Dose

Emerson Spartz discovered the internet at age 12 when he convinced his parents to let him drop out of middle school and homeschool himself. About a month later, he found a WYSIWYG editor and decided to build websites. After creating several forgettable projects, he landed on something magical: Harry Potter fandom content. The timing was perfect—he was passionate about both website building and Harry Potter, and he threw himself into it, working eight hours a day to build MuggleNet.

MuggleNet grew to 50 million page views per month through three distinct strategies. First, Spartz executed massive link swap campaigns—he manually emailed thousands of Harry Potter website owners asking for reciprocal links. "I didn't even know how to copy and paste at the time," he recalls, "so I just typed emails over and over." He sweetened the deal by offering large banners to bigger sites in exchange for small links back. Second, he obsessively studied what other successful fandoms were doing—Simpsons sites, Lord of the Rings sites, Dragon Ball Z communities—and adapted their features like polls, guestbooks, and comments. Third, he recruited aggressively, building a team of 120 part-time and volunteer writers who could produce more content faster than any competitor.

The MuggleNet playbook proved he understood how to build communities and drive growth. When it came time to launch Dose, Spartz applied the same principles but at scale and with technology. Instead of manual link swaps, he would use algorithms. Instead of recruiting 120 volunteers, he would hire elite writers and engineers. He placed prominent "write for us" calls-to-action throughout MuggleNet, attracting the first contributors by clearly explaining why they should participate.

At Dose, Spartz built an entirely different machine—not based on volume of writers, but on data science. With a 50-person team that's mostly engineers and data scientists (not traditional media people), he created a series of proprietary machine learning systems. Kepler scours the internet to identify content with high viral potential. Writers produce content in Dante, an internal CMS. Lindell tests headlines and thumbnails. Lovelace identifies top performers. Darwin distributes to the right audiences. Timely content can move from idea to live publication in an hour; other pieces take about a day of testing. For testing, Spartz recommends a baseline approach for bootstrapped creators: write 10-20 headlines per article, ask five friends to vote on their favorites, and use whatever wins. At Dose's scale, they spend between $5-$10 to test headlines and thumbnails on Facebook before publishing, running multiple rounds of A/B testing across different audience segments. The company doesn't rely on email (just 500K subscribers compared to 27 million social followers) but instead pushes content through social distribution where it resonates most.

Dose produces approximately 50 articles per day (about 1,500 per month) across its main domain and OMGFacts, reaching 15 million unique monthly visitors and 27 million social followers. Having raised $35 million, the company currently monetizes through programmatic advertising. But Spartz's bigger vision is to commercialize the distribution and virality engine itself—building a software product that lets brands guarantee performance on native advertising the way Dose guarantees viral reach on its own content. By turning art into science, and then systematizing that science, he's built what may be the most predictable viral media machine in existence.