Viral Loop

by ADAM L PENENBERG

  FIG. 3. Growth rates for viral coefficients of .6, .9, and 1.2.

  With Ning, Andreessen estimates the viral coefficient is a whopping 2.0: each person who signs up is worth, on average, two people (compounded daily). The next day she brings in four members; on day 3, eight; within a week, 128 people, each worth double her weight in virtual clones. The curve is almost twice as steep as the one that represents a viral coefficient of 1.2, and that is how Ning has been able to double in size roughly every five months. “It’s the power of compounding, predictable growth rates,” Gina Bianchini, Ning’s cofounder and a former Goldman Sachs investment banker, says.

  When Ning veers from its projected growth pattern, it’s usually because of a performance issue. Perhaps pages aren’t downloading fast enough—even milliseconds can affect virality—or there is a change in the size or shape of user pages, all of which can put downward pressure on the viral coefficient. At one point Ning slipped and Bianchini’s engineers tracked it back to a new sign-up system, which required registrants to retype passwords (for security purposes). They changed it back and Ning’s viral coefficient popped back to 2.0. In January 2009, Ning shut down its red light district, finding that adult-themed networks didn’t generate enough advertising or premium services to cover their costs—plus its ad partners were leery of porn. These kinds of networks also resulted in a disproportionately high number of Digital Millenium Copyright Act notices. What’s more, the more legal adult networks there were on Ning, the more illegal ones posting kiddie porn popped up, which required contacting the FBI. While Ning initially lost 20 percent of its page views from blacklisting porn-themed social networks, traffic bounced back within five weeks, and Ning was back on track, growing faster than ever.

  [ ANATOMY OF A START-UP ]

  A thirty-five-year-old northern California native, Bianchini met Andreessen after receiving her MBA from Stanford and launching a software start-up that tracked and measured advertising. Andreessen sat on the board of the company, which went under in the dot-com crash; he and Bianchini dated for a spell before becoming friends. Andreessen, of course, had built Netscape and off-loaded it to AOL for $4.2 billion in 1999. Even before he sold his next company, Opsware, an automated network and server company, to Hewlett-Packard for $1.6 billion, he had begun casting about for his next billion-dollar act. One thing he knew, it wouldn’t be a Web business.

  Still smarting over the end of the bubble, when the NASDAQ lost four-fifths of its value from its peak in March 2000, Andreessen believed the Web was closed for business. But Bianchini, who was consulting for Michael Ovitz, saw things differently. As Warren Buffett put it, “Get greedy when others are fearful and fearful when others are greedy.” Bianchini had been hearing about a bevy of smart start-ups she dubbed “garage viral businesses” that were thriving while Rome burned. There was Craigslist, PayPal, Hot or Not, Crushlink—a million-dollar email dating scheme started by two Harvard physics students—and Birthday Alarm, which generated $3 million a year in revenue by automatically reminding users of important dates. And, of course, Google, which was en route to a $100 billion market cap, and a young start-up from Los Angeles called MySpace, which launched in 2003 and was showing great promise. Bianchini believed it was the perfect time to start a new viral venture.

  Slowly Andreessen began to see the light. He realized the entire sector was underfunded, since all anyone seemed to hear was bad news. Bianchini provided data that showed there was a five-year gap in Internet investment, and it dawned on him that whoever got to market early would have a tremendous advantage. Then twenty to twenty-five media companies would go shopping when they realized they were in danger of falling behind. When he thought about what was working on the Internet, he came up with eBay, PayPal, MySpace, and sites that enhanced social behaviors like blogging, photo sharing, and reviews. He wondered about combining all of them into a single platform that could grow virally. Over brunch with Bianchini, Andreessen suggested they focus on creating social networking platforms, and Ning (“peace” in Chinese), seeded initially with $1 million of his own money, was born.

  They make an exquisitely odd pair: he is the gangly six-foot, four-inch former coder with the huge egg-shaped head, famous for having posed barefoot on a throne for the cover of Time; she is the petite, feminine jock, a former field hockey standout at Stanford, a woman from working-class roots anointed a “Web 2.0 Hottie” by Valleywag (a fact she concedes only reluctantly). He’s the grand visionary, the company chairman whose speech unfurls in sheets of sound and who peppers his emails with smiley face emoticons, because he learned long ago that his droll sense of humor did not travel well over the asynchronous Web; she’s the effervescent CEO, a social networker at heart who prefers roaming Ning’s hallways to the isolation of working at home. He once won Customer of the Year at Hobee’s, a local restaurant chain, for eating there every day for a year; she waitressed at a Hobee’s in high school.

  In the early days, not everyone knew what to make of Ning. The company spent three years designing and building the site’s underlying platform; a year into that process, it released a couple dozen social applications to begin testing and refining what they had made. Those simple applications led Michael Arrington to post an entry on his TechCrunch blog entitled “Ning RIP?”: “The reality of Ning is that it has lost whatever coolness it had, no one uses it, and Ning is going to have a very hard time getting people’s attention.” But Arrington did an about-face eighteen months later, after Andreessen and Bianchini attracted one hundred thousand online groups in about six months. “Everyone wants a social network of their own, and Ning is here to give them one,” Arrington wrote. “The company sure has come a long way since I pronounced it dead in early 2006. Sometimes I like it when I’m wrong.”

  [ DOUBLE VIRAL LOOP ]

  Significantly, viral loop companies are a form of organizational technology. In the way that Google doesn’t own the Internet (it just seems that way) but simply helps you find what’s out there, Bebo, Facebook, Friendster, Flickr, MySpace, and YouTube don’t create content—their audience does. To Skype or not to Skype may be the question, but what if no one were on the other end to answer? Would you “tweet” if no one else were on Twitter? PayPal is a quasi bank for online transactions; it doesn’t mint money. These viral loop companies provide an environment that is, in theory, almost infinitely scalable, relying on the wisdom of crowds to create or aggregate masses of material to fill it. The more people, the more content; the more powerful the lure for those sitting on the sidelines, the more value the company has.

  There is also a multiplier effect, because the bigger the viral network, the faster it grows. (Double 10,000 users in a month and you get 20,000; double 50 million and you get 100 million.) Some of the biggest names on the Internet rode this viral wave to stratospheric heights until ultimately hitting saturation. EBay went from online garage sale to megasite because sellers attracted buyers who attracted more sellers and buyers. Google has pursued a similar strategy outside of search: under every set of ads it serves up sits a link to its AdSense program, which encourages more website owners to join (and in Google’s case, joining leads to cold hard cash, which amplifies the viral effect).

  The same mechanisms are at work with the spread of any network, whether it is the telephone, fax, cell phone, instant messaging, email, or Skype. Each additional user is worth more than any individual user. Eventually almost everyone joins such a network, the way that everyone has a telephone or an email address, because the value to being on it is so huge as a result of everyone else being on it. Nicholas Economides, a professor of economics at New York University’s Stern School of Business, characterizes it as “a network effect”: “The more connections you have, the more nodes, the more people, the more valuable it will be,” he says.

  What’s more, viral networks can be stacked. PayPal came into being because buyers and sellers on eBay needed a way to complete transactions online, since most sellers without traditional storefronts coul
dn’t process credit cards. YouTube took off by piggybacking on the success of MySpace, becoming the go-to site for video posters and watchers alike. Flickr grew in tandem with the entire blogosphere. Google’s Gadget Ads was spun off as a mini viral network to serve ads on the tens of thousands of widgets, which themselves were layered over MySpace, Facebook, Bebo, and the rest.

  This stackability will only increase as the walls between social networks and websites crumble, hastened by the creation of OpenSocial, which provides a common programming standard so that applications can run across multiple websites. Ning, Google, LinkedIn, Yahoo, and MySpace are members; the coalition has taken the Facebook platform concept and applied it to the entire Web, meaning that a widget that works on one site will work on all the others. Think of it as the cyber equivalent of introducing standard railroad gauge during the Industrial Revolution, which helped spur American economic development coast to coast.

  What separates Ning from other viral networks, however, is that it benefits from what Andreessen calls a “double viral loop.” It spreads two ways because every network creator is a user and every user is a potential network creator. Say someone sets up an Angelina Jolie network with ten members, which grows as each person draws in others. An adoption site breaks off, a Jon Voight hate group rises up, a Brad Pitt love club forms, a Lara Croft nostalgia net appears, spawning a legion of anime spinoffs. Soon you have two, three, five networks, all expanding simultaneously. In the meantime, the original group continues to attract users. Ning swells like a river fed by an ever-growing number of tributaries.

  [ TO THE PROMISE LOOP ]

  Once a viral loop takes hold you can accurately predict how fast your user base will grow. That’s because a viral loop expands according to something known as a power law curve, characterized by a soft-edged L shape (see Figs. 4–7). It has other names, too: the 80–20 rule, Pareto’s law, the law of the vital few, the principle of factor sparsity, the long tail. The concept originated in 1906, the inspiration of Italian economist Vilfredo Pareto, who realized that 20 percent of the population owned 80 percent of the property in Italy. He fashioned power law curves to posit that wealth follows a “predictable imbalance,” leading to a winner-take-all society.

  Although no one knows exactly why, power laws describe a dizzying array of natural and unnatural phenomena—everything from ranking the size of planets, galaxies, and lakes by volume to Newtonian physics. If you charted Earth’s cities by population, listed everyone in the world in descending order of wealth, analyzed sales of music or the concentration of endowments for public and private colleges, with 20 percent of the schools possessing 80 percent of the endowments, you would end up with a curve similar to Pareto’s curve for Italian land distribution. Pareto’s law has been called on to characterize blog traffic, with, roughly speaking, the top 20 percent of bloggers attracting 80 percent of the readers and determining the relative popularity of websites. While the general shape of the curve remains the same, Pareto’s law is not a hard and fast rule; the 80–20 rule is just an approximation. Exact percentages vary depending on the exact phenomenon under the microscope.

  Chris Anderson, in The Long Tail: Why the Future of Business Is Selling Less of More, relies on Pareto’s thesis to explore online commerce and illustrate a heavy tail argument, that 98 percent of all the possible choices get chosen by someone, and the 90 percent available online often account for half the revenue and two-thirds of the profits. The point of the “long tail” argument is that in certain Internet, digital, and frictionless domains, the dominance of the “head” of the curve (a minority of things generating a majority of activity or revenue) should be somewhat less than what you would normally see in the real world.

  Applied to music it is easy to see. In the real world, retail stores stock only a small percentage of the CDs that are available. Walmart, for example, may only have three hundred CDs on its shelves, so consumers don’t have the option of buying music further down the tail, which means sales are highly concentrated toward the top three hundred CDs. Then it’s possible it becomes a 1:100 rule: 1 percent of the CDs generate almost 100 percent of the revenue. On iTunes, on the other hand, every track and CD is available, digitally, frictionless, easily findable and searchable. In the digital domain, because the long tail is more accessible, the head is less dominant than in the real world. Instead of 1:100, maybe it’s 5:80—5 percent of the CDs account for 80 percent of the revenue, with the remaining 95 percent of releases accounting for 20 percent of the revenue—and 20 percent is a lot more than zero percent, especially in a large market. Google has been a huge enabler of this longer tail because a product near the tip is both findable and accessible.

  If Ning were represented by a traditional offline power law curve, 20 percent of its largest social networks would generate 80 percent of its page views, and a few huge networks would dominate usage and traffic. Instead, while Ning’s top two hundred networks are responsible for the lion’s share of traffic, they’re not responsible for all of it, and, in fact, not much more than half. Many networks down the long tail are also thriving. Ning is seeing a power law curve distribution, but the head is squeezed, while the tail extends outward. And it expects this heavy tail to hold sway in the future, with loads of midsize to small networks with niche audiences exhibiting the most growth despite being microtargeted and relevant to relatively small numbers of people. Users discover these networks via email invitations (users inviting other users), embedded widgets (users embedding widgets in blogs and other places, leading people back to the networks), and that old mainstay: Google searches, in addition to simple word of mouth or users browsing or searching on Ning itself.

  If Ning keeps growing at this rate, it will reach the promised loop as it tips to a point of nondisplacement, adding users even if it does nothing and becoming virtually impregnable. After PayPal blossomed as an online transaction power, eBay launched a competing service and failed miserably (eventually buying PayPal instead). To combat YouTube, Google and Yahoo launched rival online video sites, neither of which went anywhere. There’s the likelihood that LinkedIn has become unassailable by aggregating huge pools of users. Ning also seems destined to achieve this point of nondisplacement. In only one instance, Friendster, has a company fallen apart after achieving this kind of reach, and it was largely done in by technical failures—network meltdowns and outages—that drove users into the embrace of MySpace and Facebook. Twitter, too, has had trouble scaling; its network is buggy and unreliable. While it is possible it could follow Friendster to ignominy, it has an enviable connection with a deep, loyal pool of users who have weathered severe outages as its network grows.

  FIG. 4. Each white sphere represents a new Ning user, each line an invitation to join. The starbursts define the extent and growth pattern of a single network, centered on its creator. (Ning.)

  FIG. 5. In some nings, the creator invites most members. When subsequent users bring in new members, clusters form to reflect the viral chain. (Ning.)

  FIG. 6. Ning benefits from a double viral loop, since a member of one social network will often set up another social network on a different topic. (Ning.)

  FIG. 7. This viral effect means each Ning member is equal to two users, compounded daily. That’s how Ning grows and expects to amass millions of networks with tens of millions of users and billions of page views. (Ning.)

  Mass audiences on this scale carry serious potential, hence the fantastical valuations viral loop companies have achieved. That’s because if you can get that many people to use your product, someone somewhere will pay you to reach them. Even if your strategy is limited to getting big and bought, you have the opportunity to auction off yourself (and your millions of users) to the highest bidder and let the next guy worry about wringing revenue from your audience. MySpace, which News Corp. bought for $580 million and Murdoch estimates is worth close to $6 billion today, chose this route. So did YouTube (Google, $1.65 billion), PayPal, and Skype (eBay, $1.5 billion and $2.6 billion,
respectively). Flickr cashed out early for a modest $40 million (to Yahoo). Or you can try to monetize those bodies yourself. If Facebook skimmed a buck a month off hundreds of millions of members, that would mean yearly revenues in the billions.

  That’s not exactly a radical idea. Google achieved a $100 billion market cap by vacuuming up nickels, dimes, and quarters from search ads and counting clicks on advertisements. Craigslist monetizes a scant 2 percent of its site, giving away the rest in the form of free classified ads. The belief is that massive audiences like these can generate huge returns in the same way a 0.25 percent increase in sales tax for a municipality can yield millions in revenue.

  [ SHARED CHARACTERISTICS OF VIRAL LOOP COMPANIES ]

  This doesn’t mean launching a successful viral loop company is easy. Far from it. Creating a deliciously spreadable product is merely the first step. Then comes the hard work of ramping up a business, and that’s where some real challenges await, as Hong and Young learned with Hot or Not. Viral loop companies succeed for many of the same reasons, while unsuccessful ones seem to fail in different ways. For every eBay and Facebook there are heaps of social start-up failures you’ve never heard of. In addition to Friendster, which simply couldn’t scale, there is Tribe, for example, an online community that never caught on; Plaxo, creator of a “smart address book” that didn’t serve an important enough need; and Flixter, which lets users share movie reviews but has had trouble attracting repeat traffic.