Last January comScore reported 70% of the total U.S. Internet audience watched at least one Internet video stream that month. It’s been reported that there are more than 12 million hours of video programming currently hosted online. YouTube has been reported to have more than 5.1m hours of content. With such an enormous, and rapidly increasing, pool of content one of the bigger challenges is how to effectively index and search the data. Without organization, it is too easy as the old saying goes “to be swimming in information but thirsting for knowledge.”
A host of video search companies are lining up to try and address the problem. The approaches vary. Some are focusing on text-to speech technology, others are looking elsewhere. It’s a new market, but one with such potential, that advertisers, eager for contextual delivery, are salivating over the potential.
One of the primary cataloging and search standard used so far has been keywords and meta-tags. They are user controlled. To work someone has to manually identify the content and title it appropriately. Social networking sites allow a wide range of people to share that responsibility (by adding unique “tags” as identifiers) but the trouble with user defined search is that it’s only effective when users use the same vocabulary. If I say tom-ay-to and you say to –mah-to we could get different results. Two people who are shown the same video might index it under two different subjects’ headings or with divergent keyword tags.
For advertisers hoping to serve ads alongside, or even in, the videos, this problem can be additionally vexing. They need contextual accuracy. They want to be able to target their advertising appropriately to their viewer – demographically, contextually, etc. But absent good indexing it’s difficult to accurately serve relevant advertising.
A new generation of technology is being developed to try and bring more automation and control to the index and search of video. Google, at the top of the food chain, is working on better techniques in house for YouTube and is porting its AdSense system for use with video. Other search giants, and social sites like MySpace. A host of startups, lean and mean with investor capital and less bureaucratic overhead than some bigger competitors, are also trying to crack the same problems.
Yesterday, San Francisco start up Blinkx unveiled what it is calling the first contextually relevant video ad platform. The platform, which was previously codenamed Project Trilby and will know be called Adhoc, has lofty ambitions. The service relies on video organization technologies including a speech-to text transcription element and a visual analysis protocol. Blinkx is hoping their tools will provide marketers means to customize both the appearance and timing of ads (shown pre, post and mid roll in the video stream). Blinkx has also announced it’s signed a deal to allow Real Networks to use Blinkx search technology underlying its ad network.
Elsewhere, in the research triangle of North Carolina near Duke University, another startup called Digitalsmiths just raised a $6m Series A financing to build its own contextual advertising technology for video. In San Diego, well funded and well connected IPTV company, VEOH, has rolled out a suite of applications to help collect, organize and distribute the depth of video programming that’s floating around. A beta of its player is now available. Veoh bills itself as a virtual television network that organizes, showcases, and delivers clear, full-screen video programming to anyone with a broadband connection. Effective video search is an essential part of their value proposition. In Redwood City, CA YuMe Networks, a well funded startup with backing from Accel Partners and Khosla Ventures, is trying to build a multiplatform video ad network that can target and distribute ads to PC’s, TV’s or mobile devices, by stream or download.
The present reality is video search is a nascent industry just laying a foundation but with IPTV, and various video content technologies and offerings growing exponentially, the significance of search is expanding comparably. Little money is being made in online video search/contextual advertising yet, but the opportunity is large. In the Blinkx announcement, the company’s founder Suranga Chandratillake characterized things by noting: "online video advertising [has been] a kind of Frankenstein’s monster – an attempt to cobble together technology that was built for Text Web banner advertising and apply it to an entirely new medium."
Television and Internet are converging. Advertisers are looking for ways to monetize the change. As companies battle to own the content, or lead in the distribution, as set-top and desk-top do battle, others are seeing opportunity in the niche of managing and organizing what’s out there. It’s not a bad place to focus. A TV Guide can be complicated enough with 100 channels to sort through, but with 100,000 (or some equivalent appropriate to a place-shifted, time-shifted video world)? That could be downright ugly.
As said before, we’re collectively “Swimming in information but thirsting for knowledge.” Video search companies are hoping to help quench the thirst.