Despite some improvement, nuanced portrayals of identity are still hard to find.

11 verticals

978 ads

4,171 characters

Recently, the Geena Davis Institute on Gender in Media analyzed character depictions in the most‑watched ads on YouTube between 2015 and 2019.1


The data in this report was collected using a content analysis methodology, an approach that is ideal for systematically analyzing the content of communications. A team of 13 researchers carefully evaluated 978 advertising videos on YouTube. These ads represent the top 100 ads from 11 advertising verticals. (Some ads could not be coded because they featured a language other than English or did not include characters.) Videos were uploaded between Jan. 1, 2015, and March 31, 2019. View totals were only considered through May 31, 2019. Ads were randomly assigned to each researcher and hand-coded on the basis of character. Prior to initiating the work, the research team engaged in codebook development, training, and tests for reliability. Inter-rater reliability was achieved in terms of both absolute agreement and Cohen’s Kappa measures.

Race and ethnicity were determined from skin color, facial features, and context (such as the race of a character’s family). In this article, we are using the terms “person of color” or “BIPOC” to refer to any character who was not identified as white.

Researchers were asked for their best assessment of a character’s gender. Sexuality was determined by characters’ apparent enduring attraction (emotional, sexual, or romantic) to men, women, or both sexes, and coded “heterosexual” unless cued otherwise.

The study identified physical disabilities as impairments in body function, cognitive disabilities as impairments in brain function, communication disabilities as impairments in the ability to communicate, and major mental illness as indicated by obvious evidence (such as a diagnosis) or less formal cues (for example, hearing voices).

Representation in ads watched by U.S. audiences was determined by considering characters in the 439 videos that had majority-U.S. views.

Although race and gender are social constructs, racism and gender discrimination exist. The criteria used in the study were necessary for measuring disparities in representation across ads, and to reflect the scale of the problem in advertising media.

Thanks to the work of activists in and outside the industry, advertising has made strides in representation.

Nearly 40% of characters depicted were characters of color.

At a glance, diversity in advertising seems to be on the rise.

BIPOC characters were nearly 3X as likely as white characters to be shown in leadership positions.

Though the real world has yet to catch up, new data suggests that creative teams are casting with diversity in mind.

But overall, an aggregate increase in diversity is no guarantee of inclusion across race, gender, or other aspects of identity.

Despite making up over 18% of the U.S. population, just 6% of characters most seen by U.S. audiences were Latinx.

In a subset of ads aimed at Americans, about 40% of whom are BIPOC, only 31% of characters were people of color. Latinx characters were particularly underrepresented.

Globally, 35% of BIPOC characters were women, compared with 43% of white characters.

While 74% of ads featured at least one woman, 61% were white. Once cast, women of color face an institutional lack of access to makeup artists, stylists, and photographers. Eurocentric norms also affect women outside the industry: Black women are 80% more likely to change their hair due to biased expectations at work.

2 in 10 Middle Eastern characters were women.

Misogyny compounds the exclusionary effects of racism. And because few population surveys acknowledge Middle Eastern heritage, ads and other media rarely reflect the millions of Middle Eastern women around the world.

Latinx characters were nearly 3X as likely to appear partially nude, and 2X as likely to appear in revealing clothing.

Sexism and racism often converge. The study found that many ads reinforce sexist stereotypes that hypersexualize and objectify Latinas: 88% of Latinx characters shown in revealing clothes were women.

Asian characters were about half as likely to be depicted driving.

Repeatedly and thoroughly debunked stereotypes about Asian drivers still cause harm. Despite appearing most often in automotive ads, Asian characters were far less likely to be shown in cars. When they were, less than half were women.

About 2% of ads depicted at least one LGBTQ+ character.

Intersectional representation includes not only race and sex, but ability, body type, sexual orientation, gender identity, age, and other characteristics.

Almost all characters with disabilities were depicted as white.

Depictions of the disability community remain rare, with only 2% of ads featuring someone with a disability. Those limited portrayals are overwhelmingly white.

Across business verticals, advertisers missed opportunities to challenge and subvert stereotypes.

Black characters were almost 2X as likely to be depicted in comedic roles.

Due to a lack of representation behind the scenes, unchecked bias can perpetuate toxic narratives — compounding the discrimination Black communities face in the real world.

Compared with other groups, Latinx characters were overrepresented at sporting events.

Reducing communities to one-dimensional stereotypes robs them of their complexity, upholding myths that can shape their lives in harmful ways.

White characters were nearly 2X as likely to be portrayed as especially intelligent.

Homogenous creative teams often spread a view of the world that centers around themselves. White characters were about 4% more likely to be shown outdoors, and represented 69% of characters in Media and Entertainment, the least diverse vertical.

Less than 1% of characters were Indigenous. Only 4% had speaking roles.

Native American, Hawaiian, Alaskan, and Pacific Islander characters rarely got to speak. When people are seen but not heard, it suggests they’re being tokenized. Portrayals that deny people their agency have been shown to cause psychological harm.

Does this ad elevate the stories and reflect the diversity of traditionally underrepresented groups?

Does this portrayal challenge tropes and stereotypes?

Are the characters depicted with agency?

Does my team reflect the communities whose stories we want to tell? Are we collaborating with people from those communities?

Ask these questions to help your team evaluate creative decisions at every stage of the process.

The images we consume shape how we see the world. They have the power to challenge toxic narratives or reinforce them, and to lift the burden of a lie or double its weight. By hiring diverse teams, brands and agencies can draw on a wider range of knowledge to tell inclusive stories that represent real audiences and reflect the world as it is.

By Marianna Nash

Editor, Think with Google


Designer: Kate McManus; Product Lead: Casey Fictum; Production Lead: Jenny Maughan.


Geena Davis Institute on Gender in Media, AR, AU, BR, CA, EG, FR, DE, GB, IN, ID, IE, IT, IL, JP, MY, MX, NL, PK, PH, PL, RU, SA, ZA, ES, TH, TR, UA, AE, U.S., VN, An Intersectional Analysis of Race Representation in Advertising, Aug. 28, 2020.