Walmart Unleashes Facebook App to Find Gifts People Want
Searching online wish lists, guessing or just directly asking people what they want as gifts is so 20th century, so Walmart is launching a new approach for the holidays -- its Shopycat app that scours Facebook profiles, updates, comments and likes to come up with gifts for your friends.
The giant retailer's @Walmartlabs unit has been testing the app since August, gaining about 6,000 users to date, but will launch it more widely this week among Walmart's 10.6 million Facebook fans and via the retailer's YouTube channel. The goal is to get users to buy gifts at Walmart.com, but Shopycat also shows availability at a user's nearest Walmart store and in some cases even directs them to the websites of rival retailers such as Bed Bath & Beyond or Hot Topic for items Walmart doesn't carry.
"Gifting is fundamentally a social activity," said Venky Harinarayan, co-head of @Walmartlabs e-commerce and social commerce development unit based in Mountain View, Calif. "It's about people connecting with other people through products. It's a huge overall business opportunity. It's something that 's not done very well on the web today."
Facebook is a rich trove of data for gift recommendations, he said, both because people's friends and families tend to be there and because they're constantly dropping hints, whether they know it or not.
At least that 's the theory behind Shopycat. It gathers clues not so much from what people put in their profiles among their interests, said Shopycat Product Director Robert Yeu, as on their updates, comments, app usage and more.
Interests listed in profiles get less weight because people tend to fill in that information when they join Facebook and update it infrequently. As @WalmartLabs has tested Shopycat, it's found that users' most recent activity plays a big role in making satisfactory gift suggestions, Mr. Harinarayan said.
Even so, it can be tough to find great gifts just based on what people say or do on Facebook, so @WalmartLabs has been tweaking Shopycat's algorithm. After the early tests, the Shopycat algorithm has been adapted to recognize that certain products, such as toilet paper -- which Shopycat actually suggested for an advertising agency executive -- don't make good gifts, Mr. Harinarayan said.
Even after several weeks of tweaking, some recipients weren't pleased. An industry association executive who Ad Age ran though Shopycat noted that he already had nearly every item on his list of suggested gifts, including an electric razor he'd already bought at Amazon. "For anyone wanting to get me something for the holidays," he wrote in an e-mail, "I've got a better shopping app: The telephone. Call me and I'll tell you what I want."
Mr. Harinarayan said @WalmartLabs has already learned that when someone mentions a book on Facebook, they've probably already read it, so they use that to inform recommendations of similar books or glean interest in products. Likewise Shopycat has moved from recommending music people mention in their profiles to books, special editions or merchandise related to the musicians.
Shopycat also has added a new feature -- the ability to look up its recommendations for your own gifts as they'll be seen by friends. Mr. Yeu also expects people to share Shopycat recommendations with friends of friends to get their input on whether it's appropriate for the recipient, and see if they want to chip in.
Early on, Shopycat seemed to recommend the same items on the gift lists of many people -- with the Presto Pizzazz Pizza Oven a common choice, akin to the electric tea kettle The Onion has lampooned as a popular choice of aunts and stepdads for kids they barely know. In recent weeks, as the algorithm and databases have grown, suggestions have grown more individualized.
But people who are less active on Facebook are more likely to get more generic suggestions. "More data is always better," Mr. Harinarayan said, but added that some items -- such as iTunes and Walmart gift cards -- are also in the mix to offer a range of price points.