Main content

Expedia is investing billions in data to create personalized 'travel-graphs'

Derek du Preez Profile picture for user ddpreez March 23, 2014
The online travel industry is going to compete on data in the future, not price

Like many others within the retail sector, the travel industry has been transformed by consumers having cheap, pervasive access to the internet. Instead of hitting the high-streets and shopping malls to find agents to help book flights, hotels and car rentals, consumers have taken to online alternatives that offer more choice and cheaper alternatives. Up until recently online travel agents have been offering a fairly commoditised service in exchange for discounted services – however, Expedia, the largest of its kind, is looking to change this with huge investments in data analysis to create personalized recommendations for its users. It hopes that this will drive proper brand engagement.

Andrew Warner, Expedia's senior marketing director for EMEA, was speaking recently at a technology marketing conference in London, where he started a session by asking delegates: “What the f*ck is big data?”. A question most of you reading this have asked ourselves a number of times, I'm sure. Expedia has approximately 150 sites that operate in 70 countries around the world, with 50 million visitors a month and 200 mobile app downloads a minute, so this is a company with plenty of data.

To give you an idea of how much Expedia is investing to take advantage of this data, last year it spent £500 million in research and

Andrew Warner Expedia
development, as well £2 billion on marketing.

Warner said:

“At our heart we are a technology company, we are a travel retailer that invests heavily in technology to make travelling better for consumers.

“What this all means is that we have a lot of data. But what do we do with it and why does it matter? We can't just sit around theorising about the future, we have to address stuff in real time and we have to serve customers in real time, 24 hours a day, as best we can.” 

Warner wants Expedia to go back to an old fashioned way of doing business. He said online retailers have largely used advancements in retail technology to take costs out of transactions, which has led to a commoditised service and a low barrier to entry. However, he wants Expedia to be a company that knows its customers so well that when they come to search for travel options, it can largely predict what they need.

He said:

“The metaphor I use is that online travel agents are a bit like ticket machines, you punch in where you want to go, a price comes up, your ticket comes out and off you go. If it's not what you are after you move on to the next one. 

“Imagine going back to a time when people know you. All of this data we get from people should enable us to anticipate the needs of consumers - to personalise, target, give consumers stuff that is contextually relevant.”

The travel graph - a £100 billion a year opportunity

Expedia believes that online travel agents have traditionally forced consumers into a very fixed format of search, because the engines have been formatted and based on an interface that was designed for travel agents to use, rather than consumers. Warner argues that, however, when you are travelling for holiday, as a family, or for business, a lot of the time you don't know what you are looking for and consumers rely upon the results provided by the travel agent – but these aren't personalised to each individual.

“That's something we need to change. We want to be able to show you stuff that you need that's useful for you, it's relevant before you have even asked for it. 

“We reckon that if we can crack that, where we know what customers want when looking for a holiday, that's an opportunity of £100 billion a year.”

delta plane
Expedia aims to tap into this £100 billion a year opportunity by building a travel graph for each of its customers. Warner said that by structuring the data in a certain way, Expedia can get to know its users very well. He explained that Facebook and LinkedIn have been doing this for a while, where they constantly prompt their users to fill in fields and answer questions about themselves. By structuring the information they are willing to share, Expedia is able to make recommendations which feel personal to each user – rather than returning a standardised result for whoever is searching.

Warner said that for a standard trip from LA to New York, Expedia has 65,000,000,000 different combinations of travel for each consumer – given variations in flight times, airlines, car rentals, hotels, offers. The travel graph will help lift the most important results to the top, based on the requirements of each person searching.

“That number is massive and of course no consumer is going to be able to search through such a huge number, and so we have to give them the most relevant things as quickly as possible, make it as relevant as possible, to hopefully get that conversion and make that booking.

“What we realised is that all of this information that we are getting from consumers allows us to do what LinkedIn and Facebook do, build a travel graph. We can actually start to see what people book, what the signals are, the combinations that they use, to build this really clever comprehensive algorithm that enables us to give control to the consumer. 

“So we see returns by structuring the data better. This gives consumers relevant information in the right context. This is a fundamentally different way that we talk about the product, a very simple trip that we talked about earlier (LA to New York) – there's lots of complication behind it, but we need to present it in a way that's right for the device that the consumer is using with the recommendations that's right for them. There's also opportunities there for up-selling if we get this right.”

Building a dynamic, real-time itinerary 

Expedia had previously been running on Microsoft systems, which Warner described as “pretty crappy” and a legacy platform that was build for a previous digital age, the dot com era. As a result, Expedia has fundamentally reengineered its platform in-house (although Warner wasn't specific about which platforms Expedia is now using) to enable the company to structure this data, but to also be able to continually test responses to changes on Expedia sites. To get an idea about how much this has been ramped up, Expedia did 30 times more tests in 2012 than it did between the whole of 2005 and 2011. This doubled in 2013. (I've previously written on this in more detail at Computerworld UK – see here).

“We can now try stuff out. Different types of consumers, different presentations of our product, on different devices – we can measure how it converts, we can measure their feedback and then we can start to really refine this. If you are a small digital travel agent you just cant do this because you don't have enough scale. You don't have enough people to be able to do this fast enough.”

“What that then enables us to do is change the way that we present our business to consumers. Search starts to become more intuitive, if we structure the data in the right way we can allow people broader search terms and get them stuff more quickly – we require less of them up front. This is stuff that is happening today, if you go to our US site that has more of this stuff already implemented, it also allows us to provide greater utility to the consumer. This is proper brand engagement.”

Some of the new things that Expedia is trying out include dynamic recommendations, such as telling a consumer that if they stay in a location for two days longer they save £50, or if they get a flight leaving 10 minutes earlier they can save some money, or how many seats there are


left on a flight. “Really useful stuff that helps them make better decisions,” said Warner.

A click to chat functionality is also being introduced, where consumers searching for a trip can speak to an agent if they need to – not only this, if that consumer has been using Expedia and developed a comprehensive travel graph, they will be instantly routed to an agent that deals with the type of trip that they are looking for.

“Travel is more like a stock market than a static market nowadays, so that's why sometimes when you search flight price and you go through to the booking page and suddenly it changes because they are changing all the time. 

“We can start to surface that data for consumers – we can start to show them what has been booked the most for the type of service they are looking for, where's busy, where demanding is increasing, what other people like them are booking. This enables the consumer to understand what is going on in the market.”

However, this isn't just for the desktop. If an Expedia customer is using its mobile to search for a hotel, for example, Expedia will instantly show them hotels that are close to them and available for that same night – because that is typically what someone searching on a mobile for a hotel ends up booking. Also, if they are typically a price conscious consumer (based on information on their travel graph), Expedia will surface cheaper hotels to the top of the results.

What Expedia is hoping is that it can create a dynamic itinerary for customers that follows them around, after they have booked all their travel needs.

“The itinerary is where it starts to get exciting. This is where we can start stretching that brand engagement out with them in their pocket, providing genuine utility based on the dates that we have to enable them to have a better travel experience. We present your itinerary to you in a dynamic way, in chronological order. 

“You can make changes to a booking during your trip, we can give you alerts during your trip, such as a flight delay or what gate you need to be at, we will tell you what time you need to check out, whether there is a traffic delay on the way to the airport, etc. We can do this dynamically, we know where you are, we know what you've booked. 

“In the future, when we know you are in a certain city, but we also know it's raining, we want to push out an offer for the aquarium. Other brands aren't doing this because they don't have that R&D investment. This allows us to surprise people and excite people.”


This is really innovative stuff and the level of commitment to data (as well as the level of spend) is impressive. Expedia is really trying to create a personalised experience for its customers, one that is with you wherever you go, on every device and delivering useful information in real-time.

However, it will be interesting to see how many consumers buy into this experience. I get the feeling a lot of people using online services for travel are largely doing so because it is cheaper. I don't book a flight with Expedia, or SkyScanner, or whoever, because they know what time I like to fly or which airline I prefer – I can figure that out – I book with a website because they are the cheapest. But maybe that's just me.

However, the dynamic itinerary that travels with you after you have booked your trip is an interesting proposition, if it works and if it is useful. Customers may feel real engagement with Expedia if they feel like they are getting a better deal in the long run. Expedia is betting on data, but only time will tell if this beats price.

A grey colored placeholder image