The best travel agent is an AI algorithm

If you have ever consulted the web for trip advice, you have probably landed on Expedia’s website. The travel site offers travelers useful tools such as travel recommendations, smart shopping and price tracking to ensure that you get the best bang for your buck. You might be shocked to learn the recommendations aren’t actually made by humans.

“Whether it is about providing insights to our travelers, driving better personalized experiences for both our traveler needs as well as partner needs, everything – from our search, sort, our abilities to drive fraud and risk protection – is powered by AI and ML across the board,” says Expedia Group CTO Rathi Murthy. 

It turns out an AI-driven algorithm knows more about your travel preferences than a human ever could. 

“There is really a lot of complexity for a human eye to detect what our algorithms can,” says Murthy. “There’s many signals on the traveler side, as well as the partner side that our models are able to catch.”

Expedia’s platform processes over 600 billion AI predictions a year, which are powered by 70 petabytes of data. To put in perspective just how massive that amount of data is, one petabyte is equal to 20 million tall filing cabinets, according to Teradata. 

The data that Expedia collects to learn about your personal preferences comes from your profile, your choices and most importantly patterns in your behavior. The company also uses supplier data to help give you the most useful and accurate recommendations.

Also: Metaverse, AI and ‘super apps’: Watch out for these top tech trends in 2023 

“We have loads of data that we can power to drive those experiences for our travelers and partners in a way no one else actually is positioned to. And everything we do today is really on harnessing that data, AI and ML,” says Murthy.  

To transform this data into products and services and develop ML models, Expedia has over 6,000 software developers, data scientists and designers. 

Essentially, every keystroke and mouse move on the site helps Expedia better understand your preferences and deliver features like price tracking, predictions, and one of its latest rollouts, guest experience scores. 

Unveiled in May 2022, guest experience scores take signals from customer reviews, experiences and more to give hotel properties a score. That score drives how properties are presented to site visitors and the visibility of the property. 

“This really takes into account a bunch of signals, like what our ratings and reviews are on this particular property,” says Murthy. “So a bunch of machine learning and mathematical models help us detect how good this particular supplier is, so we can provide more confidence to our travelers as they are booking.”

In addition to using AI and ML to bring better experiences to its customers, Expedia has also been able to harness the power of AI to reduce business costs. 

Also: How a real estate company is using AI to find and remove racist language from property documents 

A large part of Expedia’s platform is being able to provide customer service to its customers who might be experiencing troubles with their travels and need immediate assistance. This was especially true when the pandemic first hit and people had to cancel trips, flights or figure out if they were even able to travel. 

“During COVID times, we were getting a huge onslaught of contact center calls with flight cancellations, etc. So we took that moment to really build our AI/ML capabilities,” says Murthy.  

To improve its conversational capabilities, Expedia Group powers more than 29 million virtual conversations. This implementation has saved the company more than 40% in variable agent costs. 

“We’ve already seen savings of several millions of dollars just by leveraging our automation and AI/ML on the contact-center front,” says Murthy. 

Concerns with these types of chats typically come down to whether they are efficient and actually solve the problem. However, 60% of travelers are self-servicing using Expedia’s automated chat tool, and as more people use it, the smarter and more helpful it becomes. This applies to all of the company’s algorithms. 

“The more you use us, the more you tell us about you, the better we get with our models,” says Murthy. “We have a continuous learning intelligence service that makes our AI and machine-learning models better and smarter every day, so we can continue to personalize your experiences.”