Of all the interviews I did for my book, Viral Loop, the one I enjoyed most was with Michael Birch. This was about six months after he sold Bebo, the social network he had largely bootstrapped with his wife, Xochi, for \$850 million to AOL. Together they held a 70% equity stake in Bebo and walked away with \$595 million – enough, I suppose, to buy a small one-bedroom apartment in the Bay Area. Then AOL did what it does best and after absorbing Bebo ran it into the ground. Now Birch has announced on Twitter that he’s buying back “his baby,” as he’s called it, for a paltry \$1 million eight years after founding it and five years after selling.

Yes, Birch was lucky, but he also made his own luck. While he’s often viewed as a ponytailed Brit who came out of nowhere to hit pay dirt, the man paid his dues. The first three startups he launched after toiling for six years in London for an insurance company, failed miserably. Each was designed to be a viral business but they didn’t scale. Nevertheless, Birch learned from his mistakes. It was these failures, I would argue, that made Bebo possible.

Birch’s first startup was called Lemonlink, a Web-based self-updating address book, which went live in January 2000. The viral hook consisted of users entering their friends and acquaintances’ contact information, and each person added to a user’s address book would be notified by email and asked for contact details. But Lemonlink grew slowly. Its viral coefficient was barely .5: each person led to only ¼ of an additional person signing on to the service, which was too low to result in exponential growth.

For a viral business the Holy Grail is a viral coefficient above 1, where each user is responsible for bringing onboard more than one additional person (on average). One user becomes two, then four, eight, to a million and more. Not unlike taking a penny and doubling it every day for a month. By the end of a week you’d have 64 cents and within two weeks, \$83.92; by day 30, about \$5.4 million. After day 38? You’d have more than \$1 billion (of course, most businesses taper off well before then).

Lemonlink’s laggard growth, Birch figured out, was partly because it depended on prospective users downloading an application off the web, which many resisted. It also relied on email notifications, but spam was a growing irritation, and people became less likely to respond to clickable entreaties. Birch coded a plugin for Outlook, Microsoft’s email program, which pushed the viral coefficient to .7, but that wasn’t good enough.

Still living in the U.K. Birch and his wife tried a babysitting business they thought would be inherently viral because parents of young children tend to know one another. But it was too niche. With money from remortgaging their house running low, Birch partnered with his brothers on startup number three, a decidedly non-viral business: online family wills. Borne less of passion than desperation, he wasn’t surprised when it, too, floundered.

Finally, Birch hit pay dirt in 2001 with the fourth business, Birthday Alarm, which began as a way to remind people of their friends’ birthdays.

“It was incredibly simple, an incredibly short viral loop,” Birch told me. “Everyone has birthdays so there is no reason you wouldn’t email everyone you know to ask them when it is.”

To seed the site, Birch paid \$99 to the “Cool Site of the Day” and on the first day 500 people signed up. The site grew slowly with Birch tweaking things until he could get it right. Initially he had plastered the site with privacy notices (promising not to sell users’ emails, etc.) but found it slowed page load times. Faster downloading led to an up-tick in virality and no one complained about the missing privacy notices. Simplifying the instructions also helped. In fact, the rule seemed to be the simpler he made things, the more viral the site became. People, it seemed, were turned off by anything that required them to think.

The big breakthrough came when he added a copy and paste function. That way, users didn’t have to type in friends’ email addresses and birthdays; they could in two keystrokes transfer them from other programs. This pushed the viral coefficient above 1.0, with 10,000 people a day joining. The site generated revenue from banner ads through advertising networks that took a 30-percent cut, and the Birches added e-cards, since it was natural that users would want to send greetings to their friends. By the time the couple relocated to San Francisco to be closer to Xochi’s family, the site was bringing in \$10,000 a month – enough to rent an apartment and a tiny 20-square foot office in the Valley’s highly inflated real estate market.

Two years after Birthday Alarm’s birth the site practically ran itself, and Birch sought other viral challenges. Not Bebo, at least, not yet. Enamored with Friendster, Birch decided to create his own social network in 2003, which he named after Beatles drummer Ringo Starr. It took him 13 days to code and within three months ringo.com had amassed 400,000 registered users. He was thrilled to have cracked the viral code but lacked the resources to scale. Despite the large mass of users, the site made almost no money and Birch couldn’t afford to pay for servers or additional staff. He had two choices. Look for venture funding or sell.

In December 2003, he received three offers in one week and sold ringo.com to online quiz purveyor Tickle.com for a couple of million dollars in private stock. Part of the deal required Birch to agree to a non-compete clause, which meant he couldn’t start a social network site for 18 months. This gave him ample time to continue experimenting.

He returned to his very first idea—a self-updating address book, applying everything he had learned along the way. He called it Bebo, a name he chose because it was short, snappy and utterly nonsensical (like Google and Yahoo) so that users could project their own meaning onto it, well worth the \$8,000 the domain cost. Later he tuned it into an acronym for “Blog Early, Blog Often.” He put the link up on Birthday Alarm, seeding it with just a few members, and it “went ridiculously viral,” so much so he took the link down after 7 days. The viral coefficient: 2.5. One million people signed up, 300,000 of them on day nine alone. By the end of a month, he had a few million members. But the concept wasn’t sticky; no one was coming back. After signing up, people didn’t bother to update their address books—and what good was an address book if only 10 percent of your friends were in it?

Eyeballing the data, Birch noted another pattern. Those with the biggest address books –say,500 names – were the first to sign on, and they tended to spread it to others with similarly large virtual rolodexes. That, he saw, accounted for the initial viral spike. Over time, the sign up rate decreased as people with leaner address books joined. But someone with 10 addresses was virtually impossible to reach because he was part of the small is beautiful flock, included in other less-connected people’s address books.

“So you end up exploding within all these well-connected people and filtering down to less connected people,” Birch said. “Your viral growth diminishes to the point you end up getting a viral factor of one,” which is where it remained, decreasing to 30,000 new members a day. Nevertheless, he had built a network of 6 million users. At the time, Friendster had already imploded and MySpace registered 9 million unique visitors.

While his non-compete prevented Birch from combining pictures with profiles – the definition of a social network in his agreement – there was nothing to prevent him from adding photo share to an address book, and that, too, proved reasonably viral. After the non-compete bar came down he and his wife, excited about the possibilities, spent three months redeveloping Bebo into a full-blown social network with profiles, pictures, blog and invitation system. It would be, at its essence, a community enabler, helping people to get to know one another without the societal constraints that characterized public interaction.

When Birch unveiled the revamped site, he expected, with all of the new functions he had added, it would take off. It didn’t. Traffic remained flat. Although it occurred to him that Bebo might simply be too late to the social network party he didn’t panic. MySpace was grabbing the kids and Facebook was targeted to college students. Bebo, on the other hand, by virtue of starting out as an auto-updated address book, had keyed in on an older demographic. But teens and 20-somethings are always the earliest adopters – they have a lot more time on their hands than adults and less money to spend – and social networks are free. It would take time to bring in the over 30 crowd.

He also suspected that repositioning the Bebo brand had sapped some momentum. While he counted more than 6 millions members, most had simply inputted their names and names and addresses their friends. When he browsed the network he encountered a sea of barebones profiles. It was like stepping in to your local pub and finding only a few customers. “We needed a critical mass number of people who had filled out their profiles to make it interesting,” Birch says. In the interim he kept tweaking the site, making it run faster, trying to improve the user experience. For two months Bebo treaded water. Then the profiles began to fill up with photos, comments and overlapping conversations between dozens of friends going off in dozens of different directions.

Birch realized that Bebo had finally attained a viral loop. From a base of 6 million users, the site was increasing page views by 10 percent a week. The steep course correction had proved the right move, but it was growing in a surprising direction. Another teen population had discovered Bebo: British youth, who took to the site with the enthusiasm American teens had when the Beatles first crossed the Atlantic. Over the next several months it spread to Ireland, Scotland, New Zealand and Australia. As one British commentator put it, teenagers were “taking to Bebo quicker than they can pop a can of paprika-flavoured Pringles.”

Bebo was forming viral clusters: It proved so popular in Ireland that several colleges banned access before the start of summer exams when students complained they couldn’t get near a computer terminal to work on their theses with so many peers logged on to Bebo. One college estimated half its bandwidth was consumed by students on Bebo while another found more than half of its computer terminals were logged on to the site at any given time. Birch estimated that 500,000 Irish youth were members, which, as one newspaper pointed out meant “virtually every teenager in the country has their own page on Bebo.” It ended up the largest trafficked sites in all of Ireland, passing Yahoo, MSN and Google.

Within 10 months, Bebo counted 24 million members leafing through 2.5 billion monthly page views. By the time Birch sold it to Aol on March 13, 2008, Bebo had grown to 40 million users spending an average of 40 minutes a day on the site.

Now, after Bebo had fallen into a state of disrepair and was viewed as a social network has-been, Birch has it back. I don’t know what he’ll do to bring it back. Perhaps he can’t, given how much as changed in five years. But I’ll be watching with interest.

Because Birch understands the power of the viral loop. And that means anything is possible.

Image: Wikimedia