• Choose a category
• All
• Classic-Fiction

BUCKETS, MULLETS, AND CROWDS

Nowhere are the virtues of a measure-and-react strategy more apparent than in the online world, where the combination of low-cost development, large numbers of users, and rapid feedback cycles allows for many variants of virtually everything to be tested and selected on the basis of performance. Before Yahoo! rolled out its new home page in 2009, for example, the company spent months “bucket testing” every element of the design. Roughly 100 million people have Yahoo! as their home page, which in turn drives a great deal of traffic to other Yahoo! properties; so any changes have to be made with caution. Throughout the redesign process, therefore, whenever the home-page team came up with an idea for a new design element, a tiny percentage of users—the “bucket”—would be randomly chosen to see a version of the page containing the element. Then through a combination of user feedback and observational metrics like how long the users in the bucket stayed on the page, or what they clicked on, and comparing them to ordinary users, the home-page team could assess whether the element created a positive or negative effect. In this way, the company was able to learn what would work and what would not in real time and with real audience data.3

Bucket testing is now routine. Major Web companies like Google, Yahoo!, and Microsoft use it to optimize ad placement, content selection, search results, recommendations, pricing, even page layout.4 A growing number of startup companies have also begun to offer advertisers automated services that winnow down a large suite of potential ads to those that perform the best, as measured by click-through rate.5 But the measure-and-react philosophy of planning is not restricted to learning how consumers will respond to options they are presented with—it can also include the consumer as a producer of content. In the media world, this view is exemplified by what Huffington Post cofounder Jonah Peretti calls the Mullet Strategy, after the much-maligned hairstyle characterized by “business up front, party in the back.”

The Mullet Strategy starts from the conventional view that user-generated content is a potential gold mine for media companies, in part because users can greatly amplify and extend the content of, say, a news story. But even more, it allows users to participate in conversations around the story that change the nature of the experience—from pure consumption to participation—thereby increasing their engagement and loyalty. Just as is true in real gold mines, however, a lot of user-generated content is closer to dirt than gold. As anyone who reads popular blogs or news sites can attest, many user comments are wildly inaccurate or simply dumb, and some of them are downright mean. Regardless, they do not constitute the kind of content that publishers want to promote or that advertisers want to be seen next to. Moderating online comments is the obvious solution to this problem, but it tends to alienate users, who resent the oversight and want to see their comments posted without filters. Editorial oversight also doesn’t scale well, as the Huffington Post quickly discovered: A handful of editors simply can’t read fast enough to keep up with potentially hundreds of blog posts every day. The solution is the Mullet Strategy: In the back pages where few people will see any particular story, let a thousand flowers bloom (or a million of them); then selectively promote material from the back to the front page, with all its premium advertising space, and keep that under strict editorial control.6

The Mullet Strategy is also an example of “crowdsourcing,” a term coined in a 2006 Wired article by Jeff Howe to describe the outsourcing of small jobs to potentially very large numbers of individual workers. Online journalism, in fact, is increasingly moving toward a crowdsourced model—not just for generating community activity around a news story but also for creating the stories themselves, or even deciding what topics