Using facial analysis for demographic insights
Are there some questions that only images can answer? Is the human face our gateway into understanding the consumer in a way that words could never equal? These are questions we ask a lot at Crimson Hexagon because we know how important they are to the future of social analytics. But knowing that these questions are critical is the easy part. The hard part is getting to the bottom of them.
That’s why Crimson Labs, our R&D arm, has been exploring the use of facial analysis to better understand audiences by identifying attributes such as age, gender, sentiment, and ethnicity. In our experimentation we found several cases where the demographic information derived from text was vastly different from what we found in images.
Consider this hypothetical example: if we were to conduct a demographic analysis of the audience talking about the popular Disney movie ‘Frozen’ using solely text attributes, we would find that the age of that audience is skewed older (35+). However, the people discussing ‘Frozen’ on social is not the only audience we’re interested in. It’s also important to consider the audience made up of the individuals in the photos, which in many cases are young children. While the parents may be posting the content, it is the children that are actually appearing in the content.
This is an important distinction to make as it can help brands and marketers differentiate between their customer (parent) and their end user (the child).
Another use case for facial analysis involves Ralph Lauren. According to census data, most births in the United States occur in August. Ralph Lauren is known for its popular line of high-end children’s clothing. During this time of year, there is a flood of images of newborns sporting new outfits. As a result, Twitter and Instagram feeds are filled with cute pictures of babies in Ralph Lauren onesies. This, of course, causes a spike in the mentions of Ralph Lauren on social platforms.
Using facial analysis, we would be able to detect the shift in audience demographics and alert Ralph Lauren that the spike is due to an increase in the purchase of baby clothing and not necessarily an increase in the purchase of women’s or menswear.
Without the human face, this type of analysis is impossible. Text can tell us a lot about certain audiences and their preferences, but without facial analysis, we’re missing a huge piece of the puzzle.
For additional insight into image recognition and analysis, watch our webinar with Crimson Hexagon SVP of Product Errol Apostolopoulos.