Analytics Media

Parsing Upworthy’s Big Pivot

Parsing Upworthy’s Big Pivot

Equating Upworthy with clickbait has always obscured the real secret to their success: the company is among the web’s most deft practitioners of sophisticated analytics.

  • Since launching in 2012, Upworthy has done a few things very well.

    Upworthy perfected the socially-optimized “curiosity gap” headline. This was both a blessing and a curse—the site attracts around 20 million unique visitors per month, but it is synonymous with clickbait, an image they haven’t quite been able to shake even after shifting away from those thoroughly unloved headlines.

    Upworthy mastered curating the web—finding high-quality content with bad headlines, then tinkering with the framing to give it better headlines and help it find an audience. That has been Upworthy’s niche for the past three years. But that’s changed now that Amy O’Leary, formerly of the New York Times, has taken the helm as Editorial Director. Now, Upworthy is producing its own 5,000 word articles about the fast food industry. Its staffers are called “editors” instead of “curators.” This week, O’Leary released a new report on “Upworthy’s Editorial Vision,” which helps to define this pivot in Upworthy’s content strategy.

    As the company transitions to content creation, it’s worth asking what makes Upworthy different. Equating Upworthy with clickbait has always obscured the real secret to their success.

    It isn’t the clickbait headlines. Plenty of imitators picked up on Upworthy’s rhetorical stylings, leading Peter Koechley to announce “sorry we kind of broke the internet last year” at The Guardian’s 2015 Changing Media Summit. In fact, Upworthy has been moving away from “you won’t believe what happened next”-style headlines for years. They started using those headlines because they tested well. Then everyone else started using the same headlines. Then users got tired of them. So they didn’t test so well anymore, and Upworthy’s style shifted.

    In fact, “clickbait” doesn’t do justice to Upworthy’s viral model. The point of Upworthy was never to maximize clicks. It was to maximize shares. If people click on a headline and are disappointed with what they find, they generate a single page-view. (That’s nice if you have an advertising-based business model. Upworthy doesn’t, and never has). If people click on the headline and like what they find, then they’re more likely to share that piece of content through Facebook and Twitter. And that’s how the audience grows and progressive messages break out of their Filter Bubbles.

    It isn’t Facebook optimization. Facebook is the main driver of online news traffic today. Companies used to focus on “search engine optimization,” reasoning that the way people find anything new online is through google searches. Today, companies focus more on social optimization, recognizing that the balance of online power has shifted. But Facebook can be a cruel master. Tweaks to the Facebook algorithm can send your traffic plummeting. If you are too successful at generating traffic through Facebook, the site might just decide to duplicate your model and bring that traffic in-house.

    It isn’t the content itself. Part of what has made Upworthy different from its copycats is the focus on substantive content with a progressive bent. Upworthy, so to speak, is people. Hiring the right curators with the right taste and judgment has been a core element of finding and shepherding share-worthy social content. But that content has always come from mining the web. All of Upworthy’s top hits have come from some other media producer.

    No, what makes Upworthy special is their approach to analytics. Tracking the right data, thinking about data, and using data correctly has always been the thing they get more right than anyone else.

    Analytics was at the core of their initial launch. Eli Pariser and his team realized that simple A/B headline testing (the same type of testing that he’d done for years as Executive Director of MoveOn) could help build a massive audience for important stories that simply had never been framed right.

    Analytics was what drove them to clickbait. (Except it was actually sharebait.)

    And then analytics was what drove them away from clickbait. In February 2014, Upworthy stopped focusing on pageviews and unique visitors and started focusing on “attention minutes.”  

    The switch from pageviews to attention minutes is emblematic of what sets Upworthy apart from its competitors. Everyone uses some form of analytics these days. It’s relatively simple to measure pageviews, clicks, retweets, and shares. Analytics is basically how we keep score in modern-day journalism.

    But all this digital data can easily lead you astray, particularly if you’re measuring the wrong thing. Optimizing for pageviews leads to a bunch of stupid slideshows. Optimizing for clicks leads to curiosity-gap headlines, regardless of the story they link to. Optimizing for fundraising leads to an endless string of dire warnings.

    Take a look at slides 17-20 in O’Leary’s slide deck. Upworthy has spent a year and a half gathering rich data on attention minutes from tens of millions of visitors. That’s data they have and their copycats don’t. And now they’re starting to use that data for their own storytelling. Whatever happens next, they’ll monitor, listen, and learn, and adapt.

    The hard part of the new digital journalism isn’t the math or the software code that drives analytics and A/B testing. The hard part is figuring out what the right metrics are. Focus on the easy stuff and you’ll create a powerful engine driving you inevitably towards irrelevance. The center of Upworthy’s big pivot is a focus on the hard parts of analytics. That focus drove them to be the social web’s biggest curators. Now it will drive them as the social web’s latest content creators.

#PDF Analytics Election 2016



“Journalists should be pushing campaigns to answer ethical questions about the work they’re doing.”

  • Personal Democracy Forum is in less than three weeks, and we’re reaching out to some of the speakers for a quick preview of their respective talks and panels. First up is Ethan Roeder, who ran the data department for the Obama campaign in 2008 and 2012. Roeder stopped by Civic Hall yesterday to lead a brown bag lunch talk on political campaigns and data. We caught up with him afterwards to ask a few questions about the PDF panel he’ll be moderating, on the future of political analytics.

    You’re moderating a bipartisan panel at Personal Democracy Forum on the digital politics of 2016, featuring campaign consultants and analysts Scott Tranter, Kass Devorsey, and Masa Aida. What questions are you going to raise?

    The thing that I’m most interested in is the relationship between outside actors (like I360, Civis, and BlueLabs). The second thing would be innovations in analytics. One of the things I’ve heard is that building an individual-level candidate support model is a trivial affair in 2014. If building an individual-level of support model is no longer an advantage, what is?

    Roeder mentioned that he hasn’t yet figured out if he’s going to ask a framing question to get the panel started, and if so, what it will be. Perhaps people with strong opinions on the subject could offer suggestions in the comments.

    What questions should the media ask campaigns about their use of data this election cycle?

    Here’s what I hope doesn’t happen. As I mentioned in the talk, in 2012 we did a very effective job at hosing the media. And as a result they just came up with their own narratives. It wasn’t necessarily in our best interest but we effectively deflected all of their questions

    What I would hate to see happen is the media learn all the wrong lessons from that. Like Micah said, I hope they just don’t ask, “Does he have a Pinterest page?” There is a “Does he have a Pinterest page” equivalent for data, technology, and analytics. For example, “Are you using online data to target people?” Of course they’re using online data to target people.

    I think that journalists should be pushing campaigns to answer ethical questions about the work they’re doing. “Is there a line you won’t cross?” That’s the question I would ask. If you can’t give me a definition of your own understanding of the lines you won’t cross, it probably means you’re crossing a line.

    I’m less interested in the privacy question than Micah is, although that is a conversation that should happen. I’m more interested in the ethics.

    Did you run into any ethically dubious activities when you were working on campaigns?

    No specific instances.

    Campaigns exist to win. The only motivation of a campaign is to win. The only ethics of a campaign is whatever won’t hurt their chances of winning. I’m less interested in naughty things a particular campaign staffer did, and more interested in asking “What is your ethical compass? How do you determine what you will or won’t do?” I don’t think campaigns have an answer to that.