The booking funnel is one of the most optimized digital experiences ever built. The trip itself has never been designed at all. That gap is about to become a competitive problem.
A note on perspective: Travel technology has been optimized around a single transaction for its entire digital existence: booking. The experience of the actual trip, navigating a disrupted itinerary, managing a multi-leg journey across airlines and hotels, knowing what to do when something goes wrong at 11pm in a city you do not live in, has never received the same design investment as the funnel that sold the seat. That is about to matter in a way it has not before. AI travel agents are arriving as a new user type, and IATA’s NDC transition is forcing a rebuild of the data infrastructure that supports travel booking anyway. The CEO who reads this as a technology migration will complete the migration and maintain the funnel.
The travel booking funnel is extraordinary. Decades of search advertising, conversion optimization, A/B testing, and pricing algorithm development have produced a system that finds the right traveler, surfaces the right options, and completes the transaction with remarkable efficiency. The funnel knows when you are likely to book, what price will close the decision, and which seat upgrade you are statistically likely to add.
What happens after the booking confirmation email lands is a different story.
The traveler whose 7am flight is canceled at 5:30am is interacting with a system that was built for airline operations, not for the person who needs to get somewhere. The traveler whose hotel has no record of a reservation made through a third-party OTA is navigating the gap between two separate systems that were never designed to communicate with each other on the traveler’s behalf. The traveler managing a four-city itinerary across two airlines, three hotels, and a rental car is holding together a plan in their own head because no system they can access has the full picture.
This is not a customer service problem. It is a design problem. The travel industry has spent 25 years building the experience of choosing and buying a trip. It has never seriously designed the experience of taking one.
That design debt is coming due in 2027, for two reasons that are arriving simultaneously.
The Infrastructure Is Being Rebuilt Anyway
The global travel booking system runs on infrastructure that is older than the commercial internet.
The Global Distribution Systems, Sabre, Amadeus, and Travelport, were built as mainframe-era inventory management tools connecting airlines to travel agents in the 1960s and 1970s. They became the backbone of how travel inventory is sold: airlines publish fares into the GDS, travel agents and online booking tools pull from them, and the transaction completes through a protocol that predates the web by two decades.
The GDS is the HRIS of travel. It reflects the airline’s inventory categories, not the traveler’s journey. It was built for the people administering it, not for the people traveling inside it.
IATA’s New Distribution Capability standard, NDC, is an IATA-defined API protocol that enables airlines to sell travel services directly through modern data connections rather than through legacy GDS screens. NDC has been in development for years, but 2027 is the year the transition reaches a threshold that changes the market. Major carriers including American, Delta, United, Lufthansa, and British Airways have been progressively moving their best inventory, enhanced fares, seat upgrades, ancillaries, and bundled offers, to NDC-connected channels. OTAs and travel agencies that have not completed NDC integrations are seeing degraded inventory: worse prices, less availability, ancillaries that do not exist in the GDS booking path.
This means every organization in the travel distribution chain is rebuilding its data connections to the airline inventory layer right now, whether it wants to or not.
The compliance version of that rebuild is a technical migration: replace legacy GDS integrations with NDC API connections, maintain the booking funnel on top of the new data layer, capture the improved inventory access, and move on. That version preserves the existing experience with better data underneath it.
The design version asks a different question: now that the data infrastructure is being rebuilt anyway, what experience should it support?
What the AI Travel Agent Actually Needs
At the same time the booking infrastructure is being rebuilt, a new user type is arriving in the travel market.
AI travel agents, purpose-built planning and booking assistants in tools like Google Travel, Apple Intelligence, Booking.com’s AI planner, and standalone services, are beginning to handle the full travel planning and booking workflow on behalf of a traveler who states a goal and authorizes the agent to execute it. This is not a better search experience. It is a different kind of user, one that queries structured travel data, compares options across criteria the traveler specified, books the trip, and monitors the itinerary through completion.
That user has completely different information needs than the human browser the booking funnel was built for.
A human browser evaluating a flight needs price, departure time, duration, and stops. The marketing and social proof signals that drive booking conversion, the photography, the customer reviews, the upgrade prompts, are all designed for the human psychology of the purchase decision.
An AI travel agent evaluating a flight on behalf of a traveler needs: schedule reliability data, change and cancellation policy in machine-readable form, baggage allowance structured precisely, seat selection availability by category, historical on-time performance, connection viability given airport layout and typical transit time, and the ability to access the traveler’s itinerary context to understand what this flight connects to.
Most of that information is either unavailable through booking APIs, buried in unstructured text, or inconsistently formatted across carriers. The agent that is trying to book the best option for a traveler with a 55-minute connection in a hub it knows is tight cannot reliably access the data it needs to make that call. The result is a recommendation based on incomplete information, or a refusal to recommend at all.
The carriers and OTAs whose data is clean, structured, and agent-readable will exist in the AI agent’s decision set. The ones whose data lives in marketing copy and legacy GDS text fields will be skipped, not because their product is worse but because the agent cannot reliably reason about it.
This is the retail agent-mediated commerce problem, applied to travel. And the NDC transition is the moment when the data layer gets rebuilt anyway, which makes it the best available window to build it right for both user types simultaneously.
The Two Travel Strategies for 2027
There are two ways a travel organization can respond to the converging pressures of NDC migration and AI agent adoption.
The first strategy is funnel maintenance.
The organization completes the NDC migration because the inventory access requires it. The booking funnel is preserved on top of the new data connections. The experience the traveler sees is largely unchanged. AI features are added to the funnel itself: a chatbot that answers booking questions, a personalization layer that surfaces relevant options faster, a disruption notification that tells the traveler their flight is delayed.
This strategy produces a faster, more capable version of the experience that already exists. The traveler whose flight is canceled still has to navigate a rebooking flow designed around the airline’s operations workflow. The traveler whose itinerary is disrupted still has to call customer service or manage the rebooking themselves. The AI agent trying to book on behalf of a traveler still encounters structured data that is 60 percent complete for what it needs.
The funnel maintenance strategy completes the technical migration and misses the design opportunity the migration created.
The second strategy is trip design.
The organization makes a different bet: the NDC migration is the moment to build the data infrastructure for the full journey, not just the booking transaction. That means treating the itinerary as a persistent, structured data asset that travels with the traveler through every stage: planning, booking, pre-trip, in-trip, and post-trip. It means building agent-readable product data for flights, hotels, and ground transportation that is complete enough for an AI agent to reason about reliably. And it means designing the in-trip experience, the disruption management, the itinerary adjustment, the real-time support, with the same investment the booking funnel has historically received.
None of this abandons the booking funnel. It extends the design investment to the part of the product the customer is actually inside when the relationship matters most.
What the CUX Officer Has to Build That the Product Team Cannot
The Chief User Experience Officer in a travel organization facing this transition has a mandate that the booking optimization team and the technology migration team cannot fulfill alone.
The booking team is expert in conversion. They know what drives revenue per session, which design patterns reduce abandonment, how to sequence upsell offers to maximize attachment. They are building for the funnel.
The technology team is expert in the NDC migration. They know which APIs to connect, how to map legacy GDS data to NDC schemas, what the integration testing looks like. They are building for the data layer.
What neither team is building is the trip experience: the design of what happens to the traveler after the confirmation email, organized around the traveler’s perspective rather than the airline’s or hotel’s operational workflow.
That is a CUX problem. And it is not a problem that gets solved by a better disruption notification. The notification that tells you your flight is canceled is not an experience design. The experience design is what happens next: who handles the rebooking, what options the traveler sees and how quickly, what the agent does on the traveler’s behalf if one is authorized to act, and what the traveler has to do themselves if one is not.
Three problems the CUX officer has to solve that the booking optimization team cannot.
The first is itinerary data architecture. The traveler’s full itinerary, flights, hotels, ground transportation, reservations, and the connections and timing dependencies between them, needs to exist as a structured data asset that can be accessed by both the traveler’s own AI agent and the travel organization’s customer support systems. Most travel organizations hold fragments of this data in separate systems that cannot communicate. Building the itinerary as a first-class data asset is a design and product decision, not a technology decision, and it requires organizational alignment that the product team cannot achieve without authority behind it.
The second is disruption design. Travel disruption is the moment when the relationship between the traveler and the travel organization is most consequential. It is also the moment that has received the least design investment. The experience of a disrupted itinerary, what the traveler sees, what options they have, how quickly those options are presented, what happens automatically on their behalf versus what requires their action, needs to be designed for the traveler’s situation, not for the airline’s rebooking workflow. That design exists for some premium product tiers and in some emerging AI concierge services. It does not exist as a standard experience for ordinary travelers.
The third is the agent authorization experience. As AI travel agents become a primary booking channel for a growing portion of high-frequency travelers, the question of what an agent can do on behalf of a traveler, and what the traveler can see and control about that authorization, becomes a trust and design problem. The traveler who authorizes an AI agent to manage their travel needs an experience that makes the scope of that authorization visible, the agent’s actions transparent, and the human override accessible when needed. That experience does not currently exist in any mainstream travel product. Building it requires the CUX officer’s involvement from the authorization design through the in-trip oversight interface.
What the Organizations That Move First Will Have
By 2028, the travel organizations that treated the NDC transition as a design mandate rather than a technical migration will have something the ones who maintained the funnel cannot replicate quickly.
They will have an itinerary product that handles disruption with the same efficiency the booking funnel handles conversion. The traveler whose flight is canceled at 5:30am will have been rebooked on the best available option before they open the app, with the change reflected across every component of their itinerary automatically. The traveler managing a complex multi-leg trip will have a single, accurate, real-time view of everything they have booked and everything that depends on what they have booked. That is not a feature. It is a product category that does not currently exist at scale.
They will have a data architecture that positions them for the agent-mediated travel booking wave that is already building. The carriers and OTAs whose inventory data is clean and agent-readable will capture booking volume from AI travel agents that the ones relying on legacy GDS data will miss. That volume will grow as AI travel agents become a standard tool for the high-frequency business travelers and experience-focused leisure travelers who represent disproportionate booking value.
And they will have built the infrastructure for every subsequent layer of travel AI, because an organization that designed the itinerary as a first-class data asset and built the agent authorization experience the first time does not need to retrofit those things when the next capability arrives.
The booking funnel is not going away. The assumption that it is the only experience worth building for is.
Sources: IATA, “NDC Standard and Program,” 2024-2026; Skift, “NDC Adoption Tracker: Where the Major Airlines Stand,” 2026; Phocuswire, “AI Travel Agents and the Future of Distribution,” 2026; McKinsey, “The Future of Travel: How AI Is Reshaping the Industry,” 2026; Deloitte, “2027 Travel and Hospitality Industry Outlook,” 2026; Amadeus, “NDC Transition Progress Report,” 2026; Forrester Research, “AI-Mediated Travel Booking: 2027 Forecast,” 2026; Gartner, “Hype Cycle for Travel and Hospitality Technology,” 2026.