Travel portal Expedia uses machine learning to deliver seamless customer experience, writes Izwan Ismail
THE idea of travelling is always exciting, but the thought of having to go through hundreds of travel websites, apps, airline portals, car rentals, tour agency sites as well as accommodation sites can be mind-boggling to many people.
People spend weeks and countless hours searching for the best travel-related itineraries, but many hardly get what they are hoping for.
One of the reasons for this could be the lack of information at some travel portals.
POWER OF DATA
With an understanding of what people face when they plan their holidays, travel portal Expedia has taken the next step, a high tech one, in the travel-booking process— through data science and machine learning.
The US-based global travel technology company, which aggregates fares and travel metasearch engines for Expedia.com, Hotels.com, Hotwire.com, CarRentals.com, CheapTickets, trivago, Venere.com, Travelocity, Orbitz and HomeAway has been investing heavily in technology and innovation since 2010 to make travel bookings easy.
Throughout the years the company has accumulated lots of customer feedback and data, which it now uses to provide “personalised” travel-related service.
Its Global Head of Mobile Apps Marketing and Asia Pacific Head of Marketing, Gabriel Garcia, says the purpose of Expedia is to help people go places.
“And how do we achieve that? We keep customers at the heart of innovation,” he said at a recent round roundtable discussion on the impact of technology usage on travel.
NEW ENABLER IN TRAVEL
Today’s technology advancements have opened doors to build intelligent, responsive algorithms that yield travel results catered to individual customers.
Common booking issues among travellers, like deciding how to choose a hotel based on its amenities, location, size and price can be addressed using algorithms built on classical statistics.
And as the travel industry gathers more information about booking trends, it has become clear that investing in algorithms based on machine learning is the most effective path to discover and present the best offers to customers.
Garcia, who is responsible for Brand Expedia’s Mobile App growth and adoption strategy across its 33 markets, says data has become the new enabler in travel, and Expedia is banking on this to give travel information that suits travel needs and expectations.
“We always ask what are the key customer problems, are we selling the right things, how do we make the search as seamless as possible, are we serving the right ads to the right customers in the right context, and are we giving the best lodging experience?
“From looking at places to the booking process, we want to make it smooth and seamless with the use of technology.
“At the end of the day, people don’t necessarily travel to go to a place, but for the outcome.”
DATA MINING
Over the years, companies like Expedia have collected 1,500 terabytes of data to fill 6.7 billion, 200-page books with 189 billion rows of consumer behaviour.
And with that much data, Garcia says state-of-the-art data mining system and machine learning can be applied to make travel plans smoother.
In its efforts to understand customers’ behaviour better and know their expectations of travel, Expedia uses research labs in Bellevue, Washington; London; and the newest one in Singapore to conduct studies on customers’ reactions when using its portal and mobile app.
“The fundamental thing in understanding customers is by being closer to them and we are very fortunate to have innovation labs around the world,” says Garcia.
At these labs, Expedia uses technologies like electromyography (EMG) and eye tracking glasses to better understand customer behaviour.
EMG measures emotional impact on customers, while the eye-tracking glasses used in the experiment enable a researcher to see what a test subject is looking at, whether it’s on a desktop computer or a mobile phone.
The experiment also involves many sensors placed on the subject’s face like forehead to read feelings of confusion or tension. Sensors on the cheeks read muscles that form smiles enable researchers to recognise when a user is responding positively to something.
“Through the sensors and technology, we are able to understand what delights and frustrates customers when doing certain actions at our website and on the app,” adds Garcia.
“We can tag skylines, sunrise/sunset, skyscrapers and see how these affect decisions.”
Expedia has carried out some 1,500 tests.
TWEAKING PRODUCTS
From the EMG lab results, Expedia is able to make small changes to its site to ensure it’s meeting the needs of local travellers.
This has enabled it to provide its partner hotels, for example, some guides on the types of photos people like to see at a hotel website.
Tests on Asian customers have revealed different preferences and expectations. For example, customers in Japan love visual aids at hotel websites and expect bathtubs and breakfast. A test among Indian customers, meanwhile, reveals that they like to know food options at the hotel while Malaysian customers look out for promotions and shopping options.
“With all these reactions from customers, we are able to give specific group of travellers information that matters to them, and will make their travel plans easier,” says Garcia.
“We use the machine learning algorithms to build customer data, look at user patterns and analyse the results, and give the intelligence to customers for them to make the best decision for their travels.
“Through our frequent test and learn process, we are able to make localised tweaks to country sites to better suit travellers from that country, making it easier for local travellers to plan and book,” he adds.
HOW IT WORKS
There are millions of ways to get to one place.
“At any given point, a flight from point A to point B can give you millions of search options. So how do we solve this matching problem by giving our travellers the best options, with the fastest routes, and cheapest fares, in a matter of 3 seconds? So how do we solve this matching problem? And how do we get the best rate for customers and the best, fastest route possible?
“We use data science and machine learning to solve this problem,” says Garcia.
“For example, if you’re looking at a trip from Singapore to Tokyo on any given day, you may find 30 to 50 direct flights.
But if we add one stop to that route, that number becomes a billion options, and if we add two stops, the number becomes trillions, and we are able to calculate all these actions in three seconds,” he adds.
Expedia’s Best Fare Search system sorts through millions of itineraries and flights in a few seconds and narrows them down to 10,000 most relevant ones and further narrows them down to 1,000 and finally to a few most relevant to the customer.
MOBILE FUTURE
Expedia invested US$1.2 billion (RM5 billion) in technology over the past few years and a big chunk of it goes to mobile development.
There is an acknowledgement within Expedia from a research and development perspective and with the ongoing evolution of its platform that it will need to shift its focus to become mobile first.
“Globally, we’ve seen significant growth in the use of mobile app with more than 250 million Expedia app downloaded to date. Today, 1 in 3 Expedia bookings are done on Mobile, with more than 50 percent of that coming from Asia,” says Garcia.
The mobile app is perfect for delivering a seamless customer experience, such as notifying travellers on real-time flight changes or cancellations. “We want to be a digital concierge and the app helps drive engagement and utility,” he says.
To encourage bookings on the mobile app, Expedia offers more loyalty points, among other benefits.
The mobile app opens new ways to interact with customers, including providing them with push notifications, with details on flight status, for example.
Expedia is also pursuing virtual and augmented reality.
“More and more of our customers are millennial who welcome a VR experience to view the places before going there.”
In the last two years, Expedia has introduced chatbots through Facebook Messenger and LINE in Thailand.
In Asia, people are more comfortable making a booking with their smartphones and about 60 percent of voice searches on Google Assistant are made using natural language, a sign that humans are getting more comfortable talking to machines,
“We will keep enhancing our travel related processes and with technology, we will continue to focus on customer experience, giving them the best experience,” concludes Garcia.