BuzzFeed’s New Strategy: Fishing for Eyeballs in Other People’s Streams

BuzzFeed doesn’t have any problem attracting Web traffic. The people who brought you The Dress say they’re getting 200 million visitors a month, even when they don’t have an astonishing viral phenomenon on their hands.

But CEO Jonah Peretti thinks BuzzFeed can become even more omnipresent — by focusing more on publishing his stuff on other people’s sites and services. Instead of trying to use the likes of Facebook, Twitter and Pinterest to send visitors back to his site, Peretti says his company is now happy to have its stuff live outside of BuzzFeed.

This is antithetical to conventional Web publishing wisdom, which says that you want to keep as many eyeballs as possible looking at your stuff on your properties so you can maximize the ad dollars that stuff generates. It also means you don’t have to worry about sharing ad revenue with other platforms, or scrambling to adapt if they change their minds about the kind of stuff they like.

On the other hand, we may be heading toward this world whether Web publishers like it or not.

Snapchat’s new Discover platform, for instance, which has captured the imagination of both publishers and advertisers, doesn’t send any outbound links to the people who put their stuff there. And Facebook is talking to publishers about something similar, where publishers would put their stuff in the social network’s “container.”

BuzzFeed floated this idea as an experiment last summer, when it announced a major funding round. But in the last few months Peretti has decided to turn it into a full-blown strategy, and has been laying it out for his staff in a series of briefings. Today, he’s making a public version of that presentation at South By Southwest.

Peretti and I also talked about it yesterday, and you can watch that conversation now, in the video below. He’s a pretty compelling speaker, and I’m out of the frame for the most part, so I’d suggest spending 12 minutes watching it — it’s a pretty immersive dive into BuzzFeed theory and practice. And then if you want even more, we’ve got more below that.

But! If you’re a time-pressed person who wants the whole thing summed up in one GIF, we can do that, too.

Translation: BuzzFeed says that in January, it received 420 million views via referrals from Twitter, Pinterest and Facebook. Pretty good! But it says it generated 18 billion impressions on those platforms — and it wants to start making the most of that giant opportunity.

Here’s Peretti explaining how the company figured out it wanted to head this way, and what that means in practice.

Still hungry for more BuzzFeed? We can accommodate that. Here are some edited excerpts of my chat with Peretti that didn’t make it onto the video, arranged by topic:

The “cascade” and what it tells us about Twitter’s surprising influence

Jonah Peretti: People tend to think of how much traffic they get from Facebook or Twitter. But we have recently been able to look at the cascade — what’s downstream from the initial post.

So in the case of The Dress, we have almost a million views to the BuzzFeed post, that came from a single tweet, from the BuzzFeed Twitter account. And Twitter only accounted for something like 280,000 of that. Facebook actually captured more of the cascade than Twitter. As a result of that tweet, people saw it, shared it with their friends on Facebook, and Facebook ended up getting more distribution from this tweet.

In general we’re seeing from our data that Twitter’s cascades tend to bleed into other networks more. And Facebook’s cascades tend to stay on Facebook more.

So people say, “Oh, Twitter doesn’t send that much traffic” — but we have this intuitive feeling that it’s more influential, and has a bigger effect [than people think]. Lots of people in the media see it. … The way that other blogs and publishers found our dress story was from the tweet, and they clicked through, and they re-shared, and posted it on their site, and that continued the cascade as well.

Different platforms, different content

Most of what we learned about this came from Ze Frank’s [BuzzFeed video] team. Initially they went for this approach where they were posting directly to these video platforms. And it was like flying an airplane without instruments — they only had the built-in analytics you get [from the sites] like view counts. They were doing this tight loop — an iterative process without very much data, and it was a little bit frustrating.

So we sent our data science team out to LA, and we sent one person out there for three months to work with them and build dashboards for them. They were able to pull data from the different platforms and put it together. And they started to realize things, like videos about identity and emotion do much better on Facebook. And videos about facts, and cool weird things do much better on YouTube. They started to get intuitions about what would work on different platforms.

What we’ve found is that whenever we create a team that iterates against a signal from a single platform, the videos do better and better on that platform over time. So there’s a team that really thinks about Yahoo, because Yahoo is so big, and we weren’t seeing the same kind of pop there. They focused on that, and Yahoo views went way up. By looking at a single platform, and iterating against that, you can make things connect to that audience.

How to make native ads go viral — or at least get them seen

We have an advantage, where we get to touch the content. It’s a lot tougher if you’re a pure tech solution, and someone comes to you with a really terrible ad and says, “Make this go viral.”

That’s where all of these viral agencies say, “Well, we can’t make it go viral, but we can buy the cheapest traffic possible, and get it to a million views, and we can tell them it’s viral.” So it ended up polluting what the word viral even meant. It lost its grounding in epidemiological math, and became about reaching some sort of specific number.

For us, we can look at it and say, “Here’s how to take a concept and shape it into something that has the best chance of spreading.” And then we can say, let’s just not do one thing, let’s have a portfolio.

If we do a portfolio, there’s often a big skew. So if we do 10 pieces of branded content, the top performing one or two or three are going to be a lot better than the bottom. You can also do a portfolio of formats: Do a list, do a quiz, do a video, do a Vine. So you can optimize against that.

That’s why every advertiser we ever work with always gets sharing and earned media. That’s something that when we started, a lot of people were skeptical about. It always works.

Then there’s the question of how well it works. Can you do something like “How Would You Die” that we did for HBO and “Game of Thrones,” that gets over a million views — or are you getting something that gets you 20 percent extra from sharing. The big thing is that 20 percent extra is 20 percent more than you get if you just buy something that you push in people’s faces.

This article originally appeared on Recode.net.