Expedia Prepares for Multiple AI Travel Search Futures

Expedia Group is preparing for more than one version of AI-driven travel search. During an executive interview at Phocuswright’s Travel Marketing AI Summit London, Daniel Shin Un Kang, the company’s head of organic and agentic search, outlined an approach built around multiple possible outcomes. Expedia also has a dedicated business-to-agent team working to influence AI agents to act as ambassadors rather than gatekeepers, according to PhocusWire.

“The mistake is people are trying to predict the one outcome that's going to happen. I think you can get away with predicting a range of futures,” said Daniel Shin Un Kang, head of organic and agentic search at Expedia Group, according to PhocusWire.

The report, published Sept. 22, describes a company strategy for an unsettled search environment, not the launch of a new consumer booking service. Its significance lies in where Expedia is directing its attention: toward both the traveler making a decision and the AI agent that may help shape it.

Expedia’s AI search strategy has two audiences

Kang described Expedia’s approach as a dual-audience framework, according to PhocusWire. In that framework, the travel decision process is split between consumers and AI agents. Marketing therefore has to address both, rather than treating the person searching as the only audience that matters.

He compared the concept with marketing chocolate to a child and a parent. The product stays the same, but taste may appeal to the child while price or nutritional value matters to the parent. His point was that addressing two audiences is familiar territory for marketers, even if the second audience in travel search is becoming very different.

AI agents, however, do not behave like people, Kang said. Nor do they behave alike. According to PhocusWire, he emphasized that even basic differences between models can produce different decisions and behavior. He also cautioned that agents can be more biased than humans.

That distinction is central to his argument against choosing a single predicted future. Expedia’s stated goal is to influence how agents represent travel options, but Kang’s account rejects the idea that there is one uniform AI audience to satisfy. The dedicated business-to-agent team gives that work an organizational home; the dual-audience framework describes how Expedia is thinking about it.

Readable information is only the first hurdle

Some of the immediate challenges are technical. Kang described current agents as inconsistent and sometimes simplistic, including instances in which they cannot render particular file types, PhocusWire reported. Information that works for one model may not be handled the same way by another.

Structured data offers an early advantage in that environment, according to Kang. But he expects it to become a basic requirement rather than a lasting point of difference. In his view, the competition will eventually move beyond whether agents can read or discover a company’s information.

He expects the next contest to center on underlying facts and compelling value for users, with less opportunity for companies to hide behind complexity. That is Kang’s outlook, not a demonstrated result of Expedia’s strategy. Still, it establishes an important distinction in his remarks: making an offer accessible to an agent and making that offer worthwhile are separate tasks.

For travelers, the second task is the more consequential one. An agent’s ability to find information does not, by itself, establish the value of what it recommends. Kang’s own description of variable behavior and technical limitations makes that distinction worth keeping in view.

Travelers are using AI, but booking trust is different

Separate research provides context for why travel companies are paying attention without establishing that AI has displaced traditional planning channels. An account of Phocuswright’s updated European Consumer Travel Report, published by Federturismo Confindustria, describes changing travel research and purchasing behavior in the United Kingdom, France and Germany.

According to Federturismo, online travel agencies have overtaken general search as the leading online resource for planning different trip components in the report’s findings. Consumers also reported doing less research than before. AI remained secondary to traditional channels, but its use was growing rapidly and increasingly influencing destination choices.

The influence is already substantial among those using it. Federturismo reported that 88% of travelers using AI in Germany and 92% in France said it had influenced an actual travel decision. Those figures concern AI users, not all travelers, and should not be read as evidence that the same shares are handing over bookings to autonomous agents.

The same account identifies a boundary: human assistance still prevails in many situations, and a substantial share of travelers remains unconvinced about entrusting bookings to AI. Influencing a destination choice and securing permission to make a purchase are different levels of responsibility. The research supports keeping them separate.

A portfolio of bets, not one promised outcome

Kang’s response to uncertainty extends to how marketing teams work. According to PhocusWire, he argued that teams need greater technical capability and more comfort with failure. He compared the approach with venture capital, where investors allow for unsuccessful bets while positioning others for a large outcome.

His recommendation combines work that should remain useful across different futures with a portfolio of investments, rather than an all-or-nothing commitment to one forecast. That is the organizing idea behind the interview, not a promise that any particular agent or search model will win.

The traveler-facing effects remain unproven in the supplied reporting. For now, Expedia’s move is evidence of preparation: a dedicated team, a strategy addressing people and agents, and an explicit acknowledgment that the future of travel search is not settled.