One of the hottest marketing techniques in the last year has been viral video – create a clever enough video segment, and with minimal promotion it can reach a Super Bowl ad-sized audience for a tiny-fraction of the cost. The trick is creating a video engaging enough to “go viral”, i.e., induce viewers to forward it to their friends. Now, a study has tried to use a neuromarketing approach to identify videos with viral potential:
…The research respondents were asked to observe eleven viral videos, which were pre-rated by the site’s community members on a scale of 1 to 10, to test the hypothesis that neuroscience informed measures of media engagement could accurately predict the ratings that the NewGrounds online community pre-assigned to the viral content.
In order to test the level of emotional engagement of each respondent, each was outfitted with a state-of-the-art garment with embedded biometric sensors that Innerscope uses to non-invasively measure physiologic manifestations of brain activity. While wearing this garment, the respondents observed the eleven viral videos and their biological responses, including skin conductivity, heart rate, respiration, motion and eye tracking, were captured.
“Once the data was collected we processed it through an algorithm developed for the purpose of deriving a single measure of emotional engagement,” said Dr. Carl Marci, Chief Science Officer and co-founder of Innerscope Research. “Specifically the algorithm determines synchrony, or the degree to which an audience’s state uniformly changes when exposed to a media stimulus, and intensity, the cumulative strength of physiologic response to the media stimulus.” (Press Release)
The researchers, from by One to One Interactive and Innerscope Research, report that they were able to predict the popularity ranking of the videos with 77% accuracy purely from the physiological data collected. It’s easy to see why this would be useful – subjects could be shown different edits of the same video, or different videos, to determine which were most likely to go viral. A final “optimized” video could then be released and subtly promoted.
Notably, the study did not use the tool most commonly associated with neuromarketing and neuroeconomics studies, an fMRI (functional magnetic resonance imaging) machine. While fMRI scans provide interesting detail of brain activity in (more or less) real time, they are big, expensive machines that few researchers have access to. The position of the subject and the intense magnetic fields also limit the ways that subjects can be exposed to and interact with content. A cheaper, non-invasive testing approach with good predictive ability would be a boon to neuromarketers in areas where the approach was shown to be effective.
I think it would be interesting to try to identify particular individuals whose reactions seemed to be above average in predictive ability. I’m sure in an given test, a particular viral video caused a range of reactions among the subjects. That’s a general characteristic of neuromarketing, and is why we are unlikely to ever see “super ads” that turn consumers into mindless drones. Still, I’d guess that enough testing would identify certain individuals whose reactions correlated more highly with real-world viral performance than those of other individuals.
One wonders how long it will be before Hollywood starts wiring up test audiences to make movies scarier, funnier, more romantic, etc. One suspects they already are, but if so, why do they keep releasing a significant number of movies that die at the box office?
I’d have liked to see a control where subjects also rate the videos themselves. If subjects are at least ~75% accurate in rating them, then a survey would suffice instead of all the physiology.