Expedia poised to take flight with generative AI

Artificial intelligence is poised to disrupt the travel industry, and Expedia CTO Rathi Murthy, who oversees a massive online travel network with 168 million loyalty members and more than 50,000 business partners, believes her company is well positioned to capitalize.

That’s because, outside of its top brands, which include Travelocity, VRBO, Hotels.com, Orbitz, Trivago, Wotif, and CarRentals.com, the $14 billion online travel service’s most prized possession is its data — the 70 petabytes of traveler information stored on its AWS cloud.

“Data is our heartbeat,” says Murthy, who joined Expedia Group as CTO and president of Expedia Product and Technology two years ago after serving as Verizon Media’s CTO during the 5G buildup. “We have a very giant ecosystem and with that what we have been able to do is really drive personalized behaviors across for our travelers.”

And with the expansion and consumer popularization of AI fueled by recent advances such as ChatGPT, Expedia’s extensive use of analytics and machine learning to fuel that personalization strategy should enable the company to help evolve the travel industry, even as its pool of customers and partners grows, says Murthy.

“AI is a huge enabler for us. It’s not new to us, but ChatGPT democratizes AI for everybody and bringing the two together — data and AI — is very powerful,” she says, noting that Expedia currently runs roughly 600 billion AI predictions annually from its massive data repository.

Two-pronged transformation

The Seattle-based travel giant has come a long way since being founded as a division of Microsoft in 1996 and spun off as a public company in 1999. Expedia Group has more than 21 brands and more than 50,000 B2B partners connecting them with more than 3 million hotel and rental properties and more than 500 airlines and cruise companies, as well as car rental agencies.

Still, the online travel service has strong upside potential, Murthy says, estimating that just 20% of all travel professionals and consumers use online systems for booking reservations today.

According to 6sense Revenue AI numbers, Expedia has roughly 10% share of the reservation-and-online booking market and competes with more than 20 rivals, including TripAdvisor, which owns a 46% share of all revenues; Airbnb, which takes in 23%; and Booking.com, which holds about 7% of the online reservation market.

To help Expedia capitalize on the online travel bookings potential, Murthy has devised a two-pronged approach to best competitors and keep growing Expedia’s revenues, which surpassed $12 billion in 2022.

Her core strategy consists of two distinct but interrelated transformations: the ongoing internal transformation of Expedia’s infrastructure in a unified and integrated manner and leading the transformation of the online travel industry itself, the CTO says.

“The travel industry is riddled with a lot of legacy [technology and processes] and not everybody has the technology strengths and capabilities we have,” she says, noting that Expedia is trying to position itself as the travel industry’s base platform on which to help digitize travel agencies and bring them into the online fold.

“It’s not just about transforming ourselves. It’s transforming the travel industry,” she says. “Online travel agencies make up about 20% of the travel market so there’s the 80% that belongs to small players, offline players, airlines, hotel chains, directs bookings, etc.”

The Expedia Group has methodically developed a multi-phased plan for its digital transformation and began moving workloads to the AWS cloud in 2017 — before Murthy’s arrival.  Now, more than 90% of the company’s data is stored on AWS, she says.

Murthy’s mission comes at a very opportune moment — when commercial availability of AI is exploding and targeted at ordinary consumers. To date, AI platforms have been predominantly used within corporations to build machine learning models by data scientists, software engineers, and IT experts. But increasingly, SaaS vendors are incorporating conversational AI tools such as chatbots, as well as generative AI models enabled by platforms such as OpenAI’s ChatGPT, to democratize AI as a service to be used by any consumer.

That consumerization of AI is a trend that Murthy has a keen eye on, seeing it as something that will fuel further personalization opportunities in the travel industry.

Embracing the opportunity

To that end, Expedia recently incorporated two OpenAI plug-ins to expand its conversational capabilities for travelers within ChatGPT and the Expedia service itself.

One implementation is the Expedia ChatGPT plug-in that enables users who start planning a trip on ChatGPT to select the Expedia plug-in to bring that trip to life, delivering them into a seamless booking experience in Expedia, Murthy says. The other plug-in resides within the Expedia application and enables travelers to have “conversation capabilities” about all aspects of trip planning and save hotels recommended in that conversation to a new “trip” so they can easily go back and build out their trip. It is currently available on iOS devices in English.

“Marrying ChatGPT data [such as questions from travelers] with Expedia’s travel-specific data such as traveler preferences, booking patterns, and pricing availability is super powerful,” the CTO says, adding that the combination of technologies can create a full itinerary for consumers. “We’ve seen a lot of traffic. It’s been beautiful in terms of trip planning.”

The Expedia team was able to deploy the plug-ins within three to four weeks. “This is really because we’re so mature in our AI capabilities and the APIs in our travel platform,” Murthy says.

Others are speedily incorporating ChatGPT and GPT APIs into their services and applications, including CPG multinational Unilever and omnichannel used-car retailer CarMax.

But analysts suggest such early implementations of generative AI have a way to go before delivering meaningful impact.

“At this early stage, the ChatGPT feature within Expedia’s app is limited. It doesn’t have access to real-time pricing information and users are not able to book through the feature. It also only saves hotel recommendations currently into the trip section of the Expedia app,” says Brad Jashinsky, director analyst in the Gartner for Marketers practice.

Jashinsky also cautions that it is not a certainty that consumers will use chat-powered AI systems for travel in concert with any online booking system, but if so, “it could differentiate Expedia’s experience from other online travel agencies.” 

IDC analyst Dorothy Creamer suggests Expedia is in good position for the AI era if the company works closely with partners.

“Expedia is building out its data platform on 25 years of travel and booking behavior and this level and depth of data is vital to drive a more personalized booking and journey experience in the post-pandemic era,” Creamer says, noting that the company is also “building a deep bench of data scientists to help mine value from booking data — these data points require evaluation of thousands of permutations per a single page of booking information.”

Creamer also notes that Expedia’s products are being built to provide hotels, airlines, car rentals agencies, and other partners with the building blocks to process deep customer insights and marry that with actual traveler behavior. “This will require the sophisticated analysis of data that AI and ML-driven functions and platforms can provide,” she says.

Still, Expedia’s Murthy remains fully aware of the ethical concerns surrounding AI and wants to ensure that Expedia is a fair and free marketplace for all users.

“We are keeping a very close eye on its governance and ethics and making sure there’s privacy considerations on sharing any of the information,” Murthy says. “That’s something all of us have to watch out for and work together as a concrete community to use AI for good and make sure we can use best practices across the industry.”

Artificial Intelligence, Data Management, Digital Transformation, Travel and Hospitality Industry