Step by step guide of increasing your conversion rates through diversity in ads

Dennis Ksendzov

Founder · August 31, 2026 · 8 min read

Pick many kinds of creators: For its fall 2022 Pumpkin Spice shake, Premier Protein picked 78 creators with different follower counts, body types and backgrounds. The Shorty Awards
Combine formats and messaging: Caraway Home combines different formats and messaging inside one test instead of splitting out each variation, keeping their creative testing diverse. Dara Denney, 0:24
Test many different formats: Happy Mammoth tests AI ads, grids, infographics and statics that look like videos to keep finding new creative ways in. Dara Denney, 1:45
Adapt viral content for ads: Oats Overnight's warehouse employee content is the D2C gold standard, and they now adapt viral internet content for paid ads, results unknown. Dara Denney, 10:30

The short answer

Ads that show a wider range of people sold more in the biggest study with real sales data: 3.5% more in the short term and 16% more in the long term, across 392 brands in 58 countries.
People respond more to someone who looks like them. Black British viewers scored two UK ads far higher than the general public did, and 67% of Black audiences in Nielsen's survey pay more attention to ads that reflect their culture.
No public ad library shows conversions. TikTok's Top Ads is the one place with click and conversion curves, and it does not record who is in the ad.
We read who is in the live Meta ads from MUD\WTR and AG1. 62% of the MUD\WTR ads with a person cast only people who read as White, against 53% for AG1, and neither library says which ads sold.
Brands that do it well: Aerie stopped retouching its photos and cast models of all sizes, and its own site says sales grew 20 percent in fiscal year 2015. Fenty Beauty launched with 40 shades of foundation and is reported to have sold $100 million in its first 40 days. AG1 casts wider than MUD\WTR.

Can you see conversions in the ad libraries?

No. Meta, Google, LinkedIn, Snap and Pinterest show the ad, who ran it and when. None of them show clicks or sales.

Source, side-by-side check of every ad library

*Meta Ad Library searched for Old Navy: each ad shows when it started running and on which apps, and nothing about how it did*

TikTok Top Ads is the exception

Each ad on TikTok Top Ads shows likes, a click-through ranking, a budget level, and curves for click-through, conversion rate, clicks and conversions across the video. It does not say who is cast in the ad, and it only holds ads the advertiser chose to show.

Source, TikTok Top Ads and one ad's analytics view

What we found in the MUD\WTR and AG1 ad libraries

We pulled the live Meta ads from both brands, 119 from MUD\WTR and 325 from AG1, and read who is on camera in each. 76 of the MUD\WTR ads show a person, and 184 of the AG1 ads do.

MUD\WTR, 76 ads with a person:

47 cast only people who read as White, 62%
28 have someone who reads as not White, 37%
19 have someone who reads as Hispanic or Latino
3 have someone who reads as Black
2 have someone who reads as Asian
2 have someone with deep skin
2 have someone who looks 50 or older

AG1, 184 ads with a person:

97 cast only people who read as White, 53%
73 have someone who reads as not White, 40%
39 have someone who reads as Hispanic or Latino
22 have someone who reads as Black
20 have someone who reads as Asian
14 have someone with deep skin
24 have someone who looks 50 or older

AG1 casts wider on every count. This is how people look on screen, not who they are.

The libraries cannot tell us which of these sold. The MUD\WTR ads all started between 6 August 2026 and 3 September 2026, so how long they have run says nothing yet.

AG1 ads go back to 12 February 2026, and how long they run does not split by casting. Ads with someone who reads as not White and ads that cast only people who read as White had both run a median of 12 days. 10 of the 11 AG1 ads that had run 90 days or more cast only people who read as White, and 5 of those 10 are the Hugh Jackman campaign.

Ads with a wider range of people sell more

Kantar and the Unstereotype Alliance: real sales, 392 brands

Kantar measured inclusive ads against sales data from Unilever, Diageo, Mars Incorporated, Mondelez International and Bayer Consumer Healthcare. Researchers from Oxford's Saïd Business School worked on it, and it covered 392 brands in 58 countries.

*How much better inclusive ads did, across 392 brands*

"The findings confirm that inclusive ad campaigns deliver: 3.5% higher shorter-term sales and 16% higher longer-term sales"

The same ads were 62% more likely to be someone's first choice and earned 15% more customer loyalty.

Source, Unstereotype Alliance, 16 September 2024

Facebook IQ and the Geena Davis Institute: Meta's own brand lift studies

Meta took 25 of its brand lift studies, run in 2018 and 2019, and ran 4,000 simulations. It also surveyed 1,200 people and looked across more than 1,200 brand lift studies.

*What Meta found when its ads showed more kinds of people*

"And in more than 90% of the simulations we ran, diverse representation was the winning strategy for ad recall lift"

In the survey, 71% expect brands to promote diversity and inclusion in their online ads, 59% said they are more loyal to brands that stand for it, and 54% do not feel fully culturally represented in online ads.

The simulations measure whether people remembered the ad, not whether they bought.

Source, Facebook IQ, March 2021

SeeHer and Ipsos: ads that show women well

Ads that showed women in a positive light scored in the top third of Ipsos's ad-testing database.

*What Ipsos predicts for ads that show women well*

"brands saw the likelihood of short-term sales +24% higher"

"the likelihood of stronger long-term brand relationships +28% higher"

Ipsos predicts this from its ad-testing database. It is a forecast, not counted sales.

Source, Ipsos, Strive for More

Old Navy BODEQUALITY: every body size in the ads

*Old Navy's BODEQUALITY campaign, from PMG's write-up of the campaign*

"the campaign also drove 117% higher customer acquisition, with 3X higher conversion"

According to the company, BODEQUALITY doubled the number of extended-size customers in the quarter, and 15% of them were new to Old Navy. PMG also reports net sentiment moving from 68% before the campaign to 90%.

Old Navy changed its size range and its ads at the same time, so this does not separate the casting from the clothes. The number comes from PMG, the agency that ran it.

Source, PMG's write-up of the campaign

Brands that do this well

Aerie #AerieREAL: unretouched photos, models of all sizes

*Aerie's #AerieREAL campaign, from American Eagle Outfitters' own site*

Aerie stopped retouching its photos and put models of all sizes in its ads.

"After launching #AerieReal, sales grew 20 percent in fiscal year 2015"

This comes from the company's own site, looking back years later, and it does not separate the ads from anything else Aerie changed that year.

Source, American Eagle Outfitters, 22 March 2021

Fenty Beauty: 40 shades of foundation at launch

*Fenty Beauty's launch campaign, from a write-up of how the brand was built*

Fenty Beauty launched with 40 shades of foundation and cast its campaign across the whole range.

"$100 million in sales in its first 40 days of operation"

This figure is reported in the press. Fenty has not published it, and it covers a Rihanna launch as a whole, not the casting on its own.

Source, write-up of how Fenty Beauty was built, June 2019

AG1, Old Navy and IKEA

In our own read of the live Meta ads, AG1 casts wider than MUD\WTR on every count: 22 of its 184 ads with a person have someone who reads as Black, against 3 of 76. Old Navy's BODEQUALITY campaign is above, and IKEA's ad is in the System1 test below.

People respond to people who look like them

System1: the same ad, shown to the group it features

*Frames from the ads System1 tested for its Feeling Seen study*

System1, ITV and DECA tested UK ads that feature groups often left out of advertising, once with the general public and once with the group each ad shows.

"When diverse groups see people like themselves on screen, it has the potential to create a big uplift in their emotional response"

*Star ratings from the general public and from Black British viewers*

Nationwide's ad was a 2-Star ad with the general population and 4.7 with Black British viewers. IKEA's ad scored just under 3 stars with the general population and 5.2 with Black British viewers. System1's 1-5 star rating predicts long-term brand growth.

Source, System1, with ITV and DECA

Kantar: one Bounty ad, cast three ways

Bounty made the same ad with an Asian family, a Black American family and a white family. All three versions were seen as impactful, at the 70th percentile or higher.

"The Asian and Black American family ads were considered 'modern and progressive' for society (Diversity Index of 60-70)"

That was not the case for the white family's version. On the same page, Kantar read an Instacart ad featuring an interracial couple:

"The ad was impactful (70+ percentile), emotionally engaging (80+ percentile), and the most enjoyable ads across all demographics"

These are Kantar's scores for how people felt about the ads, not sales.

Source, Kantar

Nielsen: what Black audiences say

*Nielsen's chart of who wishes they saw more of their own identity group in what they watch*

"the majority of Black audiences (67%) surveyed pay more attention to ads in media that reflect their culture"

For creators, 52% of Black audiences said they are more likely to buy when a brand partners with creators connected to their fandoms and interests, against 45% of everyone.

This is what people say, not what they bought.

Source, Nielsen, January 2026

What shoppers say

Getty Images: "63% of people prefer to buy brands that are founded by or represent people like themselves". Nearly 80% said it is not enough just to have people of different ethnicities, backgrounds and appearances in the ads. Source, Getty Images and YouGov
Kantar surveyed more than 23,000 people in 18 countries: "75% of consumers say that a brands diversity and inclusion reputation influences their purchase decisions". Source, Kantar Brand Inclusion Index 2024