Like many great things, this all begins with a bet. It was 2001, and Jonah Peretti, a graduate student at MIT at the time, was procrastinating. Instead of writing his thesis, he had decided to take up Nike on their offer to personalize a pair of sneakers. Under a recently launched program, anyone could do so from their website, NIKEiD. The only problem, at least from Nike's point of view, was that emblazoning them with the word sweatshop, as Peretti had requested, was a non-starter. Peretti, in a series of emails, demurred pointing out that in no way did the word fall into any of the categories of objectionable terms that would result in his personalization request being rejected.

Peretti, believing others might find the back-and-forth with Nike's customer service representatives amusing as well, forwarded them to a number of close friends. Within days, the emails had found their way into inboxes across the world. Major media outlets, such as Time, Salon, The Guardian, and even the Today Show, had picked up on it. Peretti was at the center of a viral sensation.

But the question that began nagging at Peretti was, could this sort of thing be replicated? His friend, Cameron Marlow, had been preparing to write his PhD thesis on viral phenomena, and was adamant that such things were far too complex for anyone to engineer. And it is here that the bet comes into play. Marlow wagered that Peretti could not repeat the success he had enjoyed with that original set of emails with Nike.

Fast forward 15 years, and Jonah Peretti leads the website whose name has become synonymous with virality—BuzzFeed. With more than 77 million unique visitors in 2015, it ranked higher than the New York Times in total reach. I think it's safe to say that Peretti won that bet.

But how exactly did Peretti do it? How did he piece together the secret formula for creating content that spreads like wildfire? In this chapter, we'll attempt to unravel some of these mysteries. We'll examine some of the most shared content and attempt to find the common elements that differentiate it from the content people were less willing to share.

The following topics will be covered in this chapter:

• What does research tell us about virality?
• Sourcing shared counts and content
• Exploring the features of shareability
• Building a predictive content scoring model