WalmartLabs Can Figure Out Gifts Your Facebook Friends Want – Forbes.
Tom Groenfeldt, Contributor
Stuck trying to find a gift for your mother-in-law, or that 16-year old niece? Put @WalmartLabs’ Social Genome to work.
Go to facebook.com/shopycat and download a small application to your computer. It will review your Facebook friends and look at their hobbies, interests and activities and create a list of gift suggestion. You can search for recommendations by name or by interest, review the suggestion and click to buy a gift. (Two drawbacks for the Facebook skeptics — it also puts your name, hobbies and gift wishes out there for friends to see, and you get enrolled in Facebook’s Timeline.)
On his blog, Anand Rajaraman, senior vice president at Walmart Global eCommerce, explains:
“For example, Shopycat notices that my friend Joe keeps posting about the Red Sox, and infers that he is a Red Sox– and therefore, a baseball — fan. Shopycat analyzes likes and shares to infer tastes as varied as Harry Potter, running, Angry Birds, sushi, yoga, and parenting to recommend gifts.
“The second step is to search across a large universe of products to find the one “wow” gift that doesn’t break the bank. Shopycat matches users’ interests to a giant catalog that includes products from Walmart.com, Walmart and sites including Barnes and Noble, RedEnvelope, ThinkGeek, and Hot Topic.” Rajaraman said Walmart understood it didn’t necessarily have the best selection of potential gifts, so it partnered with other retailers.
Before tools like Hadoop were available to work with big data, using information from social media would have been difficult, if not impossible. Now Facebook, Twitter and other sites including Flickr, have changed the sources of information available to retailers. No longer is a retailer limited to monitoring the actions of shoppers on its site or in its stores. Walmart can watch social media for trends, such as the rise in popularity of the English singer Susan Boyle as it was happening, so buyers can make sure they have the right music in stock for those who still buy CDs.
Rajaraman, a serial entrepreneur and venture capitalist investor in Silicon Valley, has developed technology to understand the sentiment in social media postings at Kosmix, a company he co-founded with Venky Harinarayan. It developed a platform called the Social Genome to organize and understand the deluge of data in status updates, tweets, blogs and videos. Walmart bought the company nine months ago and its 60 employees made the move to WalmartLabs where it is part of Walmart’s e-commerce operations. Rajaraman, who sold his comparison shopping company Junglee to Amazon in 1996 for $250 million, now talks of using the social media technology developed at Kosmix to help Walmart leapfrog Amazon.
In another example from Rajaraman’s blog: “…when I tweet “Loved Angelina Jolie in Salt,” the tweet connects me (a user) to Angelia Jolie (an actress) and SALT (a movie). By analyzing the huge volume of data produced every day on social media, the Social Genome builds rich profiles of users, topics, products, places, and events.”
So if his friend remarks on Facebook that she loves Salt, Shopycat will understand this is a prompt to suggest a DVD, rather than a a salt grinder, for a birthday present. All this data stored at Walmart sounds ominous, but the company is not building massive profiles on individual users from Shopycat or planning to target them with annoying offers.
“We don’t use it for any other marketing purposes,” Rajaraman said.
Emmett Cox, a former retail analysts for Walmart who is now at BBVA Compass Bank, said that Walmart does not store personal information like some other stores do. Target, for example, was the subject of a fascinating New York Times piece on its ability to tell when women were pregnant and begin targeting them with infant-related marketing materials.
Walmart is using only information that is publicly available on Facebook, Twitter and other social media sites.
Indeed, retail trade magazines have criticized Walmart for not using loyalty cards to develop more personalize customer information; the only card it offers is a credit card which provides modest a 1 percent rebate on its Discover card and 5 cents a gallon discount on gas with both its Discover and its plain Walmart card.
Because it has built its business around Every Day Low Pricing, a term which employees rattle off as EDLP, it doesn’t want to offer some customers lower prices through loyalty cards and gift programs. Although it may sound like a discounting cliché, EDLP is key to Walmart’s success and pretty much impossible for other stores to copy, said Cox, because other discounters have trained their customers to look for weekly sales flyers and other promotions. Cox said Kmart tried hard to persuade shoppers that it was offering everyday discounts but gave up and returned to regular sales promotions.
Rajaraman sees the use of social media as a natural extension of retailing. At a very high level, retailing is simply connecting products and consumers, and Walmart seeks to do it better. WalmartLabs is working both sides of the consumer-product relationship, he added.
In addition to monitoring social media to understand individuals, @WalmartLabs is looking for product information. It spotted a growing interest in cake pops, for example — balls of cake on sticks, sort of a combination between cake and lolly pops. By analyzing traffic on Twitter, @WalmartLabs could tell the buyers in Walmart’s Bentonvillle, Arkansas headquarters that cake pops were trending up, so merchandisers could look for suppliers of the baking kits.
Walmart opens voting today on another initiative that came out of the lab —Get On The Shelf. People who have developed a new product could make a video of it for Walmart where others could vote on which products they liked best.
“Get On the Shelf” offers anyone with a product the chance to make it to the major leagues and have their product carried by Walmart,” said the company, which pledged to help the winner find a suitable manufacturer if she or he needed production assistance. The idea for Get On the Shelf came from an engineer in the lab.
All this consumes a lot of data, and Rajaraman is an expert in big data. However he won’t say how much data Walmart has, and he contends the label is somewhat misleading these days.
“It is very hard to quantify the size of the data. It used to be possible when the data you had was your own, but now some is Twitter, some is Flickr and some is Facebook.”
He identified three important trends around big data in retailing. The sources of data are not limited to the enterprise any longer, and that is true not just for Walmart but for the world at large. Then the volume of data is huge and the data comes in many varieties.
Working with it is a challenge because many of the analytical tools and techniques for databases were developed over the last 20 years to work with much smaller data files. Faced with big data, many users try to fit the data to the tools and techniques they have. Rajaram has watched his students at Stanford address big data by sampling it, which risks losing important outliers.
“We have to develop techniques for the data size we have now rather than scale down the data.”
In addition to using Hadoop on stored data, WalmartLabs has developed Muppet for analyzing what Rajaraman refers to as fast data over a large cluster of computers.
“On top of Muppet, we employ a broad range of semantic analysis techniques, including information extraction and integration, natural language processing, and machine learning. While well known, these techniques have had to be significantly adapted, or extended, to deal with the peculiarities of social media. Finally, we have also developed techniques to effectively use crowd-sourcing and human computation in building and maintaining the Social Genome.”
The results of big data analytics often go to a marketing manager or a store manager who can act on the insights to make something happen.
“Data is the big frontier ahead of us,” Rajaraman added. “Customers live in a world with more and more data about products and more and more data about customers. The best retailer connects them in the most efficient manner. At the end, retailing will be a data game.”