Hadoop Demonstrates Open Source Power with Overstock App
December 31, 2012
By Susan J. Campbell
, TMCnet Contributing Editor
When running a business, it’s key to ensure you have the latest innovations that help you get your product to market, meet the needs of your customer base and still turn a profit. Such innovations vary according to the company and the industry, but the needs exist all the same. But, what if you could get exactly what you need and it didn’t cost you any money – what would that mean for your bottom line?
Overstock.com was able to find out the answer to such a question. According to this Wired report, a Web app built by Judd Bagley to provide a continuous stream of news stories tailored to a user’s particular tastes has led Overstock to a system that would generate product recommendations for its online shoppers – for free. This price tag (News - Alert) is a significant reduction from the $2 million annual rate to which it was accustomed.
The resulting app, MyCurrent, was developed in the O Labs, Overstock’s research and development division. This miniature initiative was founded as part of a strategy to incubate projects that could push the company in new and more profitable directions. With the growing amount of data companies gather on a regular basis to try and analyze the market, customers and opportunities, the demand for Hadoop to distribute clusters of data across multiple servers is growing. Apparently, so is the need for a personal news reader.
A successful online retailer seems like a more logical choice for Hadoop implementation rather than a personal news reader, but this deviation from the expected may be the exact reason why it has the potential to drive new success. Not only will this new app save the company money, it has also proven to provide new excitement within the company’s retail division. And, it has more to do with Hadoop than you may realize.
As it turns out, the MyCurrent project created an opportunity for Bagley and his team to come across Mahout, a project that provides a set of machine learning algorithms on an open source platform. Mahout was used to examine the news stories enjoyed by individuals in the past and then to select others stories the individual was likely to care about and enjoy. The system worked so well that O Labs replaced its system with an engine built on the Mahout foundation, along with Hadoop.
Mahout was built specifically to use with Hadoop and the success Overstock and its O Labs have had with the platform demonstrate its value across the Internet and even into online business. In fact, the solution has made such an impact that it has captured the attention of not just Overstock, but also Foursquare (News - Alert), AOL, Twitter, Yahoo and even online retail king Amazon.
As online retailers and other eCommerce sites continue to compete for leadership in saturated markets, the attention of the customer base will be harder to acquire. Personalized service is critical, as demonstrated through Amazon’s ability to cater to individual customer preferences. Hadoop and Mahout are raising the bar, allowing for the creation of personalized services through a deep understanding of consumer use and preferences, without invading privacy.
As Overstock leads the way, others are sure to follow, allowing customers to enjoy greater satisfaction, which retailers hope will lead to greater revenue.
Want to learn more about the latest in communications and technology? Then be sure to attend ITEXPO Miami 2013, Jan 29- Feb. 1 in Miami, Florida. Stay in touch with everything happening at ITEXPO (News - Alert). Follow us on Twitter.
Edited by Rachel Ramsey