Want to Go Viral? Use Facial Coding

Using facial coding to measure the effectiveness of online video ads

Considering that most of our decisions are largely dependent on the subconscious mind, marketing research must go beyond conscious and explicit and dig into subconscious and implicit measurement. Emotions play an important role in our decision-making process and affect our behavior and go hand-in-hand with motivation.

Social network advertising has made viewer and target group engagement an even more important part of the advertising machinery than before. Although brand usage has always had something to do with how we perceive ourselves, self-image has become even more related to advertising. Whether we will share certain ads on social networks depends on whether it provokes a motivating emotional response or some other motive related to the image we want to create about ourselves. Online advertising is more and more reliant on the viral effect, which depends heavily on emotions.

Facial coding: emotion and advertising

Facial coding is an objective method for measuring emotions through people’s facial expressions. Facial expressions are spontaneous and some of them are identified as universal. Modern technology has provided an opportunity for researchers to track people’s facial expressions while they are watching any material, for example video ads. The process of facial coding consists of the stimuli exposure, recording facial expressions and coding expressions into separate emotions. The program records expressions on a second-by-second basis and then an algorithm codes them into seven emotions: Happiness, Surprise, Focus/Interest, Confusion, Disgust, Fear and Sadness. The output of the algorithm is always double checked by humans (trained psyshologists), to make it more accurate and provide better interpretation.

Connecting facial coding and viral advertising

EyeSee conducted an online video study to find out more about how facial coding interviews and surveys can predict the viral effect of online commercials, including its impact on sales. The online research study focused on a sample of 1,539 U.S. respondents and 52 online commercials for the wide-known brands such as Old Spice, Pampers, Sprite, Estee Lauder, etc. The study found that facial coding is two times more efficient in predicting potential to go viral, as opposed to the use of a more traditional method, like a survey.

The study looked at 52 online commercials from seven industries. For each industry, we selected two to three leading companies. For each company, we chose one best-performing and one least-performing online commercial, based on number of views, shares, likes and comments on YouTube and other social networks.

Respondents were first exposed to two-to-five ads while their emotions were tracked through facial coding. Respondents were then asked to fill out a survey to further evaluate the ads they saw during the study.

A two-staged process was used for emotion recognition. First, an algorithm automatically detected the respondents’ facial expressions, next, a human coder double-checked the coding. The expressions were coded into seven emotions: happiness, surprise, focus/interest, confusion, disgust, fear and sadness.

The study found that surveys explain 23 percent of viral effect of online commercials, while facial coding explains 45 percent. Thus, facial coding can predict the viral performance much better than surveys and therefore can keep marketers from investing into future failures.

Emotion and viral activity

Of the seven coded emotions, happiness was found to have the strongest relation with viral activities, meaning that when people feel happy while seeing the advertisement, there is a better chance that they will share, like or comment on social networks.

Happiness is the most frequent emotion as a response to advertisement. Even if the content is disgusting viewers can express happiness as well if they feel amused while watching the ad. Percent of happy viewers drops down as the number of viral activities decreases.

Advertising should be amusing. That is when people express happiness. That does not mean that sad or disgusting content won’t encourage viewers to share the ad on social networks. Any emotional content can be beneficial for viral effect of the advertising. Happiness in facial coding results will be an indicator that your ad performs well. A high percentage of negative emotions can indicate potential problems.

The highest chance to go viral

Among all tested survey variables, three were shown as significant predictors of the viral effect:

1. If the content is perceived as distinctive,
2. If the content is NOT perceived as relaxing,
3. If respondents state they want to share it.

Evaluation based on conscious decision goes in line with subconscious measurement (facial coding): engaging ads will go viral. If the ad is emotionally engaging (amusing) it will be perceived as distinctive and not as relaxing. At the same time, viewers will be willing to share it on social networks.

Sometimes results of implicit and explicit measures differ. One of the reasons is in the dynamics of human psyche: we are not fully aware of our preferences, attitudes, motivation and emotions. And sometimes we are trying to hide something (for example if it is not socially acceptable). It is always recommended to use both methodologies: implicit and explicit. That way the results are fully validated and if they differ it can lead to powerful insights with reference to psychodynamics and social influences on human behavior.

Facial coding in pre-testing

Making an effective ad is a multi-stage process which should include pre-testing of the creative solution.

Firstly, it is important to start making an ad with awareness that viral potential depends on emotional engagement. Certainly there are numerous valuable information we would like to communicate to our customers but if they are tedious it is all just a waste of money.

Secondly, our common sense (or creative agencies’ we hire) might be misleading when it comes to the preferences of the target group, especially in the context of global strategies and cultural differences. It would be smarter to test than to make a risky investment.

Thirdly, testing the ads can provide information humans are not able to predict without research. Methodologies are constantly developing and now we are able to dig into details. It is possible to differentiate scenes based on emotional impact they provide and optimize all sorts of details to provide the greatest advertising impact. Experimental designs provide information on independent effects of different details like colors, position, size, context etc. On top of that it is possible to implement ads in more and more popular digital advertising context.

Conclusion

Using facial coding in pre-testing can help marketers to choose scenes, length, characters and all other aspects of the ad aiming to increase emotional engagement the ad will provoke. This way the ad will have the best chances to go viral. Choosing content that target group can relate to and characters they can identify with will increase the ad’s chances to emotionally engage viewers.

As expected conscious measures seem to be less important when it comes to the social network behavior. Facial expressions and revealed emotions are much more powerful clue of what people will be engaged about in their virtual worlds. The future of market research is beyond the tip of the iceberg. Explicit measures should be always combined and validated by the implicit ones.