Jeremy King was ignoring the largest retailer in the world. For a month, he'd been getting calls from a Walmart recruiter. King was used to being wooed, since he was well known in Silicon Valley as an engineer who built key parts of eBay's infrastructure. The calls kept coming. Finally, he picked up the phone and let Walmart know exactly what it would take to get him to interview. "I was like, 'Why don't you get the CEO on the phone—let him talk to me and then maybe I'll come in?'" recalls King, who didn't even know who the CEO of Walmart was. "I was being cocky. The CEO of the world's largest retailer wasn't going to meet with me just so I'd do an interview."
The next thing King knew, Walmart arranged for him to join a videoconference with CEO Mike Duke. "It was the strangest thing," King says. "Mike's office in Bentonville is the original one that Sam Walton had, complete with 1970s wood paneling. I was looking at this video, thinking, Where is this place?"
Over the next 45 minutes, though, Duke made what King calls an irresistible pitch. After years of seeing his company lag online, Duke swore that digital was now a priority for Walmart. Duke had restructured the company, placing e-commerce on equal footing with Walmart's other, much larger divisions. He had made serious investments in high-tech talent, acquiring several startups. One, a 65-person social media firm called Kosmix with expertise in search and analytics, was the impetus for Walmart rechristening its Valley operations "@WalmartLabs." Duke was looking for people who would revive the company's sites and services, and energize its entire culture. He hoped to turn a company famous for rigid, coldly effective business processes into one that's flexible, experimental, and entrepreneurial. In other words, Duke wanted to inject a bit of Silicon Valley into Bentonville, Arkansas. In the summer of 2011, King signed up as CTO of Walmart.com. "We've hired hundreds of incredibly talented people, in Silicon Valley and around the world," says Duke of his aggressive moves. "We are playing to win."
WalmartLabs is now housed in a boxy office tower in San Bruno, California, a few miles south of San Francisco. In just over a year, it has helped Walmart.com revamp its search engine; presciently identified the potential of the now red-hot "social gifting" market, where companies use social media cues to suggest presents; and this fall launched a test that offers same-day shipping to customers.
This last move is a clear signal of Walmart's serious intent to compete in digital e-commerce—and blunt the looming threat of Amazon, which has its own same-day shipping experiments. Having marginalized Barnes & Noble and Best Buy, Jeff Bezos has his eyes on a bigger target. Amazon has been moving aggressively to sell Walmart staples such as diapers, soap, pet food, and cereal, even letting customers subscribe for items they want to receive regularly. Walmart is the world's biggest grocer, and a central part of its strategy is that the millions of folks who visit its stores weekly to buy food will purchase a lot of other stuff. That's a key reason Walmart's 2011 revenue of $419 billion dwarfed Amazon's 2011 sales of $48 billion.
In e-commerce, however, Walmart is a distant challenger. The company has never broken out its Internet revenue, though in 2011, the analyst Internet Retailer estimated it to be $4.9 billion. In October, Walmart projected that global e-commerce would be $9 billion in the year ahead. Meanwhile, Amazon has been on a tear, with sales rocketing toward $100 billion annually in 2015. Analysts I spoke to believe Amazon has eaten into Walmart's sales of books, music, DVDs, electronics, and even toys. "When people started to say that Amazon was going to be the Walmart of e-commerce," notes Scot Wingo, CEO of ChannelAdvisor, an e-commerce technology and consulting firm, "that's when we started to see more signs of life from Walmart."
It would be a radical oversimplification to chalk up Walmart's digital revival solely to a hungry competitor (and Walmart execs often insist, perhaps a bit too strenuously, that they are not fixated on Amazon). No, Walmart needs to get digital because that's where its customers are headed. Soon everyone's phone will be smart enough for easy shopping. With Internet-enabled tablets selling for well under $200, lower-income families are already turning into online customers. "The way our customers shop in an increasingly interconnected world is changing," Duke says.
"I'm not going to be Chicken Little and tell you the company is going to go away if we don't get the Internet and mobile right," adds Neil Ashe, the company's top-ranking e-commerce executive. "We have an obligation to the mission to get this thing right because the customer expects it of us." Like the best Internet companies, Walmart obsesses about its customers more than its competition.
In 2012, Walmart celebrated its 50th birthday. In its first 25 years, Walmart became the world's biggest general merchandise retailer. But Sam Walton wanted to be a grocer as well. "A lot of people said that was crazy," says Joel Anderson, the CEO of Walmart.com U.S. who joined the company in 2007. "Twenty-five years ago, we couldn't even spell grocery. People thought we'd never figure it out."
Anderson says the next 25 years are about becoming a digital company. "In the first few years, were we tinkering and experimenting and not moving? There's some truth to that. But look at our history. When Walmart leans into something, it's like a tidal wave."
In April 2011, Walmart bought Kosmix for a reported $300 million. Kosmix's expertise lay in simplifying the sprawl of the web for users; its algorithms were novel because they tried to understand what a user wanted rather than just match her query text. For example, if a user searched for "presidential election," Google would return pages that contain variations on that term. Kosmix could find pages that were part of that topic even if the pages didn't contain the specific phrase. It then sorted them into categories such as candidate biographies, news stories, and polling data. Only in the past year have Google and Microsoft's Bing added Kosmix-like topic pages to their search results, but Kosmix's founders, Venky Harinarayan and Anand Rajaraman, hit upon the idea way back in 2004.
Walmart wanted to apply Kosmix's artificial intelligence to commerce, but it also wanted the real brains behind the tech. Harinarayan, 45, and Rajaraman, 40, both born and raised in India and graduates of the prestigious IIT Madras university, have been inseparable friends since they met as PhD students at Stanford in the 1990s. Both skinny and short, they have more than a passing physical resemblance, and they operate as a united pair. In the Valley, which loves its duos (such as Bill and Dave, Jobs and Woz, Ev and Biz), they're known as Venky and Anand.
The pair first tackled the problem of organizing the web in 1996, when they founded Junglee.com. In 1998, Jeff Bezos acquired the company for $250 million in stock. He realized that Junglee's technology would help its customers compare prices with other online stores, a bold move toward transparency that turned out to further solidify the company's hold on those customers. Rajaraman became Amazon's director of technology, while Harinarayan was charged with creating Amazon's Marketplace, where any merchant could sell its wares on Amazon's site. "I'd meet with Jeff for a couple hours every week," says Harinarayan. "What we came up with in 2000 was pretty much what Amazon has executed on since then." Analysts estimate 40% of the goods sold on Amazon are via Marketplace.
The duo left Amazon in 2000, and after a four-year stint as venture capitalists, founded Kosmix. The company had a unique ability to find meaningful information in the cacophony of the web. An application it built, called Tweetbeat, became one of the hottest ways to explore Twitter during the 2010 soccer World Cup, because it made it easy to discover who was talking about your favorite team or even individual players. That's part of what lured Walmart execs—they salivated at the idea of bringing this kind of intelligence to shopping. People are always offering clues about products on social media—writing reviews, liking brands, checking into stores, announcing the products they want to buy.
And what lured Harinarayan and Rajaraman, besides the money? They saw an opportunity to do something more interesting than merely replicate their work at Amazon for its rival. (In a delicious irony, Bezos profited from the Walmart deal—he was an early investor in Kosmix.) "What has changed since Amazon became big?" Rajaraman asks rhetorically. "You can connect the social experience, the in-store experience, and the online experience. Nobody could do that."
When the Kosmix team landed at Walmart in the summer of 2011, they found a mess. "The only thing Silicon Valley about Walmart was that we had an office in Silicon Valley," says Gibu Thomas, the senior VP who heads up Walmart's mobile tech team and was one of the first executives to approach Kosmix about a deal. "It was run like a traditional IT organization," he says, explaining that Walmart used outsourced, off-the-shelf systems to power key parts of its site. Worse, Walmart's 27 worldwide subsidiaries used incompatible technologies; the sites did not connect seamlessly with the stores or with Walmart's legendary supply chain.
Walmart.com's search engine epitomized its failure. "Executives at Walmart—at the board level—were running searches and saying, 'This is embarrassing,'" says Sri Subramaniam, the WalmartLabs exec who ran the rebuilding effort. If you used Walmart.com's old search engine to check out "smartphones," you'd get links to a couple of cell-phone chargers, not the iPhone. A "cotton socks" query returned results for cotton candy and balls of yarn.
The Kosmix team, so deeply ensconced in the ways of Silicon Valley, worried that cultural differences would hamper their efforts to turn the site around. "My first reaction was, Wow, this is going to be interesting," says Chris Bolte, who works on Walmart's search marketing systems. But those fears proved unfounded. More than a half-dozen people on both sides of the acquisition say that Kosmix's integration into Walmart was amazingly smooth. "I think part of it was that Walmart knew that they needed us, that this was a turnaround situation," Subramaniam says.
Their first job was to create a new search engine. It took just 10 months, with just a dozen or so engineers. Walmart will not discuss specific sales figures, but execs report that the improved search tools have increased the number of people who are converted from visitors to buyers on Walmart.com by as much as 15%. If you search for cotton socks now, you'll actually find them.
When Harinarayan and Rajaraman transformed Kosmix into WalmartLabs, they put roughly half the staff on such boring but crucial tasks. They deployed the rest as true lab workers, with the freedom to experiment in small teams on far-flung new ideas. "We organize these teams as mini startups with six to eight people," says Harinarayan, who learned from Bezos's organizational innovation of so-called two-pizza teams. "One person acts as CEO, and they have a clear business goal. We step out of the way and let these guys run it."
One of the first projects born from this approach was Shopycat, a gift-recommendation app that Walmart.com launched on Facebook before the 2011 holidays. Shopycat scans your friends' profiles to identify interesting gift ideas from their stream of likes, comments, and status updates, discerning if the "Ted" your pal is raving about is the geeky ideas festival or the Seth MacFarlane stoner comedy. Shopycat then seeks out an appropriate gift for such a stoner/thinker from Walmart's product database. Walmart says Shopycat led to an increase in purchases on the site, though it won't say by how much. For the 2012 holidays, the team built Shopycat into a section of Walmart.com called Walmart Gifts; customers will log in with their Facebook or Twitter account to get personalized recommendations.
Another clever retail application of WalmartLabs's core technology has been to use spikes in social network chatter to predict demand for out-of-the-ordinary products. Last year, the team correctly anticipated heightened customer interest in cake-pop makers based on social media conversations on Facebook and Twitter. A few months later, it noticed growing interest in electric juicers, tied in part to the popularity of the juice-crazy documentary Fat, Sick and Nearly Dead. The team sends the data to Walmart's buyers, who right now are only using it to confirm its other research. But as these signals become stronger, execs say it will play a larger role in purchasing decisions.
WalmartLabs has also created projects that just get customers to think differently about Walmart and e-commerce, including Get on the Shelf, an online contest for people to submit their own inventions to go on sale at Walmart. Get on the Shelf was a social marketing blockbuster, garnering more than 4,000 submissions, over 1 million votes, and news hits in small towns across America. Then there's Goodies, a subscription service in which Walmart customers pay seven dollars a month for home delivery of a gourmet food box—creating a discerning test market for the grocer in the process.
By themselves, none of these projects will single-handedly boost Walmart's e-commerce business. Taken together, though, they showcase a new dynamism at the retailing giant. "We're going to find ways to live at the edge," says Walmart e-commerce exec Ashe. "Every three or six months, you'll see something come out from us that will make you say 'Wow.' "
The next step, says Harinarayan, is about "scaling up Labs." But he and Rajaraman won't be part of it: In June, Harinarayan and Rajaraman announced that they were leaving WalmartLabs. To many outsiders, the abruptness of the founders' departure seemed troubling. It had been only a year since the acquisition, and they hadn't completed the "earn-out" phase, meaning they wouldn't receive their full share from the sale. Was their departure a sign that dynamic tech entrepreneurs felt smothered by Walmart's corporate culture? Or was it that Walmart could no longer tolerate leaving these hands-off leaders in charge?
Harinarayan and Rajaraman dismiss the speculation. They say they had spent eight years on the startup, and they were simply ready for time off. One Friday not long after the announcement, I meet Harinarayan at a coffee shop across the street from Kosmix's first office in downtown Mountain View. He's the picture of a relaxed man. After chatting about Walmart and Amazon for an hour, he told me he was free to keep talking, as he didn't have anything else to do that day. And he is quite sanguine about walking away from the money. "Given the fortune that Anand and I have had in our careers, if you're doing anything just for money, at this point it's going to be the wrong thing to do."
Jeremy King took over as the head of WalmartLabs, and to get a sense of where he will bring the skunk works, I visit the gleaming new Walmart store off the Almaden Expressway in San Jose where Jonathan Sherman, a WalmartLabs product manager, gives me a peek into the digital dimension being woven into this temple of American retail.
That future begins, like everything else, with a smartphone app. Walmart imagines that as you go through an average day, you'll remember things you need—milk, bread, a new tennis racquet, a toy truck for your nephew's birthday—and tell the voice-enabled Walmart app. The app will list each item's location inside your local Walmart and include product info; eventually, it will also learn your preferences and offer recommendations. And once you're actually in the store, you'll be able to summon an associate to help you.
Walmart's current iPhone app has only a few of these features: The voice-list system works very well, and, depending on which store you're in and what you're looking for, the app can sometimes locate your product.
At the moment, though, it won't show you extensive product info for all items, and it won't summon store help. The company has begun to test mobile checkout in select stores. As part of it, Walmart presents customers with a running tally of their total bill as they shop, the first explicit nod in my journey through WalmartLabs to the fact that millions of Walmart shoppers are on tight budgets.
This effort to reinvent the in-store shopping experience is an argument that Walmart's physical stores are a great asset, not a liability. "We are uniquely positioned to give customers anytime, anywhere access to Walmart by combining the smartphone, online, and the physical stores," Duke says. "Ultimately, that will give us an edge over any competitor." When I ask Walmart executives about Amazon's moves to offer more customers next-day and same-day shipping, many were amused. "It's fun to see them trying to be us," says Walmart.com CEO Anderson. "We have more than 4,000 forward-deployed fulfillment centers and we're already doing shipments from some of them. Some people call them stores."
"If you think about the last 20 years of retail, how people shop in a store has not changed," Thomas says. "The question we're asking is, how do you bring to a store the capabilities that have made e-commerce successful? With 200 million customers a week, if you can increase the average basket size by a dollar—that's billions of dollars every year." In fact, it's more than $10 billion—more than its projected annual e-commerce revenue this year.
If Walmart fails in its digital transformation, it won't be for lack of resources or possibilities. Ninety-six percent of Americans live within 20 miles of a Walmart. No one has as much money; no one has a better supply chain; no one has such a close connection with so many customers. Walmart execs know this. "We'll spend more on capital expenditures this year than Amazon has spent in its entire history," Ashe tells me (hyperbolically).
That size, however, is also Walmart's greatest enemy. WalmartLabs's two-pizza teams can come up with a thousand innovative ways to improve shopping, online and off, but none matter if the company's execution is slow and bureaucratic. And the fact is that implementing these ideas will always be complex. Every change to how items are delivered, or how customers navigate stores, or how applications work with the company's existing IT structure is a maneuver that requires the coordination of thousands of moving parts.
But Walmart can succeed online without becoming the Amazon of the web. The phrase I hear most often from Walmart people is that the only way the company will win online is "by being Walmart." And they're right. Walmart doesn't need to be something radically different. The company that mastered IT in the service of unbeatable prices must now master web technology. It doesn't need to chase Amazon so much as it needs to identify how a digital Walmart can be as much a part of its customers' lives as the stores are today.
And it has to think long term. It may take a decade or more for Walmart to be a successful digital retailer. "Somebody at one of the board meetings asked me, 'Neil, how long is this going to take, and how much is it going to cost?'" Ashe recalls. "And I said, 'It's going to take the rest of our careers, and it's going to cost whatever it costs. Because this isn't a project, this is the company.'"
A version of this article appeared in the December 2012/January 2013 issue of Fast Company magazine.