There are roughly 9,286 different methods to market something online. One of the most useful, however, is viral videos. A small budget can turn into obscene amounts of product exposure, mass brand recognition, and intense profits. Now, Orabrush managed to do this with its halitophobia videos, but other companies haven’t been so lucky. The reality is, what makes a video go viral is hard to pin down. Yahoo! Research, however, is trying to make headway on questions related to the success of viral videos. A recent project involved predicting viewership for viral, or would-be viral, content.
The study uses Yahoo! Zync, an instant messenging service that allows users to quickly spread content they’re viewing on the web. Zync information is aggregated and examined by Yahoo to determine the predictable success of media content. The ability to predict viewership is most accurate when the video has had over 10 million views, but the approach has a 95.8% success rate overall.
The hypothesis of the study is that “video which have a high YouTube view count will exhibit a unique sharing pattern” in the Zyncand other related environments. By examining this pattern in an attempt to learn what a viral video looks like, Yahoo! can then predict what videos are likely to go viral in the future. Specific patterns that the Yahoo! team stumbled on are included in the full study, which was published on the Yahoo! Research site, linked below. Information covered includes the number of sessions viewing a video and the user behavior on the video (playing, pausing, rewinding, etc.). For Yahoo!, this study is an important part of understanding not only viral content but the overall social sharing picture. Since Yahoo! has moved from being primarily a search site to being primarily a social resource, it makes plenty of sense that they’d conduct such a study.
[via the Yahoo! Research Blog]