BuzzFeed Just Cracked the Code on How Social Content Spreads, and It’s a Big Deal

By Joe Lazauskas April 29th, 2015

For a while now, BuzzFeed has been the envy of the publishing world for its ability to crunch data and figure out how to create content that’s perfectly constructed to spread through the social web. Even The New York Times couldn’t help but gush over BuzzFeed’s analytic prowess in last year’s leaked innovation report. Like a disturbingly optimistic version of Batman, the company always seem to have another advanced technological tool in its belt that makes for a unfair fight with any rivals.

On Monday, they revealed they’ve gotten their hands on the Batmobile.

In a BuzzFeed Tech blog post, publisher Dao Nguyen and data-genius twins Adam and Andrew Kelleher introduced Pound, a new proprietary technology that promises to drastically change the way everyone understands how content spreads through the web. Pound stands for Process for Optimizing and Understanding Network Diffusion, which sounds terrifyingly complex. (And it is.) But in layman’s terms, it simply shows how content spreads “from one sharer to another, through all the downstream visits, even across social networks and one-to-one sharing platforms like Gchat and email.”

So instead of just seeing your shares in buckets (100 Facebook, 50 Twitter, 30 LinkedIn, etc.) you see exactly how that content spread in a tangled web from its original sharer to other channels.

For instance, BuzzFeed analyzed how its six tweets about #TheDress diffused and spread across the social landscape. Those six tweets resulted in nearly a million views but also drove hundreds of thousands of views via other social networks. You can see the clusters of shares in the graph below—dark blue for Facebook, light blue for Twitter, and white for other publishers that picked up the story.

Image via BuzzFeed, by Adam Kelleher

The GIF below provides a look inside one of those share clusters; it’s fairly difficult to understand without reading BuzzFeed’s explanation two or three times, but essentially, you’re seeing how even though Twitter is the source of the share diffusion, the story actually spreads to many other social networks as it heads downstream.

As the explanation pointed out: “In fact, clicks from Twitter represent only a quarter of the total downstream visits rooted in the BuzzFeed Twitter account!”

Gif via BuzzFeed, by Adam Kelleher

But enough dorking out on data—what does this all mean?

Tens of millions of people share BuzzFeed’s content every month, and they’re social-first in every sense of the buzzword; 75 percent of the site’s 200 million monthly visitors come from social sources. BuzzFeed is already excellent at creating content that will thrive on social, and it tests the hell out of every element of its stories, from the headline to the order of points in a listicle, and optimizes accordingly.

If BuzzFeed can figure out how to optimize its content even further for social networks based on a deep understanding of how content spreads, the potential benefits are huge. Even just a 10 percent boost in the social performance of BuzzFeed’s stories would mean 15 million new visitors a month. (And likely more, since the benefits of social are inherently exponential.)

Lucky for us, BuzzFeed included its own list of possibilities for how it could use Pound:


  • Can we propose stories that will appeal not only to you, but also to your friends and followers?
  • Can we use Pound data to power A/B tests? Can we make the site and apps better not just for readers, but for their friends — and thereby increase the impact of our site?
  • How effective are specific promotions, not just based on first-order traffic, but on all of the downstream sharing and traffic that results?
  • Can we predict the potential reach of a story based on its content or other features about it?
  • Can we filter out the effect of big sites or celebrities promoting our content, learn what average people actually like, and produce more of the right content for everyone?


These are fascinating and exciting possibilities—particularly the ability to predict the reach of a story based on its features and content, which would allow BuzzFeed to eliminate duds and maximize hits. But the biggest implication could ultimately be in what it means for the company’s bottom line.

What this means for BuzzFeed’s business—and work for brands

In BuzzFeed’s blog post, the authors made clear Pound confirmed something they’d always assumed: Sponsored content spreads the same way as editorial content. They used a post from Target—“I Tried The Fanny Basket And It Saved My Life“—as an example. This is important because it allows BuzzFeed to use Pound to sell its sponsored content offering. Salespeople can say, “No one understands social like us; we know how to create viral content for your brand better than anyone else.” Then they show brand execs this image of Target’s viral diffusion and wait for $100 bills to fall from the ceiling.

Image via BuzzFeed, by Adam Kelleher

It’s clear that BuzzFeed is thinking about Pound largely in relation to sponsored content. After listing all the possibilities of how Pound could improve BuzzFeed editorial, the authors ask: “Finally, can we do all of the above for sponsored content? In fact, we are currently seeking beta partners to help us think about how Pound data can benefit advertisers and their audiences. “Since then, they’ve added a note announcing that they’re looking for beta partners to create data-driven content using Pound.

But in addition to sponsored content, there’s another way BuzzFeed could monetize Pound: by licensing it as a software product. If Pound truly lives up to the hype, most publishers will want to get their hands on it. BuzzFeed may not want to give away its secret sauce, but at the right price point, it could be worth it. And by the time the company does that, it’ll already be working on its next technological weapon anyway. All we can do is wait for Jonah Peretti’s LOL Signal to light up the Gotham night sky.

Image by TAGSTOCK1
Tags: , , , , , , , , ,