What, then, are the two online video vehicles?
What users currently see most often is the page-based ad, in which video and sound are streamed within an "old-fashioned" banner. Such ads now constitute the overwhelming majority of online video ads. They use existing technology, and are akin to animated banners. In their start-up form they are constrained by the bounds of the banner, although recent advances have made them expandable.
One advantage of page-based video for advertisers is that a vast and cost-effective inventory of banners is already available, with many targeting options. On the minus side is the very nature of the banners themselves: since they are delivered within unrelated content, their appearance is unanticipated and may be unwelcome. The advertiser is consequently faced with the dilemma of whether or not to incorporate audio automatically, or only on-demand. If automatic, the uninvited noise will annoy some users, while confusing others as they try to spot its source; if only on-demand, however, the ad will communicate far less effectively with those who might be receptive to it.
Stream-in-stream ads are 15- or 30-second video commercials served before the main item to users who have specifically requested to view an on-demand video. Whereas page-based ads vie for users' attention both with the page's content and with its other ads, stream-in-stream ads are shown to users who are expecting to see video (and conditioned to expect an ad first), and whose attention is clearly directed to where the video will be shown. Stream-in-stream ads offer far greater impact than page-based ads, have enormous (mostly as yet unexploited) creative and interactive potential, and are free of the issues of intrusiveness affecting page-based ads.
For publishers who offer stream-in-stream, it is, unsurprisingly, far and away their most valuable inventory. But not many offer it yet; the big three of MSN, Yahoo!, and AOL are responsible for the overwhelming majority of these ads served, with sites such as CNN and ESPN following far behind in the second tier of users. For now, there is simply not enough inventory to satisfy demand, or to justify the other development that will, I believe, ultimately make stream-in-stream the powerhouse of online video: creative not merely adapted to the Web from TV, but produced specifically for the Internet and designed to take full advantage of the Web's unique flexibility and interactivity.
Although online video is still in its infancy, the page-based form has already come close to its full potential, I think. I don't mean by this that it will not see major technological advances; but when they come, these will be introduced more to deal with the intrinsic limitations imposed by the banner format than to take advantage of any big latent potential. The stream-in-stream model, on the other hand, has barely moved out of the starting block. The scope for creative input and melding of video, graphics, and interactivity is virtually limitless, and as yet virtually unexploited. Stream-in-stream is where there will be exciting developments, but it will be a little while before critical mass is achieved and the cost of creating video specifically for the Web is justified.