Video ads are everywhere, yet consumers rarely view them in their entirety. Tuck professors Prasad Vana and Scott Neslin show how to reduce audience abandonment.
If you’ve browsed the Internet recently, you’ve probably noticed a significant uptick in the number of video advertisements popping up. Your eyes are not deceiving you. In the past five years, the online video ad industry has grown from $24 billion to $74 billion in size, and now video ads make up almost a third of all digital advertising.
What’s behind this rapid surge? The theory goes that video ads are more effective than traditional static display ads because they allow brands to tell a more vivid, engaging story that people will remember. The downside to video ads is they cost more to produce and air. They also can be intrusive and irritating to consumers.
Professor Prasad Vana (top) teaches the core Analytics I and II courses, and the Quantitative Digital Marketing elective in the Tuck MBA program. Scott Neslin, the Albert Wesley Frey Professor of Marketing, teaches Statistics for Managers, Marketing Management, Sales Promotion, Customer Analytics, Marketing Research, Marketing New Products, and Decision Analysis in the MBA program.
Notwithstanding those pros and cons, a major challenge to video advertisers is keeping the audience’s attention for the entire ad. In a new working paper, Tuck marketing professors Prasad Vana and Scott Neslin conduct a groundbreaking study that models audience abandonment as a function of the creative elements of a video. Their unique statistical model is the first to determine the right mix of themes that will hold people’s attention.
To conduct their study, the researchers obtained a database of more than 200 movie trailer video ads displayed on Facebook, the second-biggest player in the video ad industry. The movies were released in 2017 and 2018 and targeted the 18–30-year-old market. The video durations ranged from 30 to 60 seconds, and the data showed that 90 percent of those who watched the first three seconds of the video did not watch the entire ad. Consumers abandoned the ads at different times, but 70 percent of them tuned out within the first half.
Vana and Neslin sought to determine whether the video’s creative elements, such as the use of suspense, influence abandonment; whether elements are more influential at different points in the video; and whether they could predict abandonment as a function of these elements. Overall, they set out to develop a statistical model to provide insight into the dynamic impact of creative themes on audience abandonment.
One of the main research challenges here was to figure out how to quantify the creative elements. Because machine learning algorithms cannot yet capture high-level creative themes in video content, the authors recruited research assistants to comb the literature and identify 20 distinct creative elements, which they narrowed down to 16. They included elements such as action, fantasy, fear, humor, romance, suspense, and violence, among others. After that, the authors recruited nine judges blind to the objectives of the study to go through about 80 videos each and rate each quarter segment of the video on all 16 elements. This was further distilled to yield six distinct themes: action, suspense, romance, music, story, and text.
The authors divided each video into four time segments. The judges quantified the creative themes that appeared in each segment, which Vana and Neslin were able to match to the abandonment rate at each segment. They were thus able to correlate those variables and model them.
What was the upshot? Their model showed that emotional themes (action, suspense, romance, and music) generally reduce abandonment while informational themes (story and text) generally increase it. Moreover, the model indicates that different creative themes play different roles over the course of a video.
The thematic order most likely to retain the audience, in this context, was action in the beginning, romance in the middle, and suspense at the end. The strategy this suggests is, first get the audience’s attention with action, then create interest with romance, then leave them wanting more by ending with suspense.
Other genres and target audiences may require a different strategy, but the main point—using creative elements differently over the course of the video—should hold.
— Prasad Vana, Associate Professor of Business Administration
The authors caution that the research on this question is in its very early stages, but that their model can be deployed by companies looking to analyze audience abandonment of their video ads. With the right data correlated to the creative themes, firms can model audience retention during ads for everything from cars to clothes detergent.
“Other genres and target audiences may require a different strategy,” Vana says, “but the main point—using creative elements differently over the course of the video—should hold.”