Feedery gets personal with content aggregation

By Published on .

Once you get people to your website, how do you keep them there and increase your chances of getting a registration, a click-through on an ad, a download or an e-commerce transaction?


But most business media companies can't economically produce enough content with their own editorial teams, so they add user-generated content, vendor-generated content and content aggregated from around the Web.

At Geeknet, content aggregation is a core competency.

Geeknet's latest launch, Feedery, located at, is a personalized content-recommendation engine, according to Garrett Woodworth, director of product for Geeknet. He said that, more than other existing aggregation models, Feedery is intended to incorporate a reader's personal preference.

Google News, the giant in the aggregation category, organizes news stories for the widest possible audience and also for broad topics such as sports, sci/tech, entertainment and business. At Digg and Reddit, stories are chosen by members of the community. “It's based on popularity, groupthink,” Woodworth said.

At Geeknet's Slashdot—a website that aggregates Web content for an audience of 4.3 million tech-savvy people each month and generates about 6,000 comments each day—stories are chosen by the crowd, then organized and presented by a team of editors. “Because the editors are well-respected and they [select] stories their community wants to read, really cool discussions have grown up around it,” Woodworth said.

Feedery adds a new layer to aggregation as its algorithm incorporates personal preference. An individual is presented a list of stories and can vote each one up or down. The system then “learn” a person's preferences and becomes more fine-tuned as a person uses it.

However, Woodworth said, the personalized voting method has the potential to produce a numbing sameness, reducing the user's interest and engagement over time. “You want a mix; you want a sense of discovery. It's our goal to make sure you're not reading the same stuff over and over again,” he said.

Feedery's proprietary algorithm was inspired by one developed for Netflix, which put out a public challenge to produce an engine that would help it select movies for customers, combining the users' past choices and stated preferences with ratings and preferences of other members to produce the most satisfying experience for each member.

Feedery's algorithm produces personalized story lists in a totally impersonal way. It takes the individual's votes into account, based on parameters like the news source and characteristics of each story, and finds similar users in Feedery's database with similar patterns. The technology then feeds each user stories based on a combination of factors.

“We don't have to know anything about individuals except their preferences,” Woodworth said. “We collect e-mails to set up accounts so we can store the [behavioral] data, but we don't store the e-mail. This is a way around privacy issues that is getting more important.”

Most Popular
In this article: