Evri Teams Up with The Times of London, Helps Online Audience Browse the Web Better
Seattle-based Internet startup Evri is announcing today that it has formed a partnership with The Times of London, one of the UK’s leading newspapers, to provide content recommendation software for online articles. For selected stories in the Times Online, Evri’s widget shows up next to the text with a list of links to related articles from the paper’s archives (and some outside websites). The news comes on the heels of a similar deal with the Washington Post that Evri announced last month.
It’s all part of Evri’s continuing efforts to build its audience, says Neil Roseman, the company’s founder and chief executive. Evri was started in 2007, and is backed by $8 million from Paul Allen’s Vulcan Capital. The concept is to use natural language processing to understand connections between entities on the Web—people, products, things—and help you browse what you’re interested in more efficiently. While it’s not intended to replace search engines, Evri does want to change fundamentally how people browse the Web. (Roseman recently wrote about presenting some new features at DEMO 09.)
Financial terms of the Times Online partnership weren’t disclosed, but it is important for several reasons. One, it is a “very significant traffic source,” says Roseman, who points to the site’s 20 million-plus monthly users. Two, it “reinforces our focus on English language articles, not just American journalism,” he says. And three, it “helps further prove the model.”
That model says people will appreciate being directed to other content on the Web that is related to what they’re reading about, not just because it contains some of the same keywords, but because the underlying meaning of sentences in the articles is related. Roseman gives the simple example of Evri’s software being able to tell the difference between Michael Jackson, the pop singer, and Michael Jackson, an actor, from an article’s context.
But one of the technical challenges is how to understand meanings and relationships reliably in everything from a 140-character message on Twitter to a full article in The Economist, say. “That hadn’t been done before with natural language processing,” Roseman says. “The technology is definitely getting better.”
Roseman, a former vice president of technology at Amazon, says he has learned a couple of startup lessons so far. “Cracking the search and SEO [search engine optimization] jujube is even more difficult than I thought,” he says, referring to efforts to get Evri’s pages indexed by search engines like Yahoo and Google. “SEO experts have different answers, but it’s all tea-leaf reading.” Roseman says he also should have planned for deals with media companies to take a long time. “I’m reminded that the sales cycle with larger media companies takes a lot longer than expected,” he says. That’s in part because there has been a lot of turnover at news sites lately. “People are willing to try things, but when the players change, you start discussions over again.”
As for the effects of the economic climate (including the turmoil in the newspaper industry) on Evri’s strategy, Roseman says, “One good thing about being a very early-stage company is that when your revenue is zero, you can’t be hurt that much. We’re still on a two-year horizon [for revenues].” He adds that Evri is starting to see good traction and is understanding better who its users are. More broadly, he sees a “world where journalism is more important than newspapers themselves.”
For the near-term future, Roseman points to areas of “overlap between social media, deeper understanding of websites, and smarter content” as being important to exploit. We didn’t get a chance to explore this in more depth yet, but I took this to mean helping people navigate and annotate content on social networks, and connecting that information to the rest of the Web—while also bringing more value to existing news sites through links to related content.
It’s still a little too early to talk about Evri’s prospects for success in a field that includes companies like Sheffield, UK-based Ensembli, Mountain View, CA-based Kosmix, and San Francisco-based Twine (Radar Networks), which is also backed by Vulcan. Roseman says his company’s business model includes advertising, affiliate sites, and eventually licensing direct revenues from companies who use Evri’s software. But for now, he reiterates, “It’s all about building the audience.”