Predicting Viral Video Success
Why do some internet videos “go viral” and rack up millions of views while apparently comparable videos don’t? For marketers seeking to use the video channel as a promotion tool, wouldn’t it be handy if you could predict which of several videos was more likely to succeed on the Web? OTOinsights, the neuromarketing division of One to One Interactive, has released a study that correlates various physiological measures taken while viewing videos with actual viewing traffic.
Analyzing the results from various physiological traces in combination with eye tracking and interview data, the t=zero research team presents a series of three insights with regards to how users engage with internet video. These include:
1. Viewer Responses to Internet Videos are Emotionally Complex.
2. Engagement Scores Substantially Enhance Interpretability of User Ratings.
3. Viewer Engagement and Video Success are Positively Linked.
One of the more interesting findings of the study dealt with video length. The researchers found no correlation between engagement, emotion, and the length of a video. In short, as long as the video is engaging, a video longer than three minutes can score as well as a much shorter one.
After turning the data collected from the subjects who viewed videos into a composite measure of emotional engagement dubbed the QEI (Quantemo Engagement Index), the study showed a correlation with actual pageviews on the Internet. This data is from a set of YouTube videos:
Web 2.0 … The Machine is Us/ing Us * 305 * 6 million pageviews
Completely Uncalled For * 219 * 5.7 million pageviews
World of Warcraft BigBlueDress * 195 * 3.6 million pageviews
The Evil Strawberry * 181 * 944K pageviews
Piece of Mind – Vancouver Film School * 124 * 864K pageviews
While this research isn’t exactly a cookbook for creating the next YouTube phenomenon, it’s worth checking out if you are in that segment – download the full report.
And, by the way, here’s the video that generated 6 million pageviews (now up to 7.5 million):