Today is International Beauty Day, and it has us thinking about the social implications of beauty standards.
Industries like beauty, fashion, and personal care have historically lacked diversity and inclusion - but some brands are making a change, and consumers have taken note. Not only are consumers asking for more diversity from brands, new data shows that they are seeking out and supporting diverse influencers.
Diversity is clearly beneficial for all, so why is it so hard to achieve?
The underlying bias of social media algorithms
“Beauty is in the eye of the beholder.” This is an adage that was meant to express how empowering the subjectivity of beauty can be, but can this really hold true in our modern-day of social media algorithms?
According to some researchers, the answer is no.
Platforms like TikTok claim that they take a network approach when suggesting accounts to follow: “users who follow account A also follow account B, so if you follow A you are likely to also want to follow B." However, according to Marc Faddoul, a researcher at the University of California Berkeley School of Information who studies AI and disinformation, these types of algorithms run the risk of creating “coverage bias”.
"Collaborative filtering may also reproduce whatever bias there is in people’s behavior. People who tend to like blonde teens tend to like a whole lot of other blonde teens. In that sense, it's kind of expected." - Marc Faddoul
In other words, algorithms as they exist today, may actually limit the expression and discovery of diverse creators.
How data bias applies to influencer marketing
When it comes to brands collaborating with influencers, we assume that having a data-first approach helps mitigate bias.
However, there are qualitative aspects like “brand fit” and “having the right look” that are involved in the influencer vetting process that opens brands up to the risk of bias. And, having that narrow perception of qualitative elements might even be supported by data.
As data scientist Cathy O’Neil puts it, data can be used to perpetuate human bias.
“Consequently, racism is the most slovenly of predictive models. It is powered by haphazard data gathering and spurious correlations, reinforced by institutional inequities, and polluted by confirmation bias.” - Cathy O’Neil, Weapons of Math Destruction
Similar to Faddoul, O’Neil points out that if we only create predictions of success based on data of past success, opportunities will be more readily available to those who fit the model of that past success - mostly white, mostly male, mostly rich.
When you compound algorithmic bias with prejudiced predictions of success, it results in data that is inherently skewed.
Given that most industries are just beginning to address the problem of diversity, BIPOC influencers are seeing a new rise of opportunities. The challenge here is that the focus on diversity is so new that BIPOC influencers by-in-large do not have the same amount of historical performance data as their white counterparts. So if a brand, for example, uses past performance on sponsored posts as an indicator of what they should pay, BIPOC influencers may not have the same advantage as those with a “proven” history of high performing sponsored posts. This may lead to a brand working with that BIPOC influencers less, or even paying lower fees to BIPOC creators for the same deliverable.
How brands can make a difference
Navigating these issues are difficult, so here are two key actions you can incorporate into your influencer marketing strategies now.
Use data to identify trends, then engage with creativity not stereotypes.
Data is a powerful tool for surfacing trends, however how we talk about those trends matters. For example, when looking at data on the hair care category, the terms “color” and “blond” rank extremely high in frequency.
Rather than equating blond to beautiful (or only asking white influencers to collaborate on this trend), center the conversation around hair color techniques that allow people to express their individuality.
This tip in action: In 2020 “bayalage” was a coloring technique that was trending. A diverse set of influencers (both BIPOC and white) jumped on this trend, showcasing their beauty through hair creativity.
Break the data bias loop by investing in BIPOC influencers.
We already know that there is the potential for minority influencers to have mass appeal. At this moment what holds this population back is a lack of resources, historical performance data, and access. Brands that make an investment in helping under-represented creators thrive won’t just be making a positive impact on critical social issues, they’ll help their own business goals.
“76% of Gen Zers feel diversity and inclusion are important topics for brands to address.”
Marketers at all levels can use their expertise to easily help BIPOC influencers use data to their advantage. When collaborating with BIPOC influencers, help them understand and navigate platform strategy, performance optimization, and algorithm best practices. Two ways to achieve this:
- Create a campaign like Degree’s #BreakingLimits initiative. Degree didn’t just hire a diverse panel of influencers for this campaign, it committed to mentoring these influencers and investing in programs “in their own community that allow them to inspire others”.
- Discover trending influencers like Lizeth Ramirez or Deborah Bland whose audience is rapidly growing across platforms, and help them monetize their popularity.
It’s in the data: Data shows that minority influencers create big impact. In a panel of 150 minority-reaching influencers (those who over-index in Black, Hispanic, or Asian audiences), we found that since the resurgence of the Black Live Matter movement (from June 2020 to June 2021), the average Brand Vitality (VIT)* Score per Influencer is 2k on beauty content. This is significantly higher than the general population of top tier beauty influencers (those who have a following of 1M - 5M), who average about 600 VIT per influencer.
*The Brand Vitality Score (VIT) is the first metric uniquely created for measuring a brand’s performance in influencers’ content. It was designed to measure what matters—visibility (reach of content), impact (engagement generated) and brand trust (quality of content on brand image).