Upworthy is one of fastest growing media startups in history. While recent critics have been quick to attribute their success to the use of click-bait headlines, the real driver of their growth is something much more fundamental: a strong culture of testing, analytics, and optimization.

The Upworthy approach doesn’t depend on any supposed Upworthy Headline Formula™ that gets everyone on Facebook to click all their links. Formula-based success fades as social media users change their preferences. While today’s social media users may love headlines “that restore your faith in humanity,” those headlines can become tired and cliched with overuse. Instead, they’ve built a strong culture of analytics and testing that empowers Upworthy curators through instant feedback on what works and what doesn’t. As audiences on Facebook stop clicking on headlines that are in the “Upworthy signature style”, Upworthy will use their data-driven system to find new patterns and insights into what people share.

What is a culture of testing?

A culture of testing isn’t just a set of rules or best practices, but a commitment to measure the success of your choices and test your intuition with data. Instead of receiving and following social media strategy rules and guidelines, you test those assumptions with data.

Although social media managers use social media analytics to find out what posts performed well in hindsight, running a rigorous analytics program is more than looking at numbers. Upworthy has an in-house A/B testing system that sends sample headlines to a sample audience and provides curators with feedback on how headlines performed. A/B tests, analytics, and frequent audience surveys inform Upworthy’s content strategy.

What does a culture of testing look like

Although the benefits of a culture of testing are obvious, it can be difficult to imagine what a culture of analytics and optimization looks like for day-to-day operations. A profile of Upworthy by New York Magazine gives a behind-the-scenes look of how Upworthy uses their data to inform headline decisions in a constantly-changing social media landscape.

The advent of click testing seems to have been a pivotal development for touchy-feely Upworthy, with each staffer developing his or her own balance of hard data and intuition. ([Editorial Director, Sara] Critchfield likes to dismantle this binary by talking about ’emotional data,’ arguing that a gut feeling is every bit as meaningful as hard numbers.) The foremost data-lover might be Mordecai, a former actor and Howard Dean organizer—he likes to credit his audience-riling for enabling that memorable Dean ‘scream’—who beams into video meetings from Denver, and seems equally loved and head-shakingly tolerated by his colleagues, like a grumpy uncle whose tics everyone’s learned to enjoy. Eisenberg tells me with some happiness that Mordecai didn’t have a single ‘hit’ until click testing came along; now he’s the staff’s biggest advocate of using data to optimize content, testing dozens upon dozens of headline variations until one succeeds. ‘The guy’s brilliant,’ says Critchfield. ‘But because he’s so brilliant, his ideas are, like’—she mimes a scatter with her hands. ‘And the data channeled his energy in a really powerful way. He was like: I could do this, or this, or this. And the data was like: Do that one.'”

Upworthy curators are given the freedom to try new headlines and different ways to approach content, but data is integral to the process, by testing assumptions and filtering ideas. By maintaining this strong culture of testing, Upworthy will likely weather changes in the Facebook algorithm and the evolving tastes of social media users in the future.