With a Lifeline to London, Blinkx Builds the World’s Largest Video Search Index
Celebrity obsession isn’t limited to Hollywood. Here in Silicon Valley, we’ve got our own versions of Brangelina and Justin Timberlake—they just go by names like Google, Apple, Facebook, and Twitter. Part of Xconomy’s mission, though, is to look beyond the headlines for firms that are innovating in unusual ways. So it’s been a pleasure to get reacquainted lately with a San Francisco video search company called Blinkx.
I first met Blinkx’s founder, Suranga Chandratillake, back in 2004, shortly after the startup had been spun out by Autonomy, the San Francisco- and Cambridge, UK-based enterprise software company. Blinkx was taking technology developed at Autonomy to search unstructured data such as corporate documents and applying it to the consumer world, especially desktop computers. But within a year or two it had switched its focus to Web video. Even then, at the very dawn of the YouTube era, online video was burgeoning. Now, of course, it sucks up more bandwidth than anything else on the Internet. So Blinkx’s switch was prescient. And today it’s accumulated the single largest index of video content on the planet, covering more than 35 million hours of searchable content—more than Yahoo or even Google.
Blinkx hasn’t become a household name like YouTube, in part because the lion’s share of its business is providing video search services to other companies, such as Ask.com, Real Networks, and Infospace. But it’s a solid company that went public in 2007 and recently finished its first profitable half-year ever. When I caught up with Chandratillake recently (it’s pronounced Chawn-dra-TILL-a-kuh), I asked him to brief me on the company’s latest search technology, its revenue sources, and its unusual (for a U.S.-based company) decision to float its stock on the London Stock Exchange.
The tale Chandratillake told—involving years of experimentation and fine-tuning, both on the technology side and the business side—is actually far more typical of most Silicon Valley startups than the instant-riches stories that garner so much press attention.
What Autonomy, and its offspring Blinkx, are best at is working with unstructured data—stuff that doesn’t fit into the neat rows and columns of relational databases. That covers a lot, from e-mails to word-processing documents to Web pages, music, photos, and video clips. In fact, in its first couple of years Blinkx was known mainly for its desktop search tool, which indexed files on the hard drives of Windows computers and automatically grouped them into folders by context.
“Local machines were getting complex enough that you needed advanced search technology for that,” says Chandratillake. And for a short while, Blinkx’s tool was the only comprehensive desktop search utility available. But Microsoft, Apple, and Google were working hard to catch up, and they had the advantage of having hooks into their own operating systems and Web search indexes. Blinkx was overtaken, and ultimately abandoned desktop search. “What brings our team to work is not fighting a battle that everyone is fighting, but being able to define a new battlefield,” Chandratillake says now.
To survive, the company refocused its efforts on video search, which was “a great example of a place where our technology can provide a very valuable difference,” Chandratillake says. It’s also “a fast-growing area, with lots of opportunities to start new businesses,” he notes. To index a video clip—that is, to make it easier to find through an online search—Blinkx’s technology goes at the clip from every angle: it uses speech recognition algorithms to create a transcript of spoken words, it uses computer vision techniques to understand in a general way what’s being shown, and it extracts metadata such as closed-captioning information.
Today, Blinkx’s spiders constantly crawl the Web for new video, and the company also receives a constant flow of video from 720 content partners, from media giants like NBC Universal and Fox to small independent producers. If you initiate a video search at Blinkx.com or any Blinkx-powered site such as Ask.com, you might get a result from the general Web, in which case you’ll zoom off to a third-party site, or you might find a video owned by one of Blink’s partners, in which case Blinkx gets a cut of the advertising revenue generated by your view.
That’s one of the main ways the company makes money. It also has an advertising-placement platform called AdHoc, which leverages the company’s understanding of videos in its index to match them with relevant pre-roll or overlay ads. Like Google, Blinks also sells sponsored video search results, primarily to content owners looking to boost their video ads into viral territory. Then there are the licensing fees Blinkx collects by letting other sites plug into its search index (although those fees only kick in above a certain threshold—a “few thousand” searches per day, Chandratillake says). The most common search licensees: newspaper website publishers who want to show lists of relevant Web videos alongside text article, to keep visitors on their sites longer.
Thanks in part to a jump in the average size of its advertising deals, and in part to a 107 percent increase in the amount of online video available in the US between September 2009 and September 2010, Blinkx earned $27.4 million in the second half of the 2010 fiscal year—which was more than double its revenues from the same period in 2009.
“What we’ve begun to realize is that the real asset we’ve built is this amazing index, which is available through a very powerful search technology, so that you can express what you want and easily get a relevant response,” says Chandratillake. “The newspaper idea came from them, not from us. We’d always focused more on the pure video search model, but we tried it out and realized that for any given article that a popular newspaper publishes, we have hundreds of relevant videos.”
Blinkx has never raised any venture backing; its root within Autonomy meant it didn’t have the usual up-front development expenses. But by early 2007—around the time the company won the “Demo God” award at the annual Demo conference—the company realized it would need to raise money to keep growing, Chandratillake says. None of the potential venture and private-equity investors the company met with seemed to be a good fit. “Some were offering a valuation that we didn’t think made sense, others had their own vision of the technology and where the company should go,” he says. “We just didn’t see eye to eye.”
So Chandratillake chose a path that’s become rarer and rarer for startups: an initial public offering. But not on the NASDAQ or the New York Stock Exchange, which would have been the standard capital marketplaces for a Silicon Valley tech company. Blinkx went out to raise $50 million on the Alternative Investment Market (AIM), a part of the London Stock Exchange.
That plan wasn’t wholly nonsensical. Chandratillake is native of Britain, and Autonomy itself is listed on the London exchange. Still, when consultants at Citibank recommended listing the company there, “It was a really bizarre idea,” Chandratillake recalls. “The first time they suggested it to us, I didn’t really take it seriously. But the more we looked at it, the more I liked the idea.”
For one thing, AIM is set up for small, early-stage companies, and the regulatory standards it imposes are far less onerous than the Sarbanes-Oxley rules that govern public companies trading in the United States. Listing in London would also give Blinkx access to “the right kind of investors,” Chandratillake says: “Institutional investors who are interested in buying something they believe in, but not in trying to be part of the team the way VCs want to.”
The July 2007 IPO was oversubscribed by a factor of three, so the company could easily have raised $150 million or more, says Chandratillake. But he says he only wanted to raise enough money to see the company through to profitability within two or three years—which is exactly how things have turned out. (To keep the growth going, the company placed $31 million in new shares this October. The company is now valued at £221 million, or about $343 million.)
The future, for Blinkx, involves expanding onto mobile platforms—Chandratillake thinks half of the company’s traffic will come from non-PC devices by 2013—and to other video interfaces such as set-top boxes. The 110-employee company recently partnered with San Francisco and Seattle-based Evri to insert relevant videos into its “EvriThing” mobile news aggregator apps.
But “video is just one type of unstructured data,” says Chandratillake—and in a development that harkens back to its desktop search days, the startup is also helping consumers locate other types of information. Using an approach called “implicit search,” for example, the company can pull up data related to whatever is on screen. The first example of this concept in action is Cheep, a browser add-on that Blinkx launched in October to help people find the best prices for e-retail goods. When a Cheep user visits a site like Amazon, the add-on provides price comparisons for whatever product is showing in the main browser window.
“I’m really bullish on this implicit search idea,” says Chandratillake. “Browser augmentation is really trendy at the moment, with so many plugins available for Chrome and Firefox. I’m a news junkie, and you can imagine [a plugin] that displays a matrix of every information source you might need” alongside every news article.
Whether Blinkx is actually building such a plug-in, or how it might be monetized, Chandratillake didn’t say. But given the company’s history of experimentation and adaptation, it wouldn’t be a surprise to see the idea come to life. “Video is just one type of unstructured data,” says Chandratillake. “It’s just the first application of what we could do.”