23% of e-retail transactions on Thanksgiving and Black Friday came from mobile devices, according to payments security firm ThreatMetrix. However, 15.5% of retailers say ...
GraphDive takes a new spin through social recommendations
Its social media graph technology helps Zoostores unlock customers’ interests.
Chief Technology Editor
Topics: 88Hours.com, Correlation Ventures, Crosslink Capital, eBay Inc., Facebook, GraphDive, Jesus Pindado, Magento, Nikhil Behl, Shahram Seyedin-Noor, Sina Sohangir, SoundHound, Strands Recommender, Top 500, Zoostores
Zoostores specializes in selling in product niches like back-yard trampolines and $3,000 machines that fire tennis balls to practicing tennis players. One of the retailer’s biggest challenges, CEO Nikhil Behl says, is knowing when to show products from each of those niches—or from other niches like espresso machines and camping gear—to a web site visitor.
“If someone is looking for a trampoline, they don’t necessarily know we also sell tennis balls or espresso machines,” he says. “So we build composites of customers and personalize sites for them.”
Since this past summer, Zoostores, No. 236 in the Internet Retailer Top 500 Guide, has been using GraphDive to gather information from Facebook to flush out what it knows about site visitors. The aim is to improve how it customizes web pages for visitors to its e-commerce site, which is built on the open-source Magento e-commerce platform from eBay Inc. “If we learn they like tennis pro Nadal Rafael and the French Open tennis tournament, we infer that they probably like tennis and play tennis, and we’ll show them tennis ball machines,” Behl says.
And if other information such as the visitor’s ZIP code indicates she has a high income, Zoostores will show her its high-end tennis ball machine that retails for about $3,000.
GraphDive’s Social API, or application programming interface, provides a software toolkit for connecting an e-commerce site with social media sites, including blogs and forums as well as Facebook and other social networks. Zoostores for now is using it to only connect with Facebook, where it can pull data only from consumers who have agreed to share their personal information with Zoostores.
Behl says it’s too soon to say what effect the GraphDive system is having on Zoostores’ sales, which he says are on course to rise 100% year over year in the fourth quarter, but notes that it’s an important step in tying into social media to improve how the retailer engages customers. “With GraphDive, we get a mix of social data and demographic data on consumers that I’ve not see elsewhere on the same level,” says Behl, a veteran of e-commerce who helped to launch the consumer web site for Hewlett-Packard Co.
Social media, he adds, will continue to emerge as an important tool for understanding consumers and marketing effectively to them. “This will not happen overnight and will take a lot of experimentation,” Behl says.
Another way he plans to use GraphDive is on 88Hours.com, the retailer’s social flash-sale site that ties its product offers to Facebook activity. When participating consumers see a preliminary deal they like on 88Hours—a recent deal offered a 34-oz. flask for storing hot coffee or soup, marked at $10.99, a 67% discount from $32.99—they have 44 hours to Like the deal on Facebook and gather a minimum number of Likes for the same deal from friends. The minimum number of Likes varies with each deal; for the flask deal, the minimum was 150 Likes. If they reach the minimum within 44 hours, 88Hours makes the deal go live and gives the Facebook friends who have Liked the deal 44 hours to make a purchase.
One of the challenges in making 88Hours effective at driving sales, Behl says, is starting out with deals that shoppers want. “We’re hoping that with GraphDive we’ll learn more about that, and curate more relevant deals,” he says. Eventually, Behl says, he’ll also consider using GraphDive to use social media input in the personalization of e-mail and other forms of marketing.
Another company using GraphDive, products recommendations service Strands Recommender, has found that it helps to more quickly build recommended product content on its client retail sites. “A big challenge we have is that when a consumer first comes to a site, we don’t know much about them to recommend products,” says Jesus Pindado, vice president at Strands. By using GraphDive, Strands can quickly build a profile of a consumer based on her social connections, he says. “It’s a cold-start problem, but GraphDive solves it for most consumers,” Pindado says.
GraphDive charges a monthly subscription fee that varies based on the volume of information a client retailer accesses through the GraphDive Social API, plus a few cents for each API data transfer, the company says. It also charges an annual fee of a few dollars for connecting to Facebook.
Based in Menlo Park, CA, GraphDive was founded in 2011 by CEO Shahram Seyedin-Noor and chief technology officer Sina Sohangir. Prior to GraphDive, Seyedin-Noor was a founding executive of NextBio, an analytics company involved in medical research, and was an advisor to technology companies at Goldman Sachs & Co. and Bank of America. Sohangir is a former senior engineer at Netseer, an online advertising analytics company; he also worked on music and speech recognition apps for iPhones and Android mobile devices at Melodis, a company since renamed SoundHound.
GraphDive is backed by more than $1 million in venture capital from investors including Crosslink Capital and Correlation Ventures.