Why, When, and How Much to Entertain Consumers in Advertisements? A Web-Based Facial Tracking Field Study

Abstract

The authors would like to thank Rich Sadowsky for customizing the web-based facial tracking system for this research. The authors would also like to thank Andy Dreisch and Brent Sapiro of Affectiva, Inc. for helping collect the data. The first author would like to thank John Deighton, Rohit Deshpande, Michael Norton and Rik Pieters for their valuable suggestions regarding an earlier version of this paper, and especially to Anissa Mak and Sumit Malic for valuable research assistantship. The usual disclaimer regarding responsibility for mistakes applies. 2 AbstractThe presence of positive entertainment (e.g., visual imagery, upbeat music, humor) in TV advertisements can make them more attractive and persuasive. However, little is known about the downsides of using too much entertainment. This research focuses on why, when, and how much to entertain consumers in TV advertisements. We collected data in a large-scale field study using 82 ads with various levels of entertainment shown to 178 consumers in their homes and workplaces. Using a novel web-based face tracking system, we continuously measure consumers' smile responses, and their viewing interest and purchase intent. A simultaneous Bayesian Hierarchical model is estimated to assess how different levels of entertainment affect purchases by endogenizing viewing interest. We find that entertainment has an inverted U-shape relationship with purchase intent. Importantly, we separate entertainment into that which comes before the brand versus that which comes after, and find that the former is positively associated with purchase intent while the latter is not. We conducted a large-scale field study by measuring entertainment 1 more naturally, in a web setting, and address three important questions. One, is there such a thing as having too much positive entertainment in ads? If too much positive entertainment can have detrimental persuasion effects in ads, and we find it does, then, two, it becomes important to understand how much entertaining content should be used. We find evidence that as much as a quarter of TV ads have passed the optimal level by dedicating too much of the ad time to entertain the consumer. KeywordsThree, does it matter when during the ad the entertainment occurs? We find that where entertainment is placed relative to the brand's first appearance has significant impact on the consumer's intent to purchase the brand.This research also provides new conceptual understanding of how the level of entertainment in an ad is related to its attractiveness and persuasiveness in increasing purchases.To do so, we propose a new individual-specific measure to identify whether the viewer associates the entertainment experienced in the ad with the brand based on the order of its presentation relative to the brand. We define pre-brand-associated entertainment (preBAE) as that which precedes the first presentation of the brand and post-brand-associated entertainment (postBAE) as that which follows the first presentation of the brand. These types of entertainment a...

Crowdsourcing facial responses to online videos: Extended abstract

“…The Nevenvision facial feature tracker 4 was used to automatically detect the face and track 22 facial feature points within each frame of the videos. To compute the smile probability measure we used a custom algorithm developed by Affectiva.…”

“…The area under the resulting ROC curve is 0.98. We also tested how well the smile classifier performs on crowdsourced face videos from a webcam where 3 http://www.ipinfodb.com/ip location api.php 4 Licensed from Google, Inc. There are significant differences in the smile responses between people that reported liking the ads more than others.…”

“…In addition to the commercial deployment, several academic studies have used our Internet-based data collection method [23], [24], [4], [5]. Teixeira et al [4] tested 82 ads and found that the entertainment an ad causes (quantified as smiling behavior) affects purchases.…”

“…Teixeira et al [4] tested 82 ads and found that the entertainment an ad causes (quantified as smiling behavior) affects purchases. The main finding was that entertainment has an inverted U-shape relationship to purchase intent (i.e.…”

“…In relation to advertising it has been shown that facial expressions exhibited in viewers while watching commercials can predict strength of recall of the commercial [2], skipping behavior [3] and purchase intent [4], [5]. The face has been described as the window to the soul, to quote Cicero (circa 100 b.c.)…”