Fenty Beauty is insanely successful and popular for many reasons. It’s founder and the face of Fenty is the singer Rihanna. The product is high quality and affordable. The cosmetic line is unique in its inclusiveness with 40+ shades for every product. The brand is widely known among the beauty industry as providing quality cosmetics for every shade and focusing their campaigns on the beauty of underrepresented women in culture. Their original social media campaign was one of the most successful to date and adopted the tag line, “beauty for all.” Like all other beauty brands, Fenty is heavily reliant on digital marketing, and predominantly their Instagram account.
With an estimated $570 million in annual revenue, just under 10 million followers on Instagram, and only 1,500 posts, Fenty Beauty is killing it in the digital marketing game. But when looking at a brand’s successes, it is also important to consider their failures, as well as the competitor's successes and failures. This is especially true with hyper competitive markets like cosmetics. We analyzed the visual data of the Instagram accounts of Fenty Beauty and five of their competitors. Using our unsupervised machine learning algorithm, we are able to see and analyze how data sets are constantly changing, creating micro and macro trends, and can predict future trends.
Unsupervised learning can be best described as a machine learning algorithm that will specifically look for patterns in datasets that may or may not have labeled responses. This will be commonly used to find patterns within large datasets, but those using unsupervised learning may not yet have a goal in mind with what to do with the results. By using unsupervised machine learning, data can be easily put into categories that will be useful for creating content.
Cortex’s enterprise SaaS platform utilizes unsupervised machine learning to analyze the performance of visual content. It can be used on social media platforms, such as Instagram, to point out several factors of engagement. This can include what colors are most prominent and what objects are found in a person’s Instagram photos. As this data is gathered, it is then sorted into clusters based on those characteristics.
In a recent update, Cortex machine learning platform released new ways to understand the data of your visuals. We can now see the performance of trends over time, and look at how they accelerating or falling off. Global brands like Fenty Beauty feel this problem of trying to make sense of trends. Brands will spend $1.1 - $1.4 Trillion on visual creative in 2020. This leads to huge risk and waste from product marketing, to brand, to performance and field marketing. And global brands are being forced to rebuild the infrastructure for visuals or risk being disrupted by laser-focused brands.
So we take this data from these brands, and figure out how, why, and what works for engagement in a highly competitive and trend-filled market. Fenty owns the trend of representing every kind of human with any shade of skin. We can see here, that when comparing Fenty to their competitors, they perform 40% better than the median of their competitors. Fenty owns the trend of "human faces". In this particular micro trend, Fenty in pre-Black Friday, Fenty focuses on the human side of the product.
But, how do we know? In the graph above, we can see how the the visuals are engaging the audience. Theres a big peak in October, when posts are pre-Black Friday. This cluster is posts of women with a focus on their face. This cluster also shows the prominent color pink. All of this data is analyzed and clustered for conclusions to be made about what content works.
Heres another popular trend... human face with human hair. The best photos for this cluster shows the micro-trend . These photos perform 15% better than the brands average performance. Here we can see how Glossier, Colour Pop Cosmetics, Milk Makeup, Huda Beauty, and Fenty Beauty shape up against each other. The images are with women, their hair, and face as the main component of these posts. Here's one last cool thing... Cortex can now show trends over time. Fenty Instagram posts of women peak in performance and engagement in May, October, and December. There could be a lot of aspects to this high performance. But the data shows how and why, which in turn, shows how and your brand is performing against others.
So we know where brands like Fenty are performing well and engaging their audience. Do you know why your visual content isn't performing as well as it could?