What Travellers Actually Search For in 2025 — And How AI Agents Should Answer

What Travellers Actually Search For in 2025 — And How AI Agents Should Answer

Across Google, app stores and social chat, travellers are no longer typing “Paris hotels”. They’re asking agents to do things: build a 3–day food itinerary, reroute a storm-hit trip, or find the lowest-emissions flight within a budget. This post distils the most searched-for intents around AI agents in travel and tourism in 2025, and turns them into practical product and content plays for OTAs, hotels and destinations.

Why search intent has shifted from “inform me” to “solve it”

• AI answers are now in the results. Google’s AI Overviews appear on more than 13% of queries, with a measurable rise in zero-click behaviour — meaning users get a task-level answer without visiting a site. Brands need to be the data that powers those answers, not just the link beneath them. (semrush.com)
• Searchers phrase needs conversationally. Skyscanner reports rapid adoption of open-ended prompts in its Savvy Search surface (e.g., “a budget city break in October with great food”) as travellers expect an AI concierge, not a list. (skyscanner.com, ttgasia.com)
• Trip planning is collapsing into a single agent. Expedia launched Romie, a persistent AI travel companion that plans, books and re-plans in one flow — a strong signal of where consumer expectations are heading. (expediagroup.com, hoteldive.com)
• AI use is highest in mobile-first markets. Skyscanner’s 2025 planning trends highlight Singapore, India, South Korea, the UAE, Saudi Arabia and Brazil as the most likely to use AI from discovery to booking — important for localisation and language coverage. (partners.skyscanner.net)
• Top-of-funnel travel interest is diversifying. Google Trends shows all-time highs for “solo travel” in 2025 and strong interest around seasonal, budget and experience-led prompts, which shape the questions travellers ask agents. (trends.withgoogle.com)

The most searched-for agent intents in 2025 (and how to win them)

The phrases below group real traveller behaviour into intents you can target with content, product surfaces and schema. Each intent includes: a representative query, evidence, and actions to capture demand.

1) “AI travel planner” / “best AI travel agent”

Example query: “AI travel planner for a foodie weekend in Lisbon under €500.”

Evidence: Major platforms now market named AI companions (Romie, Savvy Search, TripGenie), training users to ask for a planner rather than a destination. (expediagroup.com, skyscanner.com, trip.com)

• Act: Build landing pages that explicitly target “AI travel planner” + your vertical (family, foodie, outdoors). Include prompt recipes and show a sample itinerary output.
• Act: Mark up itineraries with structured data (activity, place, opening hours, price range) so they surface cleanly in AI Overviews. (semrush.com)

2) “AI itinerary” / “create my itinerary”

Example query: “Create my 10-day Japan itinerary with onsens and rail passes.”

Evidence: Trip.com’s TripGenie added collaborative editing and travel alerts — features travellers associate with “itinerary” not just “search”. (trip.com)

• Act: Offer editable, bookable itineraries with day blocks, time estimates and map embeds; allow instant substitution (e.g., swap museum for market).
• Act: Provide a friction-free export to Google Maps / Calendar for on-trip use.

3) “AI visa checker” / “entry rules now”

Example query: “Do I need a visa for Turkey with an Irish passport — AI visa checker.”

Evidence: Providers like Sherpa supply live visa and entry-rule data to airlines and OTAs, indicating high demand for authoritative, machine-readable guidance. (joinsherpa.com)

• Act: Integrate a verified entry-rules API and expose a clear, cite-back output inside your agent chat. Cache results with timestamp and source link.

4) “Lower-emissions flight” / “CO₂ filter”

Example query: “London to Barcelona lowest CO₂ flight this Friday.”

Evidence: Google’s Travel Impact Model calculates per-flight emissions and is now exposed via an API and even a Sheets add-on, enabling third-party agents to score routes. (support.google.com, workspace.google.com)

• Act: Add a “Lower-emissions first” toggle and display grams CO₂ per passenger with context labels (lower/typical/higher).
• Act: Offer rail-first alternatives when journey time is within a user-set tolerance.

5) “AI trip on WhatsApp / iMessage”

Example query: “Plan a surf weekend from Dublin on WhatsApp.”

Evidence: Persistent, chat-native assistants are the format consumers expect; Expedia positions Romie as a conversational “buddy”, while Skyscanner encourages natural-language requests. (expediagroup.com, skyscanner.com)

• Act: Ship a messaging surface with full parity: quote, book, pay, rebook and refund inside the thread; avoid “please visit our website”.

6) “Rebook me” / “flight cancelled what now”

Example query: “My 19:40 is cancelled — rebook me via rail and keep my sea-view room.”

Evidence: Platforms are moving to real-time disruption handling, with agents monitoring flights and pushing alternatives without manual calls. (hoteldive.com)

• Act: Connect flight status, hotel inventory and ground transport APIs; pre-authorise rules (max delay, price delta) so the agent can execute instantly.

7) “Passport-free / biometric check-in”

Example query: “Passport-free airports 2025” or “face scan hotel check-in”.

Evidence: Changi has introduced passport-free biometric clearance, and interest in document-free flows is rising as more hubs trial the model. (cntraveler.com)

• Act: If you operate hotels, pilot face-match check-in via a trusted ID partner and create landing pages explaining enrolment, privacy and opt-out.

8) “AI trip for solo travel / remote work”

Example query: “Solo travel July, walkable city, coworking and safe nightlife.”

Evidence: Google Trends flags an all-time high for “solo travel” searches in 2025, with intent skewing to safety, community and budget. (trends.withgoogle.com)

• Act: Build agent prompts that score destinations for walkability, public transport frequency and late-night safety guidance from official tourism and transit data.

9) “AI budget finder”

Example query: “Best beach for €350 total including flights from Manchester.”

Evidence: Consumers want total-trip budgeting, not component pricing; Savvy Search explicitly supports budget-bounded prompts. (skyscanner.com)

• Act: Return all-in prices with a spend breakdown (transport, room, meals) and sliders to trade time, cost and emissions.

10) “AI trip ideas near me”

Example query: “Weekend within three hours of Cork by train or ferry.”

Evidence: Local and regional discovery queries spiked post-pandemic and persist as cost-sensitive travellers seek short breaks; Google’s 2025 trends note strong seasonal discovery patterns. (blog.google)

• Act: Pair GTFS transit feeds with attractions calendars to propose timed micro-breaks that actually fit the timetable.

SEO + agent strategy: how to be the answer inside AI Overviews

With AI Overviews lifting “answer snippets” directly into the SERP, your aim is to seed authoritative, structured, cited information that models can trust.

1. Publish “prompt recipes”. Pages that teach users how to ask — e.g., “Budget-under-€500 prompts for a foodie city break” — earn links and are easy for AI Overviews to reference. (semrush.com)
2. Expose emissions and constraints as data, not prose. Use standard units (kg CO₂e/ passenger) aligned to Google’s TIM labels. (support.google.com)
3. Quote, book, rebook — in one surface. Travellers expect the agent to finish the task. Expedia’s messaging around Romie sets this bar publicly. (hoteldive.com)
4. Localise for AI-ready markets. Prioritise languages and payment methods for the markets Skyscanner highlights as most AI-adoptive. (partners.skyscanner.net)

Content blueprint: pages you should publish this quarter

Product requirements for an agent that actually converts

• Constraint-aware planning. Every prompt should accept budget, time, emissions and visa constraints — and show the trade-offs transparently.
• One itinerary, many channels. The same plan must live in web, app and WhatsApp with full edit, pay and rebook parity.
• Explainability. As AI Overviews expand, users expect citations. Show sources inline (menus, museum hours, emissions factors) to build trust. (semrush.com)
• Edge fallbacks. On-device inference keeps the agent responsive when travellers are roaming or offline.

Measurement: the KPIs that prove your agent works

1. Containment rate: % of conversations completed without human intervention — segmented by intent.
2. End-to-end conversion: From first prompt to paid booking. Benchmark uplift against traditional funnel.
3. Time to resolution: Especially for “rebook me” and visa queries.
4. CO₂ delta per booking: Track emissions improvement when rail or lower-emission flights are selected. (support.google.com)
5. AI Overview visibility: Monitor inclusion rate of your brand in AI answers for priority intents. (semrush.com)

Putting it together: a sample prompt library for your website

• Budget & time: “Plan a 2-night family trip within €600, flying from Dublin after 18:00 on Fridays.”
• Emissions-first: “Three-day break from London with ≤ 120 kg CO₂e per person — rail preferred.” (support.google.com)
• Visa-sensitive: “Seven days in Vietnam for Irish passport holders — entry rules and eVisa timing.” (joinsherpa.com)
• Disruption-ready: “Auto-rebook me on rail if my flight is delayed more than 120 minutes; keep a sea-view room if possible.” (hoteldive.com)
• Solo-safe: “Solo city break in October with walkable food districts and late-night metro service.” (trends.withgoogle.com)

Conclusion: be the agent travellers are already searching for

Search behaviour in 2025 tells a clear story: travellers want decisions, not pages. The winners will design for the top agent intents — AI planner, itinerary builder, visa checker, emissions filter, disruption autopilot — and back them with verifiable data, instant execution and explainable outputs.

Treat every page, API and chat surface as fuel for two audiences: your customer and the AI engines composing answers above the fold. If you can prove relevance to both, you won’t just rank; you’ll be the plan