• Heta Desai

Billion online video ad views occur every month. We did a comprehensive study of measuring facial responses to video material gathered through the Internet and how this corresponds to advertisement efficacy. We recorded more than 12,000 facial reactions from 1,223 test subjects for 170 commercials on different industries and product categories. Reactions were computer-scored frame by frame, amounting to 3.7 million frames—a ratio unthinkable for traditional research techniques. Our results show that single-shot expressions are rare, but group responses have rich emotional patterns over time. By investigating the interaction between face response and ad performance, we show that ad liking can be reliably predicted (ROC AUC = 0.85) from webcam-recorded face data. In addition, changes in purchase intention can be predicted (ROC AUC = 0.78). Ad liking is motivated by eliciting expressions, particularly positive ones. But inducing buy intent is harder than getting crowds to laugh: high- performing ads generate strong positive responses with immediate exposure of a brand. These results identify a strong and scalable approach for predicting ad effectiveness from automatic facial reactions, in isolation of viewer self-reported feedback. They offer informative clues regarding the structure of effective ads as well.

Facial expressions, emotion, market research

"Predicting Ad Appeal and Buying Interest: Large Scale Experiment of Facial Responses to Advertisements", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.2, Issue 7, page no.377-397, July-2015, Available :http://www.jetir.org/papers/JETIR1701D08.pdf

"Predicting Ad Appeal and Buying Interest: Large Scale Experiment of Facial Responses to Advertisements", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.2, Issue 7, page no. pp377-397, July-2015, Available at : http://www.jetir.org/papers/JETIR1701D08.pdf