and available supply, measured by one publisher in one month. Few product decisions in travel marketing have this clear an evidence base, though the demand half is US measured and the supply half is not.
- **Travellers less likely to visit a destination site without AI features: 25%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/less-likely-to-visit-site-without-ai
  - Method: Sojern traveller survey, April 2026, 1,006 US adults. Stated preference.
  - The downside measure to go with the 63% demand figure. A quarter of the audience treats the absence of AI tooling as a reason to go elsewhere.
- **Pre-trip AI use for destination research: 38%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Pre-trip use among AI users, 33 markets
  - https://travel.rankwit.ai/data/ai-task-destination-research
  - Method: Booking.com Global AI Sentiment Report, April–May 2025. Shares are of consumers who use AI in travel, not of all consumers.
  - Destination research is the top pre-trip use, which places AI at the upper funnel where destination and property choice is still open. This is the stage where being named matters most.
- **Travellers who received outdated or inaccurate AI travel information: 25%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/documents/resources/research-report/travel-trends-2026/amadeus-travel-trends-2026-report.pdf
  - Sample: Amadeus traveller data, 2025
  - https://travel.rankwit.ai/data/received-outdated-ai-travel-info
  - Method: Amadeus traveller data for 2025, reported in Travel Trends 2026. Self-reported, so it captures errors travellers noticed; the true rate is likely higher.
  - This is usually a content problem rather than a model problem. When the current facts about a property are not reachable in a structured form, a model answers from whatever it last saw, and stale confidence reads as authority.
- **Millennials using AI as often as social media for trip planning: 69%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: Millennial subset of 1,006 US adults
  - For comparison: Gen Z 63%
  - https://travel.rankwit.ai/data/millennials-ai-as-often-as-social
  - Method: Sojern traveller survey, April 2026, generational subsets of 1,006 US adults. Subsample sizes are not disclosed, so the generational splits carry wider uncertainty than the headline figure.
  - The comparison is the point. For most of the last decade social platforms were the default inspiration channel for younger travellers. For a clear majority of Millennials, AI now matches it.
- **Gen Z using AI as often as social media for trip planning: 63%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: Gen Z subset of 1,006 US adults
  - https://travel.rankwit.ai/data/genz-ai-as-often-as-social
  - Method: Sojern traveller survey, April 2026, Gen Z subset of 1,006 US adults. Subsample size not disclosed.
  - Gen Z indexing below Millennials is counterintuitive and worth noting. The likely explanation is that Gen Z's social-media baseline is higher, not that their AI use is lower.
- **DMOs offering AI trip planning on their website: 22%**
  - Reference period: April 2026 (2026-04)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 103 DMO leaders
  - https://travel.rankwit.ai/data/dmo-ai-trip-planning-on-site
  - Method: Sojern survey of 103 DMO leaders, April 2026, fielded alongside the traveller survey. Small sample, so treat as indicative of direction rather than precise.
  - Set against 54% of travellers who already plan with AI and 63% who say they would use a planner on a destination site, this is the sharpest supply-demand gap in the observatory.

### Trust and autonomy

Enthusiasm is near-universal; delegation is not. The gap between the two is the real story.

North America is the most sceptical region in the dataset: 32% rarely or never trust AI-generated information, against 15% in Latin America. An AI-forward product claim that lands in São Paulo can cost conversion in Chicago, and the US is the market where showing your sources pays for itself fastest.

- **Consumers who fully trust AI: 6%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - https://travel.rankwit.ai/data/full-trust-in-ai-global
  - Method: Booking.com Global AI Sentiment Report, 37,325 respondents across 33 markets, April–May 2025. The same survey also records 91% expressing at least one concern about AI's implications. The two 91% figures are separate findings from separate questions, not a transcription error: the report's own wording is that an equal proportion of the 91% who are excited also report at least one concern.
  - The single most important number in this dataset. Enthusiasm and trust are not the same variable and they are not even close. Any strategy built on the 91% and not the 6% will misjudge how much authority travellers will hand over.
- **Consumers comfortable with AI making independent decisions: 12%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - https://travel.rankwit.ai/data/comfort-ai-independent-decisions
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, 37,325 respondents across 33 markets.
  - This is the ceiling on fully autonomous travel agents today. Not a technical ceiling, a consent one. The capability can be shipped long before the permission arrives.
- **Travellers comfortable letting AI handle most of their planning: ~6%**
  - Reference period: March 2026 (2026-03)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/en/resources/white-paper/agentic-ai-airlines
  - Sample: Amadeus traveller research, March 2026
  - https://travel.rankwit.ai/data/comfort-ai-handles-most-planning
  - Method: Amadeus traveller research conducted March 2026, reported in the airlines agentic AI white paper. Both figures come from the same instrument.
  - From a single study, so the two figures are directly comparable: near-universal use, almost no delegation. The gap between 79% and 6% is the distance agentic travel still has to cover.
- **Travellers ready to let AI handle a booking: 25–33%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Phocuswright, https://www.phocuswright.com/Travel-Research/Research-Updates/2025/From-hype-to-habit-78-percent-of-travelers-say-GenAI-improves-trip-planning
  - Sample: Phocuswright, Chat, Plan, Book: GenAI Goes Mainstream; range across surveyed markets
  - https://travel.rankwit.ai/data/would-let-ai-complete-booking
  - Method: Phocuswright, Chat, Plan, Book: GenAI Goes Mainstream, 2025. Range reported across surveyed countries. The same study finds only about a third fully trust generative AI responses.
  - Published as a range across markets rather than a single number, and we keep it that way. It sits well above the 6% who would hand over planning entirely: travellers will approve a transaction they specified, not the specification itself.
- **Consumers who always fact-check AI outputs: 42%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - For comparison: sometimes fact-check 29%
  - https://travel.rankwit.ai/data/always-fact-check-ai
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, 37,325 respondents across 33 markets.
  - Seven in ten verify at least sometimes, which means the AI answer is usually the start of a research path rather than the end of one. It also means a wrong or outdated AI answer about your property is likely to be discovered, and attributed to you.
- **Consumers who trust travel brands to use AI responsibly: 42%**
  - Reference period: 2024 (2024)
  - Geography: Global
  - Evidence type: Survey
  - Source: Travelport, https://www.travelport.com/press-releases/travelport-releases-second-annual-state-of-modern-retailing-report-travels-tipping-point
  - Sample: State of Modern Retailing consumer survey
  - https://travel.rankwit.ai/data/trust-travel-brands-ai-responsible
  - Method: Travelport State of Modern Retailing, published January 2025. Cross-industry comparison within the same instrument.
  - A relative strength that the sector rarely claims. Travel starts ahead of most industries on AI trust, which makes transparency a defensible position rather than a cost.
- **Share of consumers classified as AI enthusiasts: 36%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - For comparison: detractors 25%
  - https://travel.rankwit.ai/data/ai-enthusiast-cohort-share
  - Method: Booking.com Global AI Sentiment Report, April–May 2025. Full split: Enthusiasts 36%, Advocates 13%, Cautious 13%, Sceptics 9%, Detractors 25%. The five published shares sum to 96%; the report does not say what the remaining 4% is, so the cohorts are not a complete partition of the sample.
  - The five-cohort split is more useful than any average. A quarter of the market is actively negative on AI, which is why visible AI features can cost conversion with some audiences even as they win others.
- **North Americans who rarely or never trust AI-generated information: 32%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: North America
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Regional breakdown of 37,325 respondents
  - For comparison: Latin America 15%
  - https://travel.rankwit.ai/data/ai-distrust-north-america
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, regional breakdown of 37,325 respondents across 33 markets. The report's regions are its own: the 29% comparison figure is reported for Europe and the Middle East together, not for Europe alone.
  - A two-to-one difference between the most and least sceptical regions. Markets are not interchangeable here: the same AI-forward positioning that works in São Paulo can underperform in Chicago.
- **Latin American respondents excited about AI: 98%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Latin America
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Regional breakdown of 37,325 respondents
  - For comparison: Asia Pacific 95%
  - https://travel.rankwit.ai/data/latam-ai-enthusiasm
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, regional breakdown. Excitement is a softer measure than trust and the two diverge sharply by region.
  - Latin America leads on enthusiasm and familiarity, Asia Pacific follows at 95% excited. Regional sequencing matters when choosing where to launch an AI-forward travel product.
- **Frequent travellers comfortable using AI for trip planning: 80%**
  - Reference period: 2024 (2024)
  - Geography: Global
  - Evidence type: Survey
  - Source: Travelport, https://www.travelport.com/press-releases/travelport-releases-second-annual-state-of-modern-retailing-report-travels-tipping-point
  - Sample: State of Modern Retailing consumer survey
  - https://travel.rankwit.ai/data/frequent-travellers-comfort-ai-planning
  - Method: Travelport's second annual State of Modern Retailing consumer survey, published January 2025. The frequent-traveller subset is self-defined, and the question measures comfort rather than use.
  - Comfort among high-frequency travellers runs well above general-population trust measures. The same report places travel in the top three industries for consumer trust in AI, a better starting position than the sector usually credits itself with.
- **Travellers who received outdated or inaccurate AI travel information: 25%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/documents/resources/research-report/travel-trends-2026/amadeus-travel-trends-2026-report.pdf
  - Sample: Amadeus traveller data, 2025
  - https://travel.rankwit.ai/data/received-outdated-ai-travel-info
  - Method: Amadeus traveller data for 2025, reported in Travel Trends 2026. Self-reported, so it captures errors travellers noticed; the true rate is likely higher.
  - This is usually a content problem rather than a model problem. When the current facts about a property are not reachable in a structured form, a model answers from whatever it last saw, and stale confidence reads as authority.

### What AI is used for

The specific jobs travellers hand to an assistant, before, during and after a trip.

US task data comes from Sojern's April 2026 panel, which asked travellers and destination organisations the same questions in the same month. That pairing is rare and makes the mismatch between what travellers do and what the industry prioritises directly measurable.

- **AI use for researching attractions, restaurants and transport: 60%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: AI users among 1,006 US adults. The source's scope is attractions, restaurants and transportation
  - https://travel.rankwit.ai/data/ai-task-attractions-and-restaurants
  - Method: Sojern traveller survey, April 2026, AI users among 1,006 US adults. The published wording is 'sixty percent use it to research attractions, restaurants, and transportation', so the figure covers all three and should not be quoted as an attractions-and-restaurants number.
  - The highest task-level figure in the dataset, and the broadest task in it: the question bundles three things travellers used to look up separately. Attraction, restaurant and transport discovery is where AI has most comprehensively replaced browsing, and where a missing or wrong listing is most costly.
- **AI use for comparing prices: 56%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: AI users among 1,006 US adults
  - https://travel.rankwit.ai/data/ai-task-price-comparison
  - Method: Sojern traveller survey, April 2026, AI users among 1,006 US adults.
  - The 24-point gap between general price comparison and flight-and-hotel comparison marks the current boundary of trust. Travellers use AI to understand what things cost, then switch to a booking surface to transact.
- **In-trip AI use for translation: 45%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: In-trip use among AI users, 33 markets
  - For comparison: navigation 40%
  - https://travel.rankwit.ai/data/ai-task-in-trip-translation
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, among AI users in travel. 96% of AI users reported using it while travelling.
  - In-trip use is higher than pre-trip use for most tasks, which is easy to miss. The traveller who arrives having planned without AI will often still use it on the ground.
- **AI use for building and organising itineraries: 43%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: AI users among 1,006 US adults
  - https://travel.rankwit.ai/data/ai-task-itinerary-building
  - Method: Sojern paired surveys, April 2026. The traveller and DMO figures were fielded concurrently and published together, but they are two questionnaires: 1,006 US adults on one side, 103 destination leaders worldwide on the other.
  - Itinerary building is the task destination organisations most underestimate: only 20% of DMO leaders rank it as high-value, while four in ten travellers already do it with AI.
- **Pre-trip AI use for destination research: 38%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Pre-trip use among AI users, 33 markets
  - https://travel.rankwit.ai/data/ai-task-destination-research
  - Method: Booking.com Global AI Sentiment Report, April–May 2025. Shares are of consumers who use AI in travel, not of all consumers.
  - Destination research is the top pre-trip use, which places AI at the upper funnel where destination and property choice is still open. This is the stage where being named matters most.
- **Pre-trip AI use for local experiences: 37%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Pre-trip use among AI users, 33 markets
  - https://travel.rankwit.ai/data/ai-task-local-experiences
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, among AI users in travel.
  - Experiences are a fragmented, long-tail category where models have a real advantage over directory sites, and where small operators have the most to gain from being described accurately.
- **Pre-trip AI use for restaurant recommendations: 36%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: Pre-trip use among AI users, 33 markets
  - https://travel.rankwit.ai/data/ai-task-restaurant-recommendations
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, among AI users in travel.
  - Restaurants show the staleness problem at its sharpest: opening hours, closures and menu changes move faster than any model's training data, and this is one of the clearest cases for structured, current, machine-readable facts.

### Destinations and DMOs

National tourism organisations and destination marketers, measured against the travellers they serve.

This is the best-documented case in the section, because Sojern surveyed 103 DMO leaders alongside 1,006 US travellers in the same month. Read it carefully from a US desk: the traveller side is an American panel, while the DMO side is not broken out by country. The 32-point gap is two measurements taken at once rather than an estimate; it is simply not a US-only gap.

- **DMOs offering AI trip planning on their website: 22%**
  - Reference period: April 2026 (2026-04)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 103 DMO leaders
  - https://travel.rankwit.ai/data/dmo-ai-trip-planning-on-site
  - Method: Sojern survey of 103 DMO leaders, April 2026, fielded alongside the traveller survey. Small sample, so treat as indicative of direction rather than precise.
  - Set against 54% of travellers who already plan with AI and 63% who say they would use a planner on a destination site, this is the sharpest supply-demand gap in the observatory.
- **DMOs with a documented AI strategy: 26%**
  - Reference period: April 2026 (2026-04)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 103 DMO leaders
  - https://travel.rankwit.ai/data/dmo-documented-ai-strategy
  - Method: Sojern survey of 103 DMO leaders, April 2026. 'Documented' is self-assessed. The 64% content figure comes from Sojern's State of Destination Marketing 2026 and its 359-marketer panel, which is why the two are not expressed as shares of one another.
  - Three-quarters are operating without one. A separate and larger Sojern study (359 destination marketers, not these 103 leaders) finds 64% already producing AI-formatted content. Two samples rather than one, so this is not '64% of them', but the direction holds either way: activity is running ahead of strategy, which is usually how budget gets spent twice.
- **DMO leaders who see irrelevance risk within two years: 72%**
  - Reference period: April 2026 (2026-04)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 103 DMO leaders
  - https://travel.rankwit.ai/data/dmo-irrelevance-risk-two-years
  - Method: Sojern survey of 103 DMO leaders, April 2026. Attitudinal measure.
  - Awareness is not the constraint. Nearly three-quarters name the risk, a quarter have written a strategy, and a fifth have shipped anything. The gap is execution capacity.
- **European NTOs with dedicated AI teams or AI monitoring: ~40%**
  - Reference period: March–April 2025 (2025-03)
  - Geography: Europe
  - Evidence type: Survey
  - Source: European Travel Commission, https://etc-corporate.org/uploads/2025/09/2025_AI-in-tourism_Supporting-NTO-Operations.pdf
  - Sample: 29 of 36 ETC member national tourism organisations; the report's wording is "around 40%"
  - https://travel.rankwit.ai/data/nto-ai-adoption-europe
  - Method: European Travel Commission, Artificial Intelligence in Tourism: Assessing and Supporting NTOs' Research & Marketing Operations, Brussels, July 2025. The report's own wording is «around 40%», which we do not sharpen. Its adoption-stage chart uses a separate seven-point scale.
  - Note the precise claim: having a team or watching the field is not the same as running AI in operations, and an earlier version of this page conflated the two. Even on the looser reading, three in five national tourism bodies have neither, against travellers who already plan with AI in the majority.
- **DMOs producing AI-formatted content: 64%**
  - Reference period: 2026 (2026)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/reports/state-of-destination-marketing-2026
  - Sample: 359 destination marketers worldwide, fielded with Dynata. Quoted from the publisher's summary; the full report is behind a download form
  - https://travel.rankwit.ai/data/dmos-creating-ai-formatted-content
  - Method: Sojern State of Destination Marketing 2026, 359 destination marketers worldwide, fielded in partnership with Dynata.
  - The clearest evidence that generative engine optimisation has moved from theory to budget line. Two-thirds of destination marketers have changed how they write, not just where they publish.
- **Travellers who would use an AI trip planner on a destination website: 63%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/would-use-ai-planner-on-destination-site
  - Method: Sojern paired surveys, April 2026: 1,006 US travellers and 103 DMO leaders, fielded concurrently.
  - A 41 point gap between stated demand and available supply, measured by one publisher in one month. Few product decisions in travel marketing have this clear an evidence base, though the demand half is US measured and the supply half is not.
- **Travellers less likely to visit a destination site without AI features: 25%**
  - Reference period: April 2026 (2026-04)
  - Geography: United States
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/press-release/new-research-reveals-travellers-are-two-years-ahead-of-the-travel-industry-on-ai
  - Sample: 1,006 US adults
  - https://travel.rankwit.ai/data/less-likely-to-visit-site-without-ai
  - Method: Sojern traveller survey, April 2026, 1,006 US adults. Stated preference.
  - The downside measure to go with the 63% demand figure. A quarter of the audience treats the absence of AI tooling as a reason to go elsewhere.
- **DMOs increasing digital marketing spend: 41%**
  - Reference period: 2026 (2026)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/reports/state-of-destination-marketing-2026
  - Sample: 359 destination marketers worldwide
  - For comparison: cut spend 13%
  - https://travel.rankwit.ai/data/dmo-digital-spend-increase
  - Method: Sojern State of Destination Marketing 2026, 359 destination marketers worldwide, fielded with Dynata. Remaining share did not report.
  - A flat aggregate hiding a reallocation. With 64% producing AI-formatted content on budgets that mostly did not grow, the money is moving inside the line rather than being added to it.
- **OECD priority areas for responsible AI adoption in tourism: 6**
  - Reference period: September 2026 (2026-09)
  - Geography: APEC economies
  - Evidence type: Analysis
  - Source: OECD, https://www.oecd.org/en/publications/artificial-intelligence-and-tourism-in-apec-economies_4549486a-en.html
  - Sample: OECD policy paper on AI and tourism in APEC economies
  - https://travel.rankwit.ai/data/oecd-apec-priority-areas
  - Method: OECD policy paper 'Artificial Intelligence and tourism in APEC economies', published 9 September 2026. Qualitative policy analysis rather than a survey.
  - Policy work has moved from describing AI to setting conditions for its adoption. The OECD's framing centres on SME access, because tourism is SME-dominated and the benefits otherwise concentrate among operators with capital to invest.

### Hotels and accommodation

Adoption, outlook and the capability gap between chains and independents.

US hotel performance reporting is unusually granular, but the AI adoption figures here are global. Read the chain-versus-independent split as the structural story: it holds in every market measured, and the US independent segment is large enough that it dominates the national average.

- **Tourism businesses already using AI: 70%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: UN Tourism, https://doi.org/10.18111/9789284426065
  - Sample: UN Tourism survey with Saxion University of Applied Sciences
  - https://travel.rankwit.ai/data/tourism-businesses-using-ai
  - Method: UN Tourism report produced with Saxion University of Applied Sciences, 2025. Global survey of tourism businesses; the report does not disclose sample size, which is why we treat the level as indicative. The publisher serves the document behind an access check, so we have verified the DOI and the report, not the figure inside it.
  - The highest-authority adoption figure available, from UN Tourism. Note what it counts: operational AI inside the business, not visibility in the engines travellers use to find it. A hotel can be in the 70% and still be invisible.
- **Projected growth in tourism AI adoption over five years: +25%**
  - Reference period: 2025–2030 (2025)
  - Geography: Global
  - Evidence type: Forecast
  - Source: UN Tourism, https://doi.org/10.18111/9789284426065
  - Sample: Five-year projection in the UN Tourism report
  - https://travel.rankwit.ai/data/tourism-ai-adoption-growth-5y
  - Method: UN Tourism, 2025. Five-year projection; the underlying model and assumptions are not published in the summary, so confidence is medium.
  - A projection, not a measurement, and shown as such. From a base of 70%, a 25% growth rate implies adoption approaching saturation, at which point the differentiator stops being whether you use AI and becomes whether AI uses you.
- **European accommodations expecting positive development: 66%**
  - Reference period: February–March 2026 (2026-02)
  - Geography: Europe
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/2026-european-accommodation-barometer-challenges-concerns-and-diverging-outlooks/
  - Sample: European accommodation providers. The 72%/55% split answers a separate question from the 66% outlook
  - For comparison: chains vs independents, current economic situation 72% / 55%
  - https://travel.rankwit.ai/data/eu-accommodation-positive-outlook
  - Method: Booking.com European Accommodation Barometer, fielded February–March 2026 across European accommodation providers. The 66% measures expectations for the coming season; the 72% and 55% measure how chains and independents rate their current situation. An earlier version of this page presented them as a single comparison.
  - Two questions, not one comparison: the 66% is about the season ahead and applies to the whole sample, while the 72%/55% split is about current trading. What the split does establish is a seventeen-point gap by ownership structure, and the same ownership pattern turns up in technology readiness and cyber preparedness (the mechanism rather than a coincidence).
- **Cybersecurity readiness gap by property size: 94%**
  - Reference period: February–March 2026 (2026-02)
  - Geography: Europe
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/2026-european-accommodation-barometer-challenges-concerns-and-diverging-outlooks/
  - Sample: European accommodation providers by size band
  - For comparison: properties with under 10 staff 60%
  - https://travel.rankwit.ai/data/accommodation-cyber-readiness-gap
  - Method: Booking.com European Accommodation Barometer, February–March 2026. Self-assessed preparedness, which typically overstates readiness at both ends.
  - A 34-point capability gap on a dimension that agentic systems make more consequential, since an agent that transacts needs an exposed, authenticated surface to transact against.

### Airlines, agencies and distribution

Where agentic AI is already in production, and the data fragmentation holding it back.

The executive-level figures in this section are mostly US-based samples: McKinsey's 86 travel executives, Skift's panels. They describe the market where agentic distribution is being built first, and where the fragmentation that slows it is best documented.

- **Travel agencies adopting AI: 80%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (14 countries)
  - Evidence type: Survey
  - Source: Sabre, https://www.sabre.com/resources/research/content-fragmentation-what-i-see-vs-what-i-take/
  - Sample: Travel agencies across 14 countries
  - https://travel.rankwit.ai/data/travel-agencies-adopting-ai
  - Method: Sabre Content Fragmentation study, surveying travel agencies across 14 countries, April–May 2025. The report's own wording is 'nearly 80% of agencies already implementing AI solutions, and the remaining 20% outlining plans to do so', so the 80% is adoption rather than adoption-or-intent, which is also the denominator the productivity figure is taken from.
  - Implemented plus intending accounts for the entire sample, which is why this reads as effectively universal. Agencies are further along than destination organisations, largely because their AI use case is internal productivity rather than public-facing visibility.
- **AI-adopting agencies reporting measurable productivity gains: 40%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (14 countries)
  - Evidence type: Survey
  - Source: Sabre, https://www.sabre.com/resources/research/content-fragmentation-what-i-see-vs-what-i-take/
  - Sample: AI-adopting agencies across 14 countries
  - https://travel.rankwit.ai/data/agency-ai-productivity-gain
  - Method: Sabre Content Fragmentation study, 14 countries, April–May 2025, among the 80% of agencies that had adopted AI.
  - The most honest number in the vendor research. Adoption is near-universal; measurable return is a minority outcome. The study attributes the difference largely to data fragmentation rather than to the models.
- **Agencies working across four or more booking systems: 91%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (14 countries)
  - Evidence type: Survey
  - Source: Sabre, https://www.sabre.com/resources/research/content-fragmentation-what-i-see-vs-what-i-take/
  - Sample: Travel agencies across 14 countries
  - For comparison: seven or more systems 50%
  - https://travel.rankwit.ai/data/agencies-four-plus-booking-systems
  - Method: Sabre Content Fragmentation study, 14 countries, April–May 2025. Nearly 90% also manage four or more API integrations.
  - This is the structural reason agentic travel is harder than it looks. An agent reasoning about availability and price across ten disconnected systems will be wrong often enough to lose trust, and trust is the scarce input.
- **Executives reporting AI agents in production: 52%**
  - Reference period: September 2025 (2025-09)
  - Geography: Global (24 countries)
  - Evidence type: Survey
  - Source: Google, https://services.google.com/fh/files/misc/google_cloud_roi_of_ai_2025.pdf
  - Sample: 3,466 senior leaders across 24 countries at enterprises above USD 10m revenue that already deploy generative AI, fielded April–June 2025. The figure is a share of gen-AI-adopting enterprises, not of all enterprises
  - https://travel.rankwit.ai/data/executives-ai-agents-in-production
  - Method: Google Cloud's ROI of AI study, September 2025: 3,466 senior leaders across 24 countries at enterprises above USD 10m revenue, fielded April–June 2025. Cross-industry sample rather than travel-specific, which is the main caveat on reading it next to travel's own figures.
  - Cross-industry rather than travel-specific, which is why it sits above travel's own figures. It sets the benchmark the sector is being measured against by its own vendors.
- **US travel executives reporting AI improves personalisation: 33%**
  - Reference period: September 2025 (2025-09)
  - Geography: United States
  - Evidence type: Survey
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: 86 mostly US-based travel executives, surveyed for Remapping travel with agentic AI (McKinsey with Skift)
  - For comparison: higher-quality outputs 36%
  - https://travel.rankwit.ai/data/exec-ai-improves-personalisation
  - Method: McKinsey with Skift, Remapping travel with agentic AI, September 2025. Sample of 86 mostly US-based travel executives, small enough that the levels are directional rather than precise.
  - Executive-reported benefit is far below executive-reported adoption, the same pattern the agency data shows. Roughly a third of deployments are producing an effect the business can name.
- **Travellers who have used AI for trip planning or management: ~79%**
  - Reference period: March 2026 (2026-03)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/en/resources/white-paper/agentic-ai-airlines
  - Sample: Amadeus traveller research, March 2026
  - https://travel.rankwit.ai/data/ai-trip-planning-amadeus-2026
  - Method: Amadeus traveller research conducted March 2026, reported in the airlines agentic AI white paper. The same study found only around 6% comfortable letting AI handle most of their planning.
  - The upper bound of the published range, and from the same publisher that reports 18% for generative AI planning a year earlier. The difference is the question, not the population: this one counts planning and in-trip management, ever, with any AI tool.
- **Largest travel companies mentioning AI in annual reports: 35%**
  - Reference period: 2024 annual reports (2024)
  - Geography: Global
  - Evidence type: Analysis
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: Skift Travel 200, the largest publicly traded travel companies
  - For comparison: 2022 annual reports 4%
  - https://travel.rankwit.ai/data/skift200-ai-mentions
  - Method: Analysis of the Skift Travel 200, the largest publicly traded travel companies, in Remapping travel with agentic AI (McKinsey with Skift, September 2025). Counts any mention of AI, so it measures salience rather than deployment.
  - An eightfold rise in two years, measured in the documents companies are legally careful about rather than in marketing material. It is the cleanest available evidence that AI moved from experiment to disclosed strategy.
- **Share of travel venture funding going to AI-enabled startups: 45%**
  - Reference period: First half of 2025 (2025)
  - Geography: Global
  - Evidence type: Analysis
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: Skift venture capital tracking of travel-industry funding
  - For comparison: 2023 10%
  - https://travel.rankwit.ai/data/ai-travel-vc-funding-share
  - Method: Skift venture capital tracking, reported in Remapping travel with agentic AI (McKinsey with Skift, September 2025). Share of disclosed travel-industry venture funding.
  - Capital allocation is a leading indicator of what the distribution layer will look like in three years. Nearly half of new travel technology funding is now going to companies whose premise is AI.

### Economics

What the shift is worth, where the money moves, and which claims are forecasts rather than measurements.

Venture funding is the clearest US signal here: 45% of travel venture capital went to AI-enabled startups in the first half of 2025. That capital is overwhelmingly American, and it is a three-year leading indicator of what the distribution layer will look like.

- **Travel spending growth attributed to AI-unlocked potential: USD 1.4tn**
  - Reference period: Five years from 2025 (2025)
  - Geography: Global
  - Evidence type: Forecast
  - Source: Euromonitor International, https://www.euromonitor.com/article/ai-a-turning-point-in-global-travel
  - Sample: Euromonitor five-year travel spending forecast
  - https://travel.rankwit.ai/data/ai-unlocked-travel-spend-growth
  - Method: Euromonitor five-year travel spending forecast, published 12 November 2025. The attribution of growth to AI is the analyst's judgement; the decomposition is not published.
  - A forecast, and the most-quoted figure in the field. Treat it as a directional claim about where a research house expects growth to come from, not as a measurement of value created.
- **Share of travel venture funding going to AI-enabled startups: 45%**
  - Reference period: First half of 2025 (2025)
  - Geography: Global
  - Evidence type: Analysis
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: Skift venture capital tracking of travel-industry funding
  - For comparison: 2023 10%
  - https://travel.rankwit.ai/data/ai-travel-vc-funding-share
  - Method: Skift venture capital tracking, reported in Remapping travel with agentic AI (McKinsey with Skift, September 2025). Share of disclosed travel-industry venture funding.
  - Capital allocation is a leading indicator of what the distribution layer will look like in three years. Nearly half of new travel technology funding is now going to companies whose premise is AI.
- **EU guest nights booked via Airbnb, Booking and Expedia: 951.6m**
  - Reference period: 2025 (2025)
  - Geography: European Union
  - Evidence type: Official statistic
  - Source: Eurostat, https://ec.europa.eu/eurostat/databrowser/view/tour_ce_omr/default/table
  - Sample: Dataset tour_ce_omr: experimental statistics on nights booked via Airbnb, Booking and Expedia
  - https://travel.rankwit.ai/data/eu-platform-guest-nights
  - Method: Eurostat dataset tour_ce_omr, collected under the short-term rental data-sharing regulation. Flagged experimental by Eurostat: methods are still being consolidated and figures may be revised. On the denominator: this is a full-year 2025 figure and the comparable EU full-year total is close to 3 billion nights, which is where the one-third share comes from. The 1.80 billion figure elsewhere in this dataset covers January to July 2026 only and is not the denominator for it.
  - Roughly a third of the EU accommodation total runs through three platforms (951.6 million nights against a full-year EU total of about 3 billion) under a data-sharing regulation that makes it countable for the first time. It establishes how concentrated the distribution layer already was before AI entered it.
- **AI-adopting agencies reporting measurable productivity gains: 40%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (14 countries)
  - Evidence type: Survey
  - Source: Sabre, https://www.sabre.com/resources/research/content-fragmentation-what-i-see-vs-what-i-take/
  - Sample: AI-adopting agencies across 14 countries
  - https://travel.rankwit.ai/data/agency-ai-productivity-gain
  - Method: Sabre Content Fragmentation study, 14 countries, April–May 2025, among the 80% of agencies that had adopted AI.
  - The most honest number in the vendor research. Adoption is near-universal; measurable return is a minority outcome. The study attributes the difference largely to data fragmentation rather than to the models.
- **DMOs increasing digital marketing spend: 41%**
  - Reference period: 2026 (2026)
  - Geography: Global
  - Evidence type: Survey
  - Source: Sojern, https://www.sojern.com/reports/state-of-destination-marketing-2026
  - Sample: 359 destination marketers worldwide
  - For comparison: cut spend 13%
  - https://travel.rankwit.ai/data/dmo-digital-spend-increase
  - Method: Sojern State of Destination Marketing 2026, 359 destination marketers worldwide, fielded with Dynata. Remaining share did not report.
  - A flat aggregate hiding a reallocation. With 64% producing AI-formatted content on budgets that mostly did not grow, the money is moving inside the line rather than being added to it.
- **Projected growth in tourism AI adoption over five years: +25%**
  - Reference period: 2025–2030 (2025)
  - Geography: Global
  - Evidence type: Forecast
  - Source: UN Tourism, https://doi.org/10.18111/9789284426065
  - Sample: Five-year projection in the UN Tourism report
  - https://travel.rankwit.ai/data/tourism-ai-adoption-growth-5y
  - Method: UN Tourism, 2025. Five-year projection; the underlying model and assumptions are not published in the summary, so confidence is medium.
  - A projection, not a measurement, and shown as such. From a base of 70%, a 25% growth rate implies adoption approaching saturation, at which point the differentiator stops being whether you use AI and becomes whether AI uses you.
- **US travel executives reporting AI improves personalisation: 33%**
  - Reference period: September 2025 (2025-09)
  - Geography: United States
  - Evidence type: Survey
  - Source: McKinsey & Company, https://www.mckinsey.com/industries/travel/our-insights/remapping-travel-with-agentic-ai
  - Sample: 86 mostly US-based travel executives, surveyed for Remapping travel with agentic AI (McKinsey with Skift)
  - For comparison: higher-quality outputs 36%
  - https://travel.rankwit.ai/data/exec-ai-improves-personalisation
  - Method: McKinsey with Skift, Remapping travel with agentic AI, September 2025. Sample of 86 mostly US-based travel executives, small enough that the levels are directional rather than precise.
  - Executive-reported benefit is far below executive-reported adoption, the same pattern the agency data shows. Roughly a third of deployments are producing an effect the business can name.

### Governance, accuracy and risk

What goes wrong: stale answers, unverified claims, and the policy response taking shape.

The US has no federal AI statute covering travel, so the governing pressure comes from accuracy and liability rather than regulation. That makes the 25% who received wrong AI travel information the operative number: in this market the exposure is commercial before it is legal.

- **Travellers who received outdated or inaccurate AI travel information: 25%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: Amadeus, https://amadeus.com/documents/resources/research-report/travel-trends-2026/amadeus-travel-trends-2026-report.pdf
  - Sample: Amadeus traveller data, 2025
  - https://travel.rankwit.ai/data/received-outdated-ai-travel-info
  - Method: Amadeus traveller data for 2025, reported in Travel Trends 2026. Self-reported, so it captures errors travellers noticed; the true rate is likely higher.
  - This is usually a content problem rather than a model problem. When the current facts about a property are not reachable in a structured form, a model answers from whatever it last saw, and stale confidence reads as authority.
- **Consumers who always fact-check AI outputs: 42%**
  - Reference period: April–May 2025 (2025-04)
  - Geography: Global (33 markets)
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/bookingcom-releases-the-global-ai-sentiment-report
  - Sample: 37,325 respondents across 33 markets
  - For comparison: sometimes fact-check 29%
  - https://travel.rankwit.ai/data/always-fact-check-ai
  - Method: Booking.com Global AI Sentiment Report, April–May 2025, 37,325 respondents across 33 markets.
  - Seven in ten verify at least sometimes, which means the AI answer is usually the start of a research path rather than the end of one. It also means a wrong or outdated AI answer about your property is likely to be discovered, and attributed to you.
- **OECD priority areas for responsible AI adoption in tourism: 6**
  - Reference period: September 2026 (2026-09)
  - Geography: APEC economies
  - Evidence type: Analysis
  - Source: OECD, https://www.oecd.org/en/publications/artificial-intelligence-and-tourism-in-apec-economies_4549486a-en.html
  - Sample: OECD policy paper on AI and tourism in APEC economies
  - https://travel.rankwit.ai/data/oecd-apec-priority-areas
  - Method: OECD policy paper 'Artificial Intelligence and tourism in APEC economies', published 9 September 2026. Qualitative policy analysis rather than a survey.
  - Policy work has moved from describing AI to setting conditions for its adoption. The OECD's framing centres on SME access, because tourism is SME-dominated and the benefits otherwise concentrate among operators with capital to invest.
- **Tourism businesses already using AI: 70%**
  - Reference period: 2025 (2025)
  - Geography: Global
  - Evidence type: Survey
  - Source: UN Tourism, https://doi.org/10.18111/9789284426065
  - Sample: UN Tourism survey with Saxion University of Applied Sciences
  - https://travel.rankwit.ai/data/tourism-businesses-using-ai
  - Method: UN Tourism report produced with Saxion University of Applied Sciences, 2025. Global survey of tourism businesses; the report does not disclose sample size, which is why we treat the level as indicative. The publisher serves the document behind an access check, so we have verified the DOI and the report, not the figure inside it.
  - The highest-authority adoption figure available, from UN Tourism. Note what it counts: operational AI inside the business, not visibility in the engines travellers use to find it. A hotel can be in the 70% and still be invisible.
- **Cybersecurity readiness gap by property size: 94%**
  - Reference period: February–March 2026 (2026-02)
  - Geography: Europe
  - Evidence type: Survey
  - Source: Booking.com, https://news.booking.com/2026-european-accommodation-barometer-challenges-concerns-and-diverging-outlooks/
  - Sample: European accommodation providers by size band
  - For comparison: properties with under 10 staff 60%
  - https://travel.rankwit.ai/data/accommodation-cyber-readiness-gap
  - Method: Booking.com European Accommodation Barometer, February–March 2026. Self-assessed preparedness, which typically overstates readiness at both ends.
  - A 34-point capability gap on a dimension that agentic systems make more consequential, since an agent that transacts needs an exposed, authenticated surface to transact against.
- **Executives reporting AI agents in production: 52%**
  - Reference period: September 2025 (2025-09)
  - Geography: Global (24 countries)
  - Evidence type: Survey
  - Source: Google, https://services.google.com/fh/files/misc/google_cloud_roi_of_ai_2025.pdf
  - Sample: 3,466 senior leaders across 24 countries at enterprises above USD 10m revenue that already deploy generative AI, fielded April–June 2025. The figure is a share of gen-AI-adopting enterprises, not of all enterprises
  - https://travel.rankwit.ai/data/executives-ai-agents-in-production
  - Method: Google Cloud's ROI of AI study, September 2025: 3,466 senior leaders across 24 countries at enterprises above USD 10m revenue, fielded April–June 2025. Cross-industry sample rather than travel-specific, which is the main caveat on reading it next to travel's own figures.
  - Cross-industry rather than travel-specific, which is why it sits above travel's own figures. It sets the benchmark the sector is being measured against by its own vendors.

### Market context

The baseline volumes that AI-era percentages are percentages of.

The denominators here are European and global, not American, because the EU publishes monthly accommodation statistics and the US does not. Use them for scale; use the US panels above for behaviour.

- **Nights spent in EU tourist accommodation, year to date: 1.80bn**
  - Reference period: 2026 year to date, through July (2026-07)
  - Geography: European Union
  - Evidence type: Official statistic
  - Source: Eurostat, https://ec.europa.eu/eurostat/databrowser/view/tour_occ_nim/default/table
  - Sample: Dataset tour_occ_nim: nights spent at tourist accommodation, monthly
  - Year on year: +1.93%
  - https://travel.rankwit.ai/data/eu-nights-tourist-accommodation
  - Method: Eurostat dataset tour_occ_nim, nights spent at tourist accommodation, monthly, through July 2026. Official statistic subject to Eurostat's revision policy.
  - The denominator. Every percentage in this observatory is a share of a market operating at this scale, and the market itself is still growing modestly.
- **EU guest nights booked via Airbnb, Booking and Expedia: 951.6m**
  - Reference period: 2025 (2025)
  - Geography: European Union
  - Evidence type: Official statistic
  - Source: Eurostat, https://ec.europa.eu/eurostat/databrowser/view/tour_ce_omr/default/table
  - Sample: Dataset tour_ce_omr: experimental statistics on nights booked via Airbnb, Booking and Expedia
  - https://travel.rankwit.ai/data/eu-platform-guest-nights
  - Method: Eurostat dataset tour_ce_omr, collected under the short-term rental data-sharing regulation. Flagged experimental by Eurostat: methods are still being consolidated and figures may be revised. On the denominator: this is a full-year 2025 figure and the comparable EU full-year total is close to 3 billion nights, which is where the one-third share comes from. The 1.80 billion figure elsewhere in this dataset covers January to July 2026 only and is not the denominator for it.
  - Roughly a third of the EU accommodation total runs through three platforms (951.6 million nights against a full-year EU total of about 3 billion) under a data-sharing regulation that makes it countable for the first time. It establishes how concentrated the distribution layer already was before AI entered it.
- **Spain, international tourist arrivals: 93.8m**
  - Reference period: 2024 (2024)
  - Geography: Spain
  - Evidence type: Official statistic
  - Source: UN Tourism, https://ourworldindata.org/grapher/international-tourist-trips
  - Sample: UN Tourism international tourist arrivals, via Our World in Data. France has no 2024 entry in this series
  - Year on year: +10.09%
  - https://travel.rankwit.ai/data/spain-international-arrivals
  - Method: UN Tourism international tourist arrivals, accessed via Our World in Data, 2024. Arrivals count trips, not people: one traveller making three trips counts three times. France is absent from the 2024 cut of this dataset, so any 'most-visited' ranking derived from it alone is incomplete.
  - A worked example of why a denominator needs reading before it is quoted. Ranked on the Our World in Data compilation alone, Spain tops 2024, only because France's 2024 value is missing from it, the latest French entry being 2021. The published UN Tourism ranking puts France first, Spain second. We show Spain's figure because we can trace it; we show the ranking caveat because quoting it without one would be wrong.
- **Italy international tourist arrivals: 74.0m**
  - Reference period: 2024 (2024)
  - Geography: Italy
  - Evidence type: Official statistic
  - Source: OECD, https://data-explorer.oecd.org/vis?df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_TOURISM_INTER@DF_INBOUND&df[ag]=OECD.CFE.TOU&df[vs]=1.1
  - Sample: OECD inbound tourism, international arrivals
  - Year on year: +8.9%
  - https://travel.rankwit.ai/data/italy-international-arrivals
  - Method: OECD inbound tourism statistics, dataset DSD_TOURISM_INTER@DF_INBOUND, 2024. Harmonised to the OECD definition of international arrivals, which can differ from national series.
  - Italy is the reference market for this observatory's Italian edition and the home market of its publisher. Italia.it, the national tourism portal, is also the sourcing model this project follows.
- **Claude usage index, Australia: 6.40×**
  - Reference period: June 2026 (2026-06)
  - Geography: Australia
  - Evidence type: Platform telemetry
  - Source: Anthropic, https://www.anthropic.com/economic-index
  - Sample: Privacy-preserving aggregate usage across 121 countries
  - For comparison: baseline expectation 1.00×
  - https://travel.rankwit.ai/data/claude-usage-index-australia
  - Method: Anthropic Economic Index, June 2026. Privacy-preserving aggregate statistics across 121 countries; the index is usage share divided by working-age population share. Measures one assistant, not all AI use.
  - Rare first-party telemetry rather than a survey, and a useful corrective: AI usage intensity does not track market size. For travel brands, high-index markets are where AI-mediated discovery is already material.

## Glossary

### Generative Engine Optimisation (GEO)

The practice of making a brand's information retrievable, accurate and citable inside AI-generated answers.

GEO targets the answer rather than the result list. Where SEO competes for a position on a page of ten links, GEO competes to be one of the three or four entities an assistant names. The working inputs are different too: structured and current facts, consistent description across the sources a model already trusts, and content shaped as answers rather than pages. In travel this matters because 40% of travellers report discovering a destination through AI.

https://travel.rankwit.ai/glossary/geo

### Answer Engine Optimisation (AEO)

Structuring content so it can be lifted directly as the answer to a specific question.

AEO is the narrower, more mechanical half of GEO. It covers question-shaped headings, short direct answers placed before the elaboration, FAQ and QAPage markup, and unambiguous attribution of every figure. The test is simple: could a machine extract a correct, self-contained answer from this page without reading the rest of it?

https://travel.rankwit.ai/glossary/aeo

### Large Language Model Optimisation (LLMO)

An umbrella term for influencing how language models represent a brand, used interchangeably with GEO.

LLMO is used more in vendor marketing than in practice, and generally means the same thing as GEO. Where a distinction is drawn, LLMO leans towards how a model represents an entity in its parameters and retrieval index, while GEO leans towards the generated answer. The operational work overlaps almost entirely.

https://travel.rankwit.ai/glossary/llmo

### Search Engine Optimisation (SEO)

Making pages discoverable and competitive in a ranked list of results.

SEO has not been replaced, and treating GEO as its successor is a mistake. Models still ground answers in crawled, indexed content, so technical health, crawlability and authority remain inputs. What changed is the surface: a results page shows ten options and an AI answer names three. Positions eleven through a hundred did not have much value before, but they existed.

https://travel.rankwit.ai/glossary/seo

### AI Share of Voice

The proportion of AI answers to a defined set of prompts in which a brand is named.

Measured by running a fixed prompt set across engines on a schedule and recording which brands appear. It is the closest equivalent to a ranking in generative search, with two differences: it is probabilistic rather than positional, and it varies by engine and by week. A single spot check tells you almost nothing; a tracked series tells you whether you are gaining or losing ground.

https://travel.rankwit.ai/glossary/ai-share-of-voice

### Citation

A source an AI answer explicitly credits or links as the basis for a claim.

Citations are the visible part of grounding, and in travel they are commercially consequential: being the cited source for 'best boutique hotels in Florence' places a brand inside the decision rather than adjacent to it. Which sources get cited varies by engine, and cited sources are frequently third-party (directories, reviews, destination sites) rather than the brand's own.

https://travel.rankwit.ai/glossary/citation

### Grounding

Anchoring a model's answer in retrieved documents rather than in parameters alone.

A grounded answer is generated with source documents in context. This is why making current facts reachable matters more than any wording trick: a model that cannot retrieve your present-day information answers from whatever it absorbed during training, which in travel is often a year or more stale.

https://travel.rankwit.ai/glossary/grounding

### Retrieval-Augmented Generation (RAG)

Fetching relevant documents at query time and generating the answer from them.

RAG is the architecture behind most grounded AI answers, including AI search. The practical consequence for publishers is that retrievability is a prerequisite for citation: content that is not indexed, not parseable or not current cannot enter the context window, and content that never enters the context window is never cited.

https://travel.rankwit.ai/glossary/rag

### Hallucination

A confident, fluent AI statement that is not supported by any source.

In travel, most apparent hallucination is actually staleness: hours that changed, a property that renovated, a route that no longer runs. The model is not inventing; it is reporting the last thing it saw, in the same assured register as a current fact. 25% of travellers report receiving outdated or inaccurate AI travel information.

https://travel.rankwit.ai/glossary/hallucination

### Agentic AI

AI systems that plan and execute multi-step tasks against real systems, rather than only generating text.

The difference from a chatbot is action. An agentic system searches availability, compares options, holds inventory and can complete a booking. Travel is an obvious application and a hard one: 91% of agencies work across four or more booking systems, which is exactly the fragmentation under which an agent cannot reason reliably.

https://travel.rankwit.ai/glossary/agentic-ai

### AI Agent

A software component that pursues a goal by calling tools and APIs on a user's behalf.

Agents are judged on whether they complete tasks, not on whether their prose is good. For travel suppliers the implication is that machine-readable interfaces become a distribution channel: an agent can only book what it can reach, parse and trust. 52% of executives report agents running in production as of September 2025.

https://travel.rankwit.ai/glossary/ai-agent

### Tool Use

A model calling an external function or API to obtain data or perform an action.

Tool use is what separates a model that describes a hotel from one that checks whether a room is free tonight. It also changes what good content looks like: alongside pages for people, suppliers increasingly need endpoints for machines, with predictable shapes and honest error states.

https://travel.rankwit.ai/glossary/tool-use

### Model Context Protocol (MCP)

An open protocol for connecting AI assistants to external data sources and tools.

MCP standardises how an assistant discovers and calls a capability, which turns 'integrate with every assistant' into 'expose one server'. Travel adoption is early but real: TourismIntel runs a public MCP server for travel industry intelligence, and Sabre has built one for travel distribution.

https://travel.rankwit.ai/glossary/mcp

### Conversational Commerce

Discovery, comparison and purchase completed inside a conversation rather than on a website.

Travel is a strong fit in principle and a weak one in practice so far: trips are complex, multi-person and high-value, which is where conversation beats browsing, but also where trust requirements are highest. The delegation figures mark the limit: around 79% of travellers have used AI for travel and about 6% are comfortable letting it handle most of their planning. Agentic AI is the second attempt at the same idea with better tooling.

https://travel.rankwit.ai/glossary/conversational-commerce

### AI Overviews

Google's generated answer shown above the traditional results on a search page.

AI Overviews put a synthesised answer, with a handful of links, between the query and the result list. For travel queries that are informational (what to see, when to go, how long to spend), the overview can satisfy the question entirely, which is the mechanism behind zero-click behaviour.

https://travel.rankwit.ai/glossary/ai-overviews

### Zero-Click Search

A search that is resolved without the user visiting any result.

Zero-click is not new, but generated answers extend it from simple facts to the kind of multi-source synthesis travel research depends on. The strategic response is to be inside the answer rather than to fight for the click: being named and described accurately in the synthesis is worth more than a link nobody follows.

https://travel.rankwit.ai/glossary/zero-click

### Search Engine Results Page (SERP)

The page a search engine returns for a query.

The SERP is no longer a simple ranked list. Generated answers, panels and modules now occupy the positions that used to carry organic traffic, which is why share-of-voice measurement has partly replaced rank tracking for travel brands.

https://travel.rankwit.ai/glossary/serp

### AI Crawler

A bot that fetches web content for training or for grounding live AI answers.

Distinguishing crawler purposes is now an operational decision. GPTBot, ClaudeBot, PerplexityBot, Google-Extended and others have different roles, and blocking them has different consequences: blocking a training crawler affects future models, while blocking a retrieval crawler can remove you from answers today. Many travel sites block both by accident through a copied robots.txt.

https://travel.rankwit.ai/glossary/ai-crawler

### robots.txt

The file that tells crawlers which parts of a site they may fetch.

robots.txt now governs AI visibility as well as search indexing, and the two need separate decisions. A common and costly error in travel is a blanket disallow inherited from a template, which quietly removes a property from the retrieval layer AI answers are built on. Strict parsers end a rule group at the first blank line, so formatting mistakes can disable rules silently.

https://travel.rankwit.ai/glossary/robots-txt

### llms.txt

A proposed plain-text file at the root of a site that points AI agents at its most useful content.

llms.txt is a convention rather than a standard: a Markdown-formatted map of a site's canonical, machine-readable entry points. It is cheap to publish and increasingly common among data-led travel publishers. This observatory serves one, together with a full-text variant.

https://travel.rankwit.ai/glossary/llms-txt

### Crawl Budget

The volume of fetching a crawler will spend on a site in a period.

Budget is spent before it is earned. Slow responses, redirect chains, near-duplicate pages and parameter sprawl consume fetches that could have gone to pages that matter. For large travel sites with faceted inventory this is often the difference between being current in the retrieval layer and being a year out of date.

https://travel.rankwit.ai/glossary/crawl-budget

### Structured Data

Machine-readable markup that states explicitly what a page's content means.

Structured data converts prose into assertions a machine can use without inference: this is a hotel, this is its address, this is the price, this is the date the figure refers to. For AI answers it reduces the chance of a model guessing wrong, and it is the single highest-leverage technical change most travel sites can make.

https://travel.rankwit.ai/glossary/structured-data

### Schema.org

The shared vocabulary used to describe entities and relationships in structured data.

Schema.org supplies the types most travel sites need: Hotel, Restaurant, TouristAttraction, TouristDestination, Offer, Review, FAQPage, Dataset. Using the specific type rather than a generic one gives a model more to work with, and using Dataset and Claim types for statistics is what makes a figure quotable with its provenance intact.

https://travel.rankwit.ai/glossary/schema-org

### JSON-LD

The JSON-based syntax for embedding structured data in a page.

JSON-LD lives in a script tag, separate from the visible markup, which makes it easier to maintain than inline alternatives and is the format search and AI systems prefer. Every page in this observatory carries JSON-LD describing what it is, when it was updated and where its figures came from.

https://travel.rankwit.ai/glossary/json-ld

### Entity

A distinct real-world thing a model can reason about: a property, a city, a person, an organisation.

Models answer about entities, not keywords. A property that is described consistently across its own site, directories, review platforms and destination sites resolves to one confident entity. One described inconsistently resolves to several uncertain ones, and uncertainty is the most common reason a model declines to name a business.

https://travel.rankwit.ai/glossary/entity

### Knowledge Graph

A structured network of entities and the relationships between them.

Knowledge graphs let a system answer questions that were never written down anywhere: which hotels are walkable from a station, which restaurants near a museum open on Mondays. Travel is unusually graph-shaped: places contain places, and almost every useful query is relational.

https://travel.rankwit.ai/glossary/knowledge-graph

### Name, Address, Phone (NAP)

The core identity facts of a physical business, and the baseline consistency test.

NAP consistency predates AI and matters more now. Conflicting names, addresses or phone numbers across platforms fragment an entity, and a fragmented entity is one a model is less likely to name. It is the cheapest possible fix and still one of the most common faults in independent hospitality.

https://travel.rankwit.ai/glossary/nap

### Stale Content

Information that was accurate when published and no longer is.

Stale content is more dangerous than missing content, because a model will state it with full confidence. Closing days, renovation status, routes, prices and policies are the usual offenders in travel. Dating your facts explicitly and updating them on a schedule is the only reliable defence.

https://travel.rankwit.ai/glossary/stale-content

### Freshness

How recently information was published or verified, and how visibly that is stated.

Freshness signals help both ranking and grounding, and in travel they are often the deciding factor between two otherwise equivalent sources. Explicit review dates, visible update timestamps and structured dateModified values all help. This observatory states when each figure was last checked for exactly this reason.

https://travel.rankwit.ai/glossary/freshness

### Prompt Set

A fixed, representative list of questions used to measure AI visibility over time.

A prompt set is the instrument. It should cover the intents that actually produce bookings (category, destination, price band, season, audience) and stay stable so results remain comparable week to week. Changing the prompts and the period at the same time makes the series meaningless.

https://travel.rankwit.ai/glossary/prompt-set

### Intent Cluster

A group of prompts expressing the same underlying traveller need in different words.

Clustering by intent rather than by phrasing is what makes AI visibility measurable at all, since models respond to meaning rather than exact strings. 'Where should I stay in Florence with kids' and 'family-friendly Florence hotels' belong to one cluster and should be scored together.

https://travel.rankwit.ai/glossary/intent-cluster

### Destination Management Organisation (DMO)

The body that leads and coordinates a city, region or country as a travel destination.

DMO officially stands for destination management organisation (UN Tourism's definition, where the role is to lead and coordinate a destination, not merely to advertise it). The term is frequently rendered "destination marketing organisation", and several of the surveys cited here use that wording for their respondents. DMOs sit at the sharpest edge of the AI transition: 22% offer AI trip planning on their site while 54% of travellers already plan with AI, and 72% of their leaders say two more years of inaction risks irrelevance.

https://travel.rankwit.ai/glossary/dmo

### National Tourism Organisation (NTO)

The national-level body responsible for promoting a country as a destination.

NTOs are the country-scale case of the same problem. Around 40% of European NTOs have a dedicated AI team or monitor AI use, according to the European Travel Commission, which is a weaker thing than running AI in operations. Italia.it, run by ENIT, is the example this observatory follows for sourcing discipline.

https://travel.rankwit.ai/glossary/nto

### Online Travel Agency (OTA)

A platform that aggregates and sells travel inventory from many suppliers.

OTAs are both the incumbent distribution layer and an active participant in the AI shift. 951.6 million EU guest nights ran through Airbnb, Booking and Expedia in 2025, roughly a third of the EU total. Whether assistants route around that concentration or reinforce it is the open commercial question of the decade.

https://travel.rankwit.ai/glossary/ota

### Direct Booking

A reservation made with the supplier rather than through an intermediary.

Direct bookings avoid commission and carry the guest relationship, which is why AI-mediated discovery interests suppliers: a traveller who arrives from an AI recommendation is high-intent and has not been routed through an OTA. That only works if the model knows the property well enough to name it.

https://travel.rankwit.ai/glossary/direct-booking

### Global Distribution System (GDS)

The wholesale network connecting travel suppliers to agencies and sellers.

Amadeus, Sabre and Travelport are the GDS incumbents, and all three are now building agent-ready interfaces. The strategic question is whether agents treat a GDS as their inventory layer or route around it, which is why their current research output is so focused on granular APIs and governance.

https://travel.rankwit.ai/glossary/gds

### Revenue per Available Room (RevPAR)

Room revenue divided by available rooms, the headline hotel performance metric.

RevPAR combines occupancy and rate into one figure, which is why it is the number hotel operators watch. It is included here because AI visibility work is ultimately judged against it: citation share that does not move RevPAR is a vanity metric.

https://travel.rankwit.ai/glossary/revpar

### Average Daily Rate (ADR)

Average room revenue per occupied room.

ADR isolates pricing from occupancy. 40% of European accommodations reported ADR increases in 2026, down from 43% the year before (a useful reminder that rate growth is normalising even as demand holds).

https://travel.rankwit.ai/glossary/adr

### Occupancy

The share of available rooms that are sold.

Occupancy is the volume half of RevPAR. 50% of European properties reported occupancy growth in 2026, ten points better than the previous year, which is the context against which AI-driven demand claims should be read.

https://travel.rankwit.ai/glossary/occupancy

### Featured Answer

A short extracted response displayed above or instead of the results.

Featured answers were the first widely visible form of answer-engine behaviour and taught the discipline that AEO now formalises: lead with the direct answer, keep it self-contained, and make the attribution unambiguous. Generated answers extend the same logic to synthesis across many sources.

https://travel.rankwit.ai/glossary/featured-answer

## Sources

### OECD (Organisation for Economic Co-operation and Development)
- Type: Intergovernmental body
- Coverage: 38 member countries and partner economies
- https://www.oecd.org/en/topics/tourism.html
- Published method: https://www.oecd.org/en/data/dashboards/oecd-tourism-trends.html
- Intergovernmental body publishing the Tourism Trends and Policies series, harmonised inbound and domestic tourism statistics, and policy papers on AI adoption in tourism.
### Eurostat (Statistical Office of the European Union)
- Type: Official statistics
- Coverage: EU-27, EFTA and candidate countries
- https://ec.europa.eu/eurostat/web/tourism
- Published method: https://ec.europa.eu/eurostat/cache/metadata/en/tour_occ_esms.htm
- The EU statistical authority. Source of the monthly accommodation series (tour_occ_nim) and of the experimental platform-economy statistics covering Airbnb, Booking, Expedia and TripAdvisor (tour_ce_omr).
### UN Tourism (World Tourism Organization (UNWTO))
- Type: Intergovernmental body
- Coverage: Global, 160 member states
- https://www.untourism.int/
- Published method: https://www.untourism.int/tourism-statistics/un-standards-for-measuring-tourism
- The United Nations agency for tourism. Custodian of the International Recommendations for Tourism Statistics and publisher of guidance on responsible AI adoption across the sector.
### European Travel Commission
- Type: Trade association
- Coverage: Europe, 36 national tourism organisations
- https://etc-corporate.org/
- The association of European national tourism organisations. Runs the recurring survey of NTO digital and AI maturity and the quarterly European Tourism Trends & Prospects.
### Phocuswright
- Type: Industry research
- Coverage: Global travel distribution and technology
- https://www.phocuswright.com/
- The reference research house for travel distribution. Consumer travel trackers and the Travel Forward and State of Distribution series.
### McKinsey & Company
- Type: Consultancy
- Coverage: Global
- https://www.mckinsey.com/industries/travel/how-we-help-clients
- Management consultancy publishing travel practice research on generative and agentic AI, often with Skift Research.
### Deloitte
- Type: Consultancy
- Coverage: United States
- https://www.deloitte.com/us/en/insights/industry/transportation/2026-summer-travel-trends-survey.html
- Runs the annual US Summer Travel Survey, one of the few repeated instruments that measures AI-assisted trip planning year over year on a consistent question wording.
### Euromonitor International
- Type: Industry research
- Coverage: Global, 40+ markets
- https://www.euromonitor.com/store/explore-reports/travel
- Market research firm. The Voice of the Consumer: Travel Survey is fielded annually across tens of thousands of respondents.
### Amadeus
- Type: Vendor research
- Coverage: Global travel technology
- https://amadeus.com/en/blog
- Global distribution system and travel technology provider. Annual Travel Trends report and traveller surveys, plus white papers on agentic AI for airlines.
### Sabre
- Type: Vendor research
- Coverage: Global travel technology
- https://www.sabre.com/resources/research/
- Travel technology provider. Research on conversational commerce, content fragmentation across agencies, and practical agentic AI deployment.
### Booking.com
- Type: Platform data
- Coverage: Global, 33 markets
- https://news.booking.com/
- Online travel agency. Publishes the Global AI Sentiment Report and the European Accommodation Barometer, both with disclosed sample sizes and field periods.
### Sojern
- Type: Vendor research
- Coverage: Global, with US traveller panels
- https://www.sojern.com/reports/
- Travel marketing platform. Runs paired surveys of travellers and destination marketing organisations, which makes the adoption gap between the two directly measurable.
### Travelport
- Type: Vendor research
- Coverage: Global
- https://www.travelport.com/reports
- Travel retailing platform. The annual State of Modern Retailing report covers consumer trust in AI and in travel pricing.
### Anthropic
- Type: Platform data
- Coverage: 121 countries
- https://www.anthropic.com/economic-index
- Published method: https://www.anthropic.com/research/economic-index-june-2026-report
- AI developer. The Economic Index publishes privacy-preserving aggregate statistics on how Claude is actually used, by task and by geography, rare first-party telemetry on AI usage.
### Google
- Type: Platform data
- Coverage: Global
- https://blog.google/
- Operator of Search, AI Overviews and Gemini. Publishes the ATLAS study on AI tool usage and periodic travel-search research.