Expedia spends big money on predictive analytics search engines

Gary Morrison explained how high-speed testing can boost returns for the company

Expedia is spending over US$500 million ($521 million) a year on its in-house technology, most of which is being ploughed into new platforms that allow the company to better predict itineraries for those searching for travel options on its website.

It believes that high-speed testing and advanced analytics will allow it to better understand what consumers want when booking trips, rather than trying to boost revenues through flash deals.

Computerworld UK spoke to Expedia's head of retail, Gary Morrison, who has also previously worked for Motorola, Dell and Google. He explained the complexities behind delivering tailored results for a traveller.

"Creating travel recommendations for consumers is hard. Let's say you wanted to do a return trip, for a day, between JFK and LAX. There are eight airlines that will do that trip, which gives you about 8,000 permutations as travel options," said Morrison.

"Now let's imagine that you were okay having a layover at an airport between the two - now the number of permutations for that same day trip is about five billion. If you also say that you don't mind flying out of Newark, instead of JFK, and you don't mind landing at an airport near LAX, the permutations jump to 65 billion. This is before you've even added a hotel!"

Test, test, test

According to Morrison, there are only two companies in the world that have the technology to take those permutations and create a list of itineraries - Expedia and ITA (which is owned by Google). Expedia has invested massively in a strategic direction that relies on using a Hadoop platform that allows it to carry out a high number of tests at great speed and implement changes rapidly.

He said that tests are being carried out on even the smallest details on the website.

"In order to do this you need to have a platform that will allow you to test and learn with massive amounts of velocity," said Morrison.

"For example, if you have a page with a 'book now' button on it that is blue and you have another page exactly the same but has a 'book now' button that is yellow - we could direct 50 percent of our traffic to each site and test to see if more people bought trips on the blue or yellow button.

"If the preponderance of people that book yellow is higher, you would leave the yellow button there."

Morrison said that the volume of testing is important, as Expedia relies on lots of 'small successes' to make a difference to revenues.

"We rigorously test every piece of our experience, and the platform we have been investing in the past two years allows us to substantially increase the number of tests."

Big data = personalised search

In addition to the rigorous testing, Expedia is using the platform to improve its predictive analytics.

Morrison believes the 'daily deals' that have dominated the online travel industry in the past are not targeted and lack structure. He is now keen to create 'meaningful recommendations' by analysing how people navigate choices that are offered to them on the website.

Why predictive analytics matters
6 practical predictive analytics tools
How predictive analytics is tackling customer attrition at American Express

"For us, this is what we call big data. If you take myself, I am a father with three kids, but when on Expedia I may be going on a family vacation, a golfing trip, a business trip - the ability to figure out whether personal recommendations are relevant to me really depends on what I'm trying to do," he said.

Expedia is analysing, in an anonymous fashion, how people interact with all of its products - across the website, mobile app and tablet app. It is also integrating data collected from over 10 million reviews, which are relevant to hotels, airports, and points of interest.

"So if we analyse how people choose what is offered and what they book, integrate that into our algorithms, we can start to identify patterns. If we can then relay those patterns back to the next consumer that comes on the site, it is going to be that much more useful, people will book more," said Morrison.

"Our focus isn't on collecting personal data, it's on what people research, what they ultimately book, and then identifying the pattern."

He added: "Once that pattern is identified and we start relaying those searches to consumers, we end up with a much better product. It also allows us to work with suppliers to see if we can get some preferential rates on bookings."

Last year Expedia complained to Europe's competition regulators about Google, which is accused of using its search engine to direct users to its own services (ITA) and reducing the visibility of competing websites and services. Expedia says that Google's flight-search service, which was launched last year, "excludes any link to online travel agencies."

Join the CMO newsletter!

Error: Please check your email address.
Show Comments

Supporting Association

Blog Posts

Top tips to uncovering consumer insights for business innovation

An in-depth understanding of consumers sits at the heart of what we all need to do, but we know it’s not always easy to uncover insights that will unlock a true innovation opportunity.

Matt Whale

Managing director, How To Impact

Is your customer experience program suffering bright shiny object syndrome?

You may have heard of ‘bright shiny object syndrome’. The term is used to describe new initiatives undertaken by organisations that either lack a strategic approach, or suffer from a failure to effectively implement.

Leveraging technology to stand out in the sea of sameness

The technology I'm talking about here is data and marketing automation. Current digital marketing methodology, much as it is practiced at Bluewolf, dictates the need for a strategy that does four things: Finds the right audience, uses the right channel, delivers the right content, and does all of that at the right time.

Eric Berridge

CEO and co-founder of Bluewolf, an IBM Company

Lead Management is very important part of the process. For anyone running Facebook Lead Ads I would recommend using this service.Get your...

Dirk Lo

How this fintech startup is improving content marketing and lead generation

Read more

I am agreeing with Mr. Tyron Hayes that a measured test-and-learn approach could be missing opportunities to not only better engage custo...

brunson5862@mail.ru

CMO interview: How Curtin University’s marketing chief is using test and learn to cope with complexity

Read more

Excellent!

Dr Sadasivan,US

Shakespeare shows data and creativity aren’t Montagues and Capulets

Read more

Great article! Agreed with all... Matthew Lerner, Deeps De Silva... When a company has a great product that solves customers needs, a gre...

James Tyler

Why marketers are embracing growth hacking techniques

Read more

Very good article, Social media analytics helps in problem identification. They can serve as an early warning system for negative custome...

BizVinu

Four ways to use social media to boost customer loyalty

Read more

Latest Podcast

More podcasts

Sign in