Chuhee LeeIndependent research
All publications
Cover art: four stations on one path, one per role in a booking; the first, the answer screen, is filled.
Horizon
2026–2031
Scope
US-led · Europe, East Asia
Lens
Discover, plan, book, decide on the move
Basis
Expert forecasts
ShiftsTravelNovember 2026 edition · Research report

Trip Discovery Outlook 2026–2031

Trip discovery leaves the click: how AI answers, agents, social video and new devices will change the way people find, plan and book trips by 2031, and who keeps the traveler.

Research questionHow will AI assistants and agents, social video and new devices change how people discover, plan and book trips by 2031, and who will stand between the traveler and the seller?

By Chuhee Lee≈24 min readData as of 3 Nov 2026Next refresh: November 2027Change log

With research assistance from several AI models (Claude, Gemini, ChatGPT) · Views are the author’s own

Cite asChuhee Lee, Trip Discovery Outlook 2026–2031, November 2026 edition (data as of 3 November 2026), https://blueplanet-ai.github.io/trip-discovery/, CC BY-NC 4.0.

The report in 60 seconds

Three takeaways · each with its source
  1. 56% vs <1% Fact

    Travelers say they use AI; sites barely see it

    56% of active US travelers used AI for at least one trip in the past year, and 28% of all US adults used it to plan one in the past six months. Booking Holdings says AI chat referrals still bring significantly less than 1% of its room nights, though they are accelerating.

    Source: Phocuswright [1]; Longwoods [2]; Booking Holdings [3]
  2. −18.8 pts Interpretation

    Answers take the click, not the booking

    In a randomized experiment, Google’s AI Mode cut clicks to outside sites by 18.8 percentage points, while Booking.com and Expedia still take about two-thirds of their consumer bookings direct. The author reads this as AI moving the moment of decision, not yet the transaction.

    Source: Wang et al. [4]; Booking Holdings [5]; Expedia [6]
  3. 30% vs 8% Hypothesis

    Answers up front, brands at checkout

    IDC forecasts 30% of travel bookings executed by AI agents by 2030; today 8% of travelers are comfortable booking through an AI platform. The author’s central case for 2031 is that AI answers lead discovery and planning while travel brands and platforms keep the booking, payment and service: not yet observed.

    Source: IDC [7]; Expedia and YouGov [8]

Fact: stated by the cited source. Interpretation: the author’s reading of the cited evidence. Hypothesis: expected from the evidence but not yet observed. Figures inside every takeaway are sourced.

At a glance

One-page summary

Headline findingThrough 2031, travelers increasingly get trip ideas and plans as answers rather than clicks, mostly from Google, while travel brands and platforms keep the booking, the payment and the service.

Summary

The short version

Expert forecasts and today’s evidence point the same way for the next five years: travelers will increasingly get trip ideas, comparisons and plans as answers, from Google, chat assistants, travel apps and social video, rather than by clicking through pages of results. The checkout moves much more slowly. The author’s central case for 2031, “Answers up front, brands at checkout”, is that whoever owns the answer screen will not usually be the one who takes the payment, carries the booking and handles the cancellation. Trip discovery leaves the click; the booking stays with brands and travel platforms.

We start from a gap.

Surveys put AI use for trips at 28% to 56% of US travelers, depending on who is asked and how. Travel sites measure less than 1% of their visits or bookings arriving from AI assistants. Most of the gap is answers that never send a click, and AI inside Google counted as search.[1][2][3][9]

Answers replace clicks.

Search engines as travelers’ most-used research tool fell from 51% to 36% in about 18 months in Phocuswright’s tracking. In a randomized field experiment, Google’s AI Mode cut outbound clicks by 18.8 percentage points, yet Google reported its fastest query growth in five years.[10][4][11]

Planning becomes a conversation; social stays the spark.

AI trip planners mostly look for activities, ideas and food. 60% of US travelers say social media inspires their trips. The best models still produce a valid multi-constraint itinerary less than half the time in research benchmarks.[2][12][13]

Agents at the checkout are a forecast, not yet a habit.

IDC forecasts 30% of travel bookings executed by AI agents by 2030. Today 8% of travelers are comfortable booking through an AI platform, and Google and OpenAI both leave the hotel or travel company as merchant of record.[7][8][14][15]

Three regions, three setups.

In the US, Google adds hotel booking to AI Mode. In Europe, competition rules have stripped Google’s travel features and excluded the EEA from its new AI travel tools. In Korea and China, Naver and Trip.com keep AI inside their own apps.[14][16][17][18]

The starting point: what travelers say against what sites measure

Percent · different bases, two kinds of measure · 2026

Say they use AI for trips (surveys)
Active US travelersany AI use for a trip, past 12 months · Phocuswright
56%
Air passengers, North Americaused AI for travel planning · IATA
47%
US travelers, active planningPhocuswright
39%
US adultsused AI to plan a trip, past 6 months · Longwoods
28%
What travel sites measure
Booking.com room nights from AI chat referralsQ3 2026
<1%
Visits to travel sites from AI assistantsevery category Bain tracked · Similarweb data
<1%
Visits to 41 UK travel sites from AI assistantsJan–Apr 2026 · Adido
0.37%

Survey bars and measured bars use the same axis but answer different questions: the first share people, the second share visits or bookings. The “less than 1%” bars are drawn at 1%.[1][19][10][2][3][9][20]

Chapter 1 of 9
Baseline

Where we start in 2026

Travelers say they use AI for trips far more than travel sites can measure; the gap is mostly how answers work, not a measurement error.

What travelers say. 56% of active US travelers used AI for planning, booking or in-destination help on at least one trip in the past 12 months, up from 33% in the first half of 2025.[1] Narrower questions give lower numbers: 39% of US travelers actively use AI to plan trips (Phocuswright), 49% of air passengers worldwide and 47% in North America have used AI for travel planning (IATA, over 10,000 responses), and 28% of US adults used AI to plan a trip in the past six months, up from 19% in August 2024 (Longwoods, 1,000 adults matched to the Census).[10][19][2]

What travel sites measure. Booking Holdings says referrals from AI chat assistants still bring significantly less than 1% of Booking.com’s room nights, though they “accelerated over the last several weeks” of the third quarter, and in every travel category Bain tracked with Similarweb data, AI referrals stay under 1% of site traffic.[3][9] Across 41 UK travel sites, 0.37% of visits came from AI assistants, 88% of them from ChatGPT.[20]

Why the gap. Most AI travel answers send no click; Google’s AI answers are counted as search; and a traveler who asks ChatGPT and then opens the Booking.com app shows up as a direct visit. Survey wording widens the range: Phocuswright’s 56% counts planning, booking or help at the destination. Half of travelers who use AI in search engines still click through to the source.[1] The site’s earlier analysis of this gap is folded into this chapter.

Where the money sits. OTAs took about one fifth of US travel gross bookings in 2025, 100.3 billion dollars, and Booking.com and Expedia take about two-thirds of their consumer bookings direct, through their own apps and sites.[21][3][6]

Author’s read · confidence: highThe five-year question is not whether travelers will use AI, which most already do, but whether AI moves the click, the booking and the customer relationship, and to whom.
Chapter 2 of 9
Shift 1 · discovery

Answers replace clicks

AI answers become the first stop for trip research; clicks to travel sites fall faster than bookings, and the answer is mostly Google’s.

What experts forecast. Phocuswright says AI is “on a path to replace traditional search as the primary front door” for travel. Its tracking shows general search engines falling as travelers’ most-used research resource, from 51% in mid-2024 to 36% in late 2025.[10] Generative AI platforms such as ChatGPT reached 33% use for trip research, five times their 2024 level.[22]

What AI answers do to clicks. In a preregistered field experiment with 1,100 US Google users, switching people to AI Mode only cut the share of searches with a click to an outside site by 18.8 percentage points, lowered trust in Google’s information and raised searching on rival engines by 11.2 points. Removing AI features raised clicks by 8.8 points.[4] The study is a preprint and not about travel alone.

But Google grows. In the third quarter of 2026 Google’s search advertising revenue rose 12% to 63.3 billion dollars, query growth was the highest in five years, and AI Mode queries were three times longer than traditional searches, one in five using voice, images or documents. Google is now expanding ads in AI Mode internationally.[11] The growth costs something: search revenue growth slowed from 17% in the second quarter, and Google’s chief business officer said “the strong growth in AI Mode queries had an impact because of the lower monetization we were seeing in AI Mode”, alongside currency, the EU’s end of a customs exemption and a post-World Cup comedown.[23][11] In the experiment above, participants chose AI Mode for only 0.6% of their searches; AI Overviews appeared on 36%.[4]

Who gets cited. In 1,357 citations Gemini gave for Tokyo hotel questions, price-led questions (“cheap hotel in Shinjuku”) mostly cited OTAs, while experience-led questions drew a majority of citations from hotels, guides and other non-OTA sources.[24] Expedia says answer-engine optimization and social are two of its fastest-growing marketing channels.[6] Skift’s 2026 megatrends tell travel brands to optimize for AI-driven discovery, “GEO over SEO”, and Bank of America calls agentic AI the most significant theme for online travel in 2026, a challenge to “traditional search-and-click models”.[25][26]

What AI Mode does to clicks

Change in percentage points against today’s Google · 1,100 US users · randomized, March 2026

Google with no AI featuresvs today’s Google
+8.8 pts
Google AI Mode onlyvs today’s Google
−18.8 pts
Other effects of AI Mode only
Searching on a rival engineBing, DuckDuckGo or Yahoo
+11.2 pts

The −18.8 bar shows the size of a drop. Preprint by researchers at the University of Pennsylvania and Northeastern University.[4]

What cuts against this. Forecasts of search collapse have been early before: in February 2024 Gartner predicted search engine volume would drop 25% by 2026; Google now reports its strongest query growth in five years.[27][11] Much of the shift happens inside Google rather than away from it. AI Mode is still a small share of searches, and forced use lowered satisfaction in the experiment.[4]
Author’s forecast · confidence: medium-highBy 2031 an AI answer is the usual first step of trip research for most US travelers under 50, and most of those answers are Google’s, in Search, Maps and Gemini, rather than a new assistant’s. Clicks from search to travel sites fall well before bookings do, and travel marketing shifts from ranking pages to being cited and bookable inside answers.
Chapter 3 of 9
Shift 2 · planning

Planning becomes a conversation

Itineraries start as AI drafts that travelers check; travel companies’ own planners gain on standalone startups, and accuracy decides how far people trust them.

What people do with it. US adults who used AI to plan a trip mostly looked for attractions and activities (54%), inspiration for places to visit (44%) and dining (42%).[2] In Japan, 61% of people planning a summer trip in 2026 said they would use generative AI to look up local food, build itineraries or compare transport, and the survey’s sponsor said AI is replacing the guidebook.[28] In Cornell’s survey of 1,029 US travelers, AI ranks fourth among planning tools, behind search engines, review websites and official hotel sites; more than 60% name accuracy as the main barrier.[29]

Accuracy is the limit. In a 2024 benchmark of 1,225 realistic trip requests, GPT-4 produced a plan meeting every constraint 0.6% of the time.[30] A 2026 re-test of GPT-5.2, Claude Sonnet 4.5, Gemini 3 Pro and others found the best model succeeded 43% of the time on that benchmark, 15% or less on a harder one, and no model extracted every constraint exactly on a set of Chinese trip requests.[13] Public incidents, such as AI tools sending tourists in Peru to a canyon that does not exist, keep the risk visible.[31]

Travel companies build their own. Expedia bought the AI planner Layla in July 2026. Its chief executive said its natural-language tools “are not driving conversion right now” but give over 60% more information about what travelers want.[32][6] Trip.com says orders assisted by its TripGenie assistant rose about 400% in a year, with nearly 60% of interactions now about booking.[18] Human advisors are adding AI rather than being replaced by it: Fora, a platform for independent advisors, raised money at a 1 billion dollar valuation in July 2026.[33]

Where AI leads, travel changes. Among 2,669 foreign visitors to Seoul, the 23 who named generative AI as a leading information source spread their spending across more categories and visited the city’s iconic heritage sites less.[34] The group is small, but it is the first sign that AI planning changes where tourists go, not just how they book.

What US AI trip planners use it for

Share of adults who used AI to plan a trip · Longwoods, 1,000 US adults, July 2026

Find attractions and activities
54%
Get inspiration for places to visit
44%
Find dining options
42%

Base: the 28% who used AI to plan a trip in the past six months; subgroup size not published.[2]

What cuts against this. Benchmarks test full itineraries with many constraints; most travelers ask for ideas, which tolerate errors. Luxury travelers in the Cornell study use AI for research but strongly prefer human advisors to put complex trips together.[29] Standalone planners have a weak record: Google closed its Trips app in 2019 and folded its features into Search and Maps.[35]
Author’s forecast · confidence: mediumBy 2031 most US leisure trips start with an AI draft, a day-by-day plan or a shortlist, that the traveler checks against reviews, maps and prices. The planners that win sit inside Google, the big travel platforms and super-apps, which hold live prices and inventory; independent planners survive as features others buy.
Chapter 4 of 9
Shift 3 · inspiration

Social video stays the spark

Trips are still first imagined on social video; AI turns saved posts into plans, and trust in friends, creators and reviews decides which ideas survive.

Social leads inspiration. 60% of US travelers say social media inspires their trip ideas, but trust in friends, credible creators and reviews decides which ideas move forward, McKinsey found in a survey of more than 1,000 travelers and diaries from 26 travelers. It describes trip planning as a set of overlapping “orbits” rather than a funnel.[12] In Spain, Italy and the UK, 43% say they have booked a trip or activity after seeing a viral post or video, and 48% of Gen Z name TikTok as their main source of travel inspiration (TUI Musement, a seller of tours and activities).[36]

AI arrives later in the journey. In Skift Research’s diary study, AI accounted for 21% of touchpoints at the inspiration stage and 37% across inspiration and research combined, but 6% at booking. US travelers’ comfort with AI for ideas and research (net score 58) is about twice their comfort with AI for booking (27).[37]

From video to plan. Expedia’s Trip Matching turns an Instagram Reel sent to its account into a bookable itinerary, startups turn a TikTok link into stops on a map, and Google’s AI Mode accepts images as questions.[38][39][11] Amadeus calls the result “travel mixology”: travelers combine assistants, Reddit, YouTube and brands’ own tools rather than switching from one to another.[40] The leading content travelers save, and the reviews they read, become the material AI answers cite.

AI’s share of trip touchpoints, by stage

Percent of touchpoints · Skift Research diary study, 26 travelers · updated Oct 2026

Inspiration
21%
Inspiration and research
37%
Booking
6%

The middle bar combines inspiration and research. A small diary sample; read the pattern, not the levels.[37]

What cuts against this. The social figures are surveys, two of them paid for by travel companies, and none measures how much booking value social video really creates. Social discovery also concentrates crowds: 39% in the TUI survey say they visited a tourist site mainly to take a photo or video.[36]
Author’s forecast · confidence: mediumThrough 2031 social video remains where most leisure trips are first imagined, especially for travelers under 35. AI does not replace it; it converts saved videos and posts into plans and bookings, which makes creators and reviews more valuable as the evidence AI cites.
Chapter 5 of 9
Shift 4 · booking

Agents at the checkout

Forecasters expect AI agents to book a large share of travel by 2030; today’s products keep the travel company as seller, and business travel moves first.

What experts forecast. IDC predicts that by 2030, 30% of travel bookings will be executed by AI agents; its public summary gives no method or definition.[7] For all US e-commerce, Morgan Stanley expects agentic shoppers to reach about 10% of online retail by 2030 (up to 20%), Bain expects 15–25% of purchases to be initiated, influenced or completed by agents, and McKinsey expects agents to “de-verticalize” shopping, pulling people away from single-category apps such as Expedia toward general assistants.[41][42][43] The forecasts differ mostly in what they count.

What travelers say. In Accenture’s survey of 5,003 leisure travelers in 12 countries (June 2026), 40% would fully hand comparing travel options to an AI agent but only 10% would hand over payment.[44] 44% of US travelers say they would book directly inside an AI platform and 40% would let an AI assistant handle flight and hotel bookings (Phocuswright).[22] Asked about comfort rather than intention, the numbers fall: 8% are comfortable booking through an AI platform and 68% prefer a trusted travel brand (Expedia and YouGov, US, UK and India).[8]

What products do. Google started hotel booking in AI Mode in the US in August 2026 with ten partners including Booking.com, Expedia, Marriott and Hilton; the hotel or booking platform remains merchant of record and handles customer service.[14] OpenAI’s checkout protocol is designed the same way.[15] Of more than 60 travel tools for Claude and ChatGPT checked in October 2026, five could finish a booking inside the chat, all for business travel, plus one for personal trips; Expedia, Booking.com, Kiwi.com and trivago only search and send the traveler to their own sites.[45]

Where agents already book. In China, Alibaba’s Fliggy moved its assistant from pilot to live bookings inside Alibaba’s Qwen AI app in March 2026, for flights, trains and attraction tickets; it reports AI orders up 800% during the Spring Festival campaign and conversion comparable to established OTAs.[46] There the assistant and the travel platform belong to the same company. In the US, small bookings come first: Booking Holdings says OpenTable’s native integration with OpenAI has become one of its fastest-growing channels, “from a small base”.[3] Travel executives expect more: about three-quarters of the 600 Accenture surveyed believe AI will be executing bookings, payments and disruption handling for customers within three years, and slightly more of them name AI-native platforms (21%) than airlines and hotels (19%) as the likely owner of the traveler relationship.[44]

The record. Booking without leaving the search page was announced as inevitable for a decade, until Tripadvisor closed its Instant Booking in 2026, one industry tracker notes.[47] OpenAI reportedly stepped back from in-chat checkout in March 2026.[48]

How far agents go: 2030 forecasts against 2026

Percent · different bases: travel bookings, US e-commerce, travelers, touchpoints, room nights

Forecasts for 2030
IDC: travel bookings executed by AI agentsglobal · method not published
30%
Bain: US e-commerce initiated, influenced or completed by agentsdrawn at the 20% midpoint
15–25%
Morgan Stanley: US e-commerce by agentic shoppers10% likely, up to 20%
10–20%
Today, 2026
Travelers comfortable booking through an AI platformUS, UK, India · Expedia / YouGov
8%
AI’s share of booking touchpointsdiary of 26 travelers · Skift Research
6%
Booking.com room nights from AI chat referralsQ3 2026
<1%

Ranges are drawn at their midpoint (Bain) or likely case (Morgan Stanley). The forecasts are not like-for-like: Bain counts purchases agents only influenced; IDC does not define “executed”.[7][42][41][8][37][3]

What cuts against this. Payment rails for agents now exist: Google says it is combining its Universal Commerce Protocol with an Agent Payments Protocol, and business-travel agents already book in chat.[11][45] Google’s third-quarter examples of agent checkout were all retail; travel was not discussed on the call.[11] If loyalty programs and cancellations work through agents, leisure could follow faster than this report expects.
Author’s forecast · business and simple rebooking: medium · complex leisure trips: lowBy 2031 agents book a meaningful share of business trips and simple, repeat purchases such as a known hotel, a flight change or a restaurant table. For most leisure trips the agent shortlists and hands off; the travel company stays merchant of record, owns the cancellation and keeps the loyalty points.
Chapter 6 of 9
Shift 5 · during the trip

Decisions on the move

The fastest growth is during the trip: travelers ask where to eat, how to get there and what a sign says, first on the phone, then by voice and glasses.

In-trip use is already common. Among Korean Millennial and Gen Z travelers, help during the trip is the most common use of AI (32%), ahead of transport routes and price comparison.[49] US AI trip planners look for activities and dining more than for flights or hotels.[2] Phocuswright’s headline AI figure counts help at the destination.[1]

New interfaces are built for it. Google’s first Android XR audio glasses, made with Samsung, Gentle Monster and Warby Parker, arrive this fall with turn-by-turn directions based on where you face, reviews of the restaurant you are walking past, live translation and ride ordering.[50] In Korea, Naver’s AI Briefing reaches more than 30 million people a month, and its AI Tab, with 10 million users, moves from search to purchases and reservations.[17] The site’s In-Car AI benchmark tracks the same shift in cars.[51]

What cuts against this. Travel-specific measures of in-trip AI use are rare; IATA’s survey measures planning, where flight search (21%) and cheapest fares (19%) lead.[19] In-trip errors cost more (a closed restaurant, a wrong platform), and roaming, battery and privacy limit always-on assistants abroad.
Author’s forecast · confidence: mediumBy 2031 asking becomes the normal way to make small decisions during a trip: where to eat, what to do next, how to get there, what this says. The phone remains the main device; audio glasses and car assistants take the hands-free moments. This fits the site’s thesis that time and attention on the move are the next contested space.
Chapter 7 of 9
Shift 6 · who keeps the traveler

Who keeps the traveler

Owning the answer screen and owning the customer are splitting apart; channel shares barely move in forecasts, and the US, Europe and East Asia diverge.

Four roles, not one. A booking involves four roles: the screen where the traveler decides, the inventory behind it, the merchant of record who takes the payment, and the company that handles changes and complaints. AI is moving the first role fastest. Today’s AI booking products leave the other three with hotels, airlines and OTAs.[14][15]

Channel forecasts barely move. Phocuswright projects the OTA share of US travel gross bookings rising from 20% in 2025 to 21% in 2028.[21] In the UK, the supplier-direct share of the online market moves only from 78% to 81% over its six-year forecast, and the OTA share drifts from 20% to 19% by 2029.[52] Booking.com says its direct share stayed in the mid-60s percent over the four quarters to September 2026, with its app in the high 50s, and that Connected Trip transactions, which combine several travel products, grew almost 20%; Expedia says two-thirds of its consumer-brand bookings come direct and its business-to-business bookings grew 21% in the second quarter.[3][6]

What incumbents and skeptics say. Trip.com’s chairman, James Liang: “we see AI as an evolution of the travel user interface rather than simply a new traffic channel,” and agents will become “an increasingly important discovery and inspiration channel.”[18] Altimeter Capital’s Thomas Reiner expects OTAs to decline slowly, “like the Yellow Pages — permanently damaged, never deceased.”[47] Airbnb kept out of ChatGPT’s apps.[53]

RegionSetup in 2026What it implies for 2031
United StatesGoogle adds hotel booking, price tracking and points to AI Mode; ChatGPT carries ads and travel apps; OTAs and brands keep the merchant role.[14][54][55]Google holds the answer screen; OTAs and brands compete to be the bookable option inside it.
Europe (EEA)The Commission fined Google 460 million euros for self-preferencing in search in July 2026; Google removed live prices and date filters from its travel units in September and excluded the EEA from its AI Mode flight tracking. Google says hotels report direct-booking clicks down as much as 30% after its earlier changes.[16][56][14][57]Rules slow Google’s answers in travel and keep large OTAs strong; AI assistants must also disclose they are AI.[58]
East AsiaNaver keeps AI search inside Korea’s main portal; Alibaba’s Fliggy books inside Alibaba’s own AI app; ByteDance’s Doubao cites Douyin, its sister video app, in most travel answers; Trip.com builds AI into its app and paid a 5.28 billion yuan antitrust charge, replacing its tiered hotel distribution programs.[17][46][59][18]Domestic super-apps keep discovery and payment together; global assistants play a secondary role.

Channel shares in forecasts: little movement

Percent · Phocuswright market estimates

US: OTA share of all travel gross bookings
2025
20%
2028 forecast
21%
UK: supplier-direct share of the online market
2023
78%
2029 forecast
81%

Different markets and bases: US share of all travel gross bookings; UK share of the online market. The UK forecast period runs 2023–2029.[21][52]

Author’s forecast · confidence: mediumBy 2031 Google, and in Asia the super-apps, own more of the moment of decision, while OTAs and travel brands keep most bookings, payments and service through loyalty, inventory and post-booking support. AI becomes another paid channel for them rather than a replacement, and the US, Europe and East Asia keep different setups rather than converging.
Chapter 8 of 9
Counterforces

What holds it back

Accuracy, trust, regulation, liability and closed platforms slow the shifts, especially at checkout; none stops AI-assisted discovery.

ForceEvidenceEffect on the shiftsKind
AccuracyBest model 43% on a planning benchmark; more than 60% of US travelers name accuracy as their main barrier.[13][29]AI drafts, people verify; slows booking by agents.Observed
Trust8% comfortable booking through AI; forced AI Mode lowered trust in Google’s information.[8][4]Favors trusted brands at checkout.Survey
Competition rulesEU fine and EEA redesign; Trip.com’s antitrust charge in China.[16][18]Limits bundling of answers and booking by dominant platforms.Observed
Pricing and consumer lawFTC orders to Delta, United, JetBlue and six others on prices set from personal data, loyalty status and algorithms (28 October 2026); UK regulator: “if an AI agent you use does something illegal, you are responsible.”[60][61]Makes travel companies cautious about letting agents set prices or book unchecked.Observed
Liability and serviceGoogle and OpenAI leave the merchant role, cancellations and support with the travel company.[14][15]Keeps the customer relationship with brands and OTAs.Company
Closed platformsAirbnb outside ChatGPT apps; big OTAs expose search, not booking, to assistants.[53][45]Assistants cannot book what platforms keep closed.Company
The pleasure of planningTravelers move through overlapping “orbits” of ideas; trust in people decides.[12]Browsing stays part of the trip for many leisure travelers.Survey
Author’s read · confidence: mediumThe counterforces act mostly on the checkout. They slow agents and keep bookings with brands, but they do little to stop travelers asking AI for ideas, plans and answers.
Chapter 9 of 9
Judgment, not forecast

Scenarios to 2031

The author’s central case splits by role: answers up front, brands at checkout. One alternative is close; agents booking most leisure trips is the least supported.

How good are such forecasts? Forecasts that a new interface would take over travel booking have mostly been early: Google’s standalone Trips app closed in 2019, booking inside metasearch faded, and Gartner’s 2024 forecast of a 25% drop in search volume by 2026 has not shown up in Google’s numbers.[35][47][27][11] Forecasts of new habits have done worse than forecasts of infrastructure.[62] This report therefore trusts forecasts of AI use for ideas more than forecasts of AI taking over the checkout.

ScenarioWhat 2031 looks likeEvidence forEvidence againstConfidence
Answers up front, brands at checkout
central case
AI answers, mostly Google’s and the big assistants’, are the first stop for ideas and planning. OTAs and travel brands keep the booking, payment and service, reached through hand-offs; travel sites get fewer but better-qualified visits.Search as most-used research tool down from 51% to 36%; AI Mode keeps the hotel as merchant of record; big OTAs expose search, not booking; channel shares stable in forecasts.Google may take more of the transaction over time; AI referrals are still under 1%, so the hand-off is not yet proven at scale.Medium-high
Platforms absorb AIGoogle, Booking, Expedia, Trip.com and Naver keep the front door with AI built into their own apps; standalone assistants stay a niche starting point for travel.Direct shares in the mid-60s percent; TripGenie orders up about 400%; Fliggy books inside Alibaba’s own AI app; Naver AI Tab at 10 million users; Google Trips folded into Search and Maps.Generative AI platforms reached 33% use for trip research; general assistants have daily habits travel apps lack.Medium
Agents bookAgents plan and book end to end for a large share of leisure trips; OTAs become supply behind them and the assistant owns the customer.IDC: 30% of travel bookings by 2030; three-quarters of travel executives expect AI to execute bookings within three years; payment protocols for agents; business-travel agents book in chat.Best model 43% on a planning benchmark; 8% comfortable booking via AI; 10% would hand an agent the payment; platforms refuse the merchant role.Low (business travel: medium)
PushbackErrors, ads in answers, pricing scandals and regulation slow adoption; AI stays a research aid and travelers keep comparing themselves.Forced AI Mode lowered trust; EU, Chinese and US regulators active; accuracy the top barrier.Use is already broad and rising; ideas and in-trip help do not need full accuracy or booking.Medium as a brake; low as a reversal

Confidence describes how well the evidence supports each story, not a probability. Gemini and ChatGPT, asked to test these scenarios, both rated the central case and “Platforms absorb AI” as best supported and agents booking most leisure trips as least supported.

What would change the picture

Quarterly

AI’s share of OTA bookings and direct share

Booking Holdings and Expedia earnings calls. AI referrals rising to several percent of room nights while direct shares hold supports the central case; direct shares falling as agent bookings rise supports Agents book. In the third quarter, Booking said AI referrals were still significantly below 1% of room nights but accelerating; a first stated figure of 1% or more would be the first crossing.[3][6]

As released

Booking tools in assistants

Whether Booking.com, Expedia and major airlines let assistants book inside the chat, and who is merchant of record. Search-only tools support the central case; in-chat booking supports Agents book.[45][14]

Quarterly

Clicks and ads in Google’s AI answers

Alphabet calls and independent studies. Fewer clicks with stable bookings fit the central case; Google extending checkout from hotels to flights and activities would move it toward Platforms absorb AI.[11][4]

Each wave

Comparable survey trackers

Longwoods and Phocuswright, with stable wording. Steady growth supports the shifts; a plateau supports Pushback.[2][10]

Twice a year

Planning benchmarks

Valid-plan rates on TravelPlanner and its successors. Rates well above half would make delegation credible.[13]

As decided

Regulators

The Commission’s DMA decisions, the FTC’s pricing study and consumer-law cases on agents. Strict liability for agents supports Pushback; clear rules could speed Agents book.[16][60][61]

Method, limits and flags

Read before citing

Question. How will AI assistants and agents, social video and new devices change how people discover, plan and book trips, and decide during them, by 2031, and who will stand between the traveler and the seller?

Scope. Leisure travel first, business travel where it shows what may reach consumers. Inspiration, planning, the booking hand-off and decisions during the trip. US first; Europe and East Asia as distinct setups.

Sources. Research papers, analysts’ and research firms’ forecasts, company disclosures, regulators, and major news and trade publications. Each reference carries its type and date. Observed, survey, company and forecast figures are kept apart and labeled.

How the research was done. Claude researched and verified the 2026 baseline and the forecasts; Gemini (research papers, government, Google material) and ChatGPT (analysts, companies, major publications) answered the same written brief; Claude opened their sources and merged the findings. Framing, the six shifts, the scenarios and conclusions are the author’s.

What this report is not. The “author’s forecast” boxes and the four scenarios are judgments based on the evidence cited. Confidence labels describe how well the evidence supports a statement, not a probability. Signposts give a direction and a source, not a numeric threshold.

Flags and open questions

  1. Third-quarter results in part. Alphabet (28 October) and Booking Holdings (3 November) third-quarter results are included, from Alphabet’s call transcript and Booking’s release and prepared remarks; Booking’s live question-and-answer session was not read, so company figures reported only from it (AI-assistant conversion rates, AI search visitors, OpenTable covers) are not used. Expedia (4 November), Airbnb and Tripadvisor (5 November) and Trip.com (date not announced) report later; their second-quarter statements are used.
  2. Forecasts, not measurements. This edition leans on what analysts, researchers and companies forecast to 2030–2031 and does not require each claim to be validated by an independent third party; each has a linked source. Many forecasters sell to the industries they forecast (IDC, Bain, McKinsey, Morgan Stanley, Phocuswright, Skift), and company figures (Google, Booking, Expedia, Trip.com, Naver) are claims. Each is labeled.
  3. Rescoped edition. An earlier draft of this report measured today’s gap between stated and measured AI use for trips; its verified figures now form chapter 1. Earlier sections on European distribution, Korean statistics and the AI trip-planning startup landscape were shortened or dropped.
  4. Research round with Gemini and ChatGPT, then Claude’s own pass. Both models answered the same written brief. Claude then searched independently for travel outlook reports, East Asian evidence and forward-looking research, which added Accenture’s traveler and executive survey, Fliggy’s live AI bookings, QuestMobile’s citation data, the Japanese survey, and the Skift, Amadeus and Bank of America outlooks. These agree with the two models’ findings; the main addition is that agent booking is already live in China where the assistant and the travel platform belong to one company. Claude opened the sources behind each new figure used here. Their main effects: the role split (answer screen versus merchant of record), business travel as the first market for booking agents, and treating the US, Europe and East Asia as different setups.
  5. Corrections from that round. Trip.com’s antitrust charge was 5.28 billion yuan, not 5.18 billion. The 460 million euro EU fine is the search part of an 890 million euro total. Alphabet’s third-quarter search revenue growth was 12%; the 17% figure belonged to the second quarter. Phocuswright’s 44% and 40% figures on booking through AI come from its July 2026 summary, not the March page.
  6. Not used because not verified or secondary only. Chinese figures on Golden Week AI planning, Alibaba’s Qwen orders and Doubao’s daily users seen only in a consultant’s blog; a figure of 36% of loyal travelers who would switch brands because of AI, not in the Accenture report; eMarketer’s in-chat checkout forecast and Gartner’s business-buying forecast (seen only through a vendor summary), Sabre’s MCP release notes (page did not load), a Sprout Social survey, a Skift article’s 2% figure (no named source), a hallucination study available only as an abstract, and a claim that 50% of searches would run through AI engines in 2026.
  7. Secondary and vendor sources. Booking Holdings’ second-quarter remarks come from an editor-reviewed transcript on Investing.com; Alphabet’s and Expedia’s from The Motley Fool. The check of AI booking tools was published by Engine, a business-travel booking company. The industry tracker quoting Altimeter Capital and Tripadvisor’s Instant Booking closure is a personal blog.
  8. Preprints and small samples. The Wang, Zhu and Zhang studies are preprints. The Seoul study’s leading-AI group is 23 people; the Skift and McKinsey diaries cover 26 travelers.
  9. Academic research looks backward. A review in Tourism Management of 170 empirical studies of generative AI in business and tourism (2022–2024) finds the field early and mostly quantitative; it offers research questions, not forecasts.[63] The forecasts in this report therefore come mainly from analysts, companies and research firms.
  10. Not found. A published method behind IDC’s 30% forecast; a forecast of AI-assistant booking revenue in travel; measured in-trip AI use; independent tests of current production trip planners; a scored archive of past travel-technology forecasts.

Change log

  • November 2026 edition
  • Added
    First edition, data as of 3 November 2026. Rescoped before publication from a study of today’s trip discovery to a five-year forecast.

About this report

Personal research by Chuhee Lee. Research, source verification and drafting were carried out with several AI models, including Claude (Anthropic), Gemini (Google) and ChatGPT (OpenAI). Every factual claim is sourced to the references listed below; topic selection, framing and conclusions are the author’s. These are personal views, not those of any employer.

References

Continue down the map

ForcesShiftsSignalsOpportunities
Follow new editions

Each report is refreshed on a schedule. Know when it changes.

Next refresh dates are listed on the home page.