Consumer Evaluations of Video Ads Portraying Human-Like Avatars

Abstract

Despite that Television (TV) remains the primary outlet for advertising through video format, most companies cannot afford to run TV ad campaigns due to the high airing cost of TV ads and the high production costs associated with the creation of commercials. A tentative solution to reduce some of these costs is a new format of video ads that is gaining popularity: avatar-based commercials. Holzwarth, Janiszewski, and Neumann (2006) define avatars as a “general graphic representations that are personified by means of computer technology.” These representations may be human-like, animals, or objects. In avatar-based video commercials, the ads contain realistic animated characters in computer-generated locations instead of actors in real places. A remarkable example is an animated Coca Cola ad named “Heist,” that won the Emmy Award in 2009 for outstanding commercial (www.emmys.com). Although the idea of incorporating avatar-based video ads to a company’s marketing efforts is tempting, the effectiveness or lack thereof of this type of ad format remains unknown. This exploratory study attempts to fill this gap. In this study, the ad effectiveness of three ads is compared across two different types of viewing instructions. More specifically, this study employed a 2 (Instruction: storyboard draft vs. final version) × 3 (Ad type: human Airwick vs. avatar Airwick vs. avatar Snow Joe) between-subjects design. Each participant was randomly assigned to view one of six possible combinations. One hundred and eighteen undergraduate students at a large Southwestern university participated in the study in exchange for extra credit. Participants completed the online questionnaire in a computer lab. Three ads were employed in this study. Two ads portrayed an indoor air fragrance carried by Airwick. These two ads were identical in content, context, duration, and narrative. However, one of the ads shows real actors and real settings whereas the other one shows human-like avatars in animated settings. The third ad was avatar-based and portrayed a new model of snow blower by Snow Joe. This ad was included to test for possible differences in familiarity across product categories. Participants were also given one of two viewing instructions. In one set of instructions, participants were told to evaluate a storyboard (draft) of an upcoming TV ad that was going to appear on nationwide television. This instruction was especially relevant for the avatar-based ads since viewers might have erroneously thought that this type of ads could only be a draft version. Furthermore, there are theoretical reasons to believe that the storyboards may induce imagination in the evaluation of the ads, thus influencing information processing. The measures of ad effectiveness were attitudes towards the brand, purchase intentions, and unaided brand recall. All study constructs were measured using multi-item measures adopted from existing literature. The results of this exploratory study show that participants reported more positive attitudes towards the brand and higher purchase intentions for an avatar-based ad than for a human-based commercial only when participants were told that the ad was a draft. In addition, the avatar-based ads performed better than human-based ads only when viewers were familiar with the product category. Further, brand recall was similar between an avatar-based ad and a human-based ad when the ad portrays a familiar product. However, brand recall was lower for an avatar-based ad when the ad portrays an unfamiliar product. These findings suggest that companies may benefit from showing avatar-based previews on their websites as much as they benefit from showing human-based final versions of the ads. In addition, the results show that familiarity with the product is an important factor for the success of avatar-based ads. However, managers should interpret these results carefully. There are several limitations in this exploratory study. A larger sample is needed to strengthen the results because small samples are sensitive to extreme responses. Furthermore, a more heterogeneous sample should be used to test for the generalizability of our results.