IN-DEPTH: It is considered that there is a uniformity of general search patterns that has long irked those in the industry who want to make search more flexible, intuitive and personalised. For a start-up, Hopper believes that to fix the user experience, first there is a need to fix the data. Hopper is built on NoSQL technology, which means it doesn’t have a rigid data schema, and it can scale out

Published: 25 Aug 2011

IN-DEPTH: It is considered that there is a uniformity of general search patterns that has long irked those in the industry who want to make search more flexible, intuitive and personalised. For a start-up, Hopper believes that to fix the user experience, first there is a need to fix the data. Hopper is built on NoSQL technology, which means it doesn’t have a rigid data schema, and it can scale out easily to store hundreds of millions, even billions of records.

By Ritesh Gupta

Hopper, a startup founded by a team of former Expedia employees, has this week raised $8M in new funding from Atlas Venture and Brightspark Ventures. The company is developing a website for consumer travel discovery.

Hopper’s travel-oriented search engine allows users to discover destinations and products using only keywords. Hopper is applying cluster computing techniques to build the world’s largest database of travel information. The company is using Machine Learning, NoSQL databases and Big Data processing to transform raw web pages into structured and organised information, enabling a faster, more complete, and more flexible search than traditional travel sites.

Hopper will open a new office in Boston, where it expects to double its engineering team over the next year. It was founded in 2007 and previously raised $2M in 2008, bringing the company’s total funding to $10M.

“We are getting our key partnerships in place and we are really eager to get those partner data feeds fully scaled up and running so that we can launch the site as soon as possible. Until then, we are heads down on extending and tuning the product,” says Hopper CEO Frederic Lalonde.

Lalonde spoke to EyeforTravel’s Ritesh Gupta about Hopper’s unique approach, its technology, and relevant issues. Excerpts:

Which gaps/ opportunities are you trying to address through Hopper?

The web revolutionised travel when it allowed consumers to book their own trips. But the tools stopped evolving over a decade ago. Planning a trip online is actually very cumbersome and time-consuming: a lot of jumping around from site to site, a lot of forms to fill out, a lot of waiting for results, and a lot of useless information.

Numerous startups have attempted to tackle this problem. But Hopper is taking a unique approach: we are using a new generation of Big Data analysis technologies. We believe that to fix the user experience, we first need to fix the data. And there is a lot of data out there.

How do you think Hopper is going to set a benchmark in the travel industry?

One of the reasons why online trip planning is so broken actually has to do with structure of the industry and the technology it uses. The whole travel industry operates on request/response data systems that only give answers to very specific questions.

In a way, the online travel industry today looks like the mapping industry did 10 years ago. There were a handful of data providers like Navteq and Tele Atlas who invested millions of dollars to collect geographic data. As a result, it cost a fortune to license that data. So as a consumer, if you wanted digital maps you had to buy a Garmin device and buy all sorts of CD-ROMs.

And then came Google Maps. They licensed all this proprietary data and they made it public through an interface built for the web from the ground up: Image tiles and JavaScript. But the real magic happened when Google combined this data with all the other data that it had already collected from the Web. And most importantly, they made it free. For users, it was revolutionary. But it also created all kinds of new opportunities for the data providers. There was a lot of resistance to making the data public, because it was their core asset. But it ended up reinvigorating the industry. There’s never been so much demand for map content.

There has been a sea change of technology in the last few years. What a startup can now do with analysis and aggregation of really big data sets would have been inconceivable a decade ago. And this has led to some incredible innovation in domains like web search, social media monitoring, finance, and health. But the travel industry is lagging behind. We want to change that.

So we are reaching out to suppliers, distribution systems and online travel agents to get a trend of open data started. When we started we had no idea how this would be received, but we are finding that almost everyone is listening, and many are eager to get to work.

Is that because they’re afraid of what Google is working on with ITA? Maybe. Either way, there is a recognition that online travel needs an upgrade, and that it has to start with the core data technologies.

How is your offering going to be beneficial for travellers in their travel planning and buying cycle?

Much of the planning is actually happening on the search engines, who lead consumers to magazine articles, blogs, forums and reviews, whereas the booking is happening on travel merchant sites. And those sites have tried to integrate planning and inspiration content, but it’s too shallow and often weakly integrated with the products.

All of this is frustrating for consumers, because in reality planning and shopping are not sequential activities, they are part of a continuous thought process of discovery. We believe that the current tools widely ignore this. So we’re building Hopper in a completely different way.

Hopper is built on NoSQL technology, which means we don’t have a rigid data schema, and we can scale out easily to store hundreds of millions, even billions of records. In order to run this, we need to build our own servers. And we use Map/Reduce, which is a really powerful framework for distributed processing. Our partners push product data to Hopper; it’s like a live feed of up-to-the-minute flight, hotel and package offers that users on their sites have queried. And so we break from the request/response paradigm. We don’t compute optimal fares and routes; it’s not our expertise. We don’t package products, and we don’t apply discounts. We are the opposite of a meta search site.

All of our focus goes to computing the relevant relationships between web content, travel products and the real-world in order to make all these simultaneously and instantly search-able by keyword. Any keyword.

A couple of years ago, social shopping for travel was being touted as the next big thing. It was being said that after all, unbiased meta-search and UGC seem to be perfectly compatible life partners. What’s your viewpoint regarding the same?

I think there is a fundamental social liquidity problem in social search. How many of your friends and family have been to this small town in Sicily and will provide you with useful information? The overlap between your social graph and the travel options in your consideration set is actually pretty thin.

In any case, we are convinced that you can’t create a social travel experience without first building a complete database of travel content. Other industries have evolved social marketplaces, but that’s because they have complete and widely available data sets. Music, books, and film, for instance, all have complete catalogues available for purchase or even download. In travel, that data is closely guarded and fragmented. And without a data set of sufficient depth and structure, you can’t really capture meaningful social signals.