While brands have been called out for a lack of diversity in their marketing, it can be a difficult issue to quantify. Longtime marketer and former Google marketing strategist Asha Shivaji started SeeMe Index to come up with that data, with some help from AI. I talked to Shivaji, the cofounder and CEO, about how marketing can be more inclusive and why it’s important.
This conversation has been edited for length, continuity and clarity. It was excerpted in the Forbes CMO newsletter.
How much of a problem is the lack of inclusive marketing today?
Shivaji: There was a KPMG study a few years back that showed the number one reason brands weren’t taking action to be more inclusive was a lack of understanding how to move forward. I think what that exemplifies in the market is that a lot of people may have good intentions to be inclusive, but don’t exactly know where to start or how to go about doing that. That’s really where our company came to be.
We are helping brands grow through inclusivity, using responsible AI and measuring everything they put in the world. In measuring everything they put in the world, we are measuring their ads, their product, their website, their external commitments across six identity dimensions: gender expression, age, body size, visible disabilities, sexual orientation and skin tone. With that 360 view, what we often see is brands leaning into one identity dimension or one of those components [with] their ads or products. But very few are consistent across the board in being inclusive. That’s what we’re really trying to help them see and realize as a gap. I think when we are working with brands, they’re so appreciative of having someone hold up a mirror to help them understand how a consumer’s really experiencing them.
How does SeeMe Index work?
We use two types of AI at a high level. One is computer vision and the other is knowledge extraction.
For the computer vision, we look at publicly available ads. We’re looking at TV spots, YouTube, TikTok, Meta, Instagram. Across all of those platforms, we’re looking for presence and screen time from a brand’s assets for different identity dimensions. It’s not just who’s in the ad, but how long they’re in the ad, and what intersectionality of identity dimensions we see.
For the knowledge extraction part, we’re looking across all the brand’s publicly available assets—their website, their product lines, commitments they make in the press—for mentions of different identity dimensions. Let’s say a brand stands for food insecurity. We know food insecurity affects different groups more than others—the LGBTQ+ community, Black and brown people. We’re always looking: Does the brand acknowledge that this platform that they stand for has identity significance as well?
There have been a lot of well-publicized issues out there with AI having issues with race and identity and getting things wrong. How do you make sure that your AI gets it right?
We position ourselves and truly want to live the belief of being a responsible AI company. To us, that means it’s not just what you measure, but how you measure it. We’re hyperconscious that any bias built into the model would be bias coming out. When we talk about skin tone, we use the Monk Skin Tone Scale that has 10 shades, versus the Fitzpatrick Scale, built by a dermatologist for the beauty industry many years ago [with] six shades: four would traditionally fall into light skin, one for brown skin and one for Black skin. With that kind of scale, you’re grouping a large number of people with different skin tones into just two shades. The Monk Skin Tone Scale also has four light skin tones, but broader tones for brown and Black skin.
Similarly, when we are thinking about gender, we specifically say gender expressions. Since gender is an internally held trait, we program the tool to look at not just male gender expression and female gender expression, but also gender nonconforming.
We also work really closely with experts and advocacy groups to bring in the latest thinking. We’re constantly tweaking the model, and literally everything we do has human oversight. It’s such a new space and a sensitive space, we feel it’s so important that we have human eyes on everything as well.
What kinds of things have you found and helped different brands act on?
We not only work with brands, but we also publish public indices. Last year, we published our first index on the beauty industry. This year, we’re launching our index for OTC drugs, which we’re excited about. We thought that by focusing on certain industries, we can help raise the bar for everyone.
When we looked at the beauty industry, there were some fascinating findings. One thing we found was in beauty ads, for each shade deeper you get on the Monk Skin Tone Scale, you lose a second of screen time—meaning people with the lightest skin tones are in ads for about 12 to 13 seconds, and the deepest only three seconds. It makes you realize that while we like to believe this is some unconscious bias coming into the editing room or the casting room, it’s happening. Once you have this sort of insight and data, people can actually take action. We’ve shared this with a lot of beauty brands, and I think jaws dropped every single time.
Another thing we found was across all of the beauty ads, only 2.9% of people had large bodies. When you start layering things on, a large body size, a deep skin tone, 55+, she’s absent. There is no woman who really looks like that in beauty ads. This is something that’s been spoken about in the advertising industry often, but once you have the data and you [have] looked at hundreds and thousands of seconds of ads and there is nobody who shows up like this, it gets very hard to argue that you’re being inclusive for these different groups.
I was struck that a lot of the brands seem to be doing really great with external DEI commitments to DEI and a diverse-targeted product lineup, but not on the ads that they put out. Why do you think it is so difficult to make the visual part inclusive while brands are doing other more complicated things that might take more company resources?
I think the challenge is few organizations have truly shifted to making this an organizational approach. We have different teams that are leaning into inclusion, but not necessarily in the same way. An example we saw, the brand had a beautiful spot with somebody disabled in it. Their website had no accessibility features. You see the gaps in an organization when one team wants to celebrate and show this consumer that they’re part of the brand, and then another part of the brand just isn’t firing on that same data.
What we try and do when we’re bringing this 360 data forward is point out those gaps for brands. You might have a great shade spectrum, but why isn’t she showing up in your ads, or why isn’t your external commitment mentioning that group of consumers, either? I think that’s helped brands understand a consumer’s not just seeing what’s on shelf. We’re also seeing what is on your YouTube channel or on your TikTok, and they’re also reading about what you’re doing in the press. All of this needs to work together, or it just screams that it’s inauthentic.
The term “DEI” has recently been losing its prominence among companies and in boardrooms. What about the way that companies are actually behaving, in terms of internal commitments, visual commitments and marketing?
Whether or not brands want to talk about DEI, they’re leaning into inclusivity because it’s good for business. A couple weeks ago we launched a paper with Circana, [which] has retail data. We compared it to our beauty index to see if there was a relationship. We saw that the most inclusive brands are growing 1.5 times faster than their competitors and the rest of the industry. Similarly, Unstereotype Alliance with Oxford launched a paper that saw more inclusive ads drive 3% higher short-term sales and 16% higher long-term sales.
Savvy brands are so data-driven today, I believe they truly know this. They just may not want to tell a story in the marketplace about their efforts leading in DEI. But I think inclusivity today is really just another way to describe personalization. As a marketer, we’ve always wanted to connect a consumer and their needs and their passions and what they believe with our brands, and inclusivity is just another lens of doing that. I think people have been telling us for a long time that they want brands to share their social values, and they want brands to stand for something. The way they spend reflects that, as well.
While this has been an issue for a long time, there aren’t too many people who have looked at it in this way. Why are you using AI here, and how did AI get to be one of the first ways to find a solution?
For us, it’s really the scale that AI allows us to analyze data. My cofounder and I met at Google and we were in roles where we were working with Google’s 30 biggest advertising clients. We kept getting questions from clients about how do I know if I’m doing my DEI efforts correctly? Am I winning with this group? Am I earning their hearts, earning their dollars? We didn’t have a real data-driven way to answer that, which goes back to that original problem. I think in the DEI and inclusivity space, nobody’s had a real data-driven approach.
That’s why we thought an index was really important, because understanding how you stack up to competitors is critically important in this space. What we see is that industries overall have gaps, but you also have a gap versus your peer set. There’s a long-term goal, there’s some closer end goals. That’s why AI was really a breakthrough to us, because we never would’ve been able to watch the number of ads we watched, and measure every single person in those ads, and how long they’re in those ads, and find the sort of insights that we’re finding. Or similarly, just looking at the breadth of data across all the brands, whether it’s their shelves or their websites or the commitments and what they’re saying in the press, the technology has helped us uncover insights that I don’t think would’ve been possible.
In your opinion, what do you see as a perfectly inclusive messaging campaign?
I think a brand should choose different groups that they want to connect with and earn their dollars, earn their loyalty, and think about how to really cater to their interests. In catering to those interests, it should show up in everything that they do and put out in the world.
As a consumer, you would want to see yourself reflected in the advertising. You’d want to see a product that meets your needs. You’d want to see external commitments that speak to causes that you care about. And my favorite is when a brand really extends themselves to allow a consumer to get involved. With more sophisticated external commitments these days, we’re seeing brands allow consumers to take trainings, donate themselves, volunteer. I think that’s the ultimate badging. When you really care about a brand, and you care about the causes they care about, it’s another avenue for you to express yourself. I think the best brands are doing that.
Is there a perfect mix of what would be presented in visual campaigns, or does that depend?
It absolutely depends on who that product’s for and who that brand stands for as well. One of the most painful executions you see when brands have the right intentions is just putting a lot of people in an ad, and the rest of the brand doesn’t reflect that. It’s like a one-hit moment where they decide to lean into a certain group, and there isn’t the follow-through.
When we’re working with brands, we help them understand where they’re stacking up against different identity dimensions. What’s interesting is sometimes they are surprised to learn they have more equity with a certain group because of actions that they’re taking, where they hadn’t even connected the dots themselves. It’s a great moment for them to lean in more, but oftentimes they’ll see they’re completely neglecting an identity dimension that matters a lot to them. It’s a great call to action for, how do we bring this all together?
I think maybe it’s having a partner show up with the data, that we can get everyone in the room together and be this neutral voice, showing them this is how you look in the marketplace. It’s not something you can argue with. Your competitor has more people of a certain identity dimension in their ads, or is creating more shades or marketing their products in a certain way. It’s inarguable when you look at the data.
Where should brands that are concerned about this start?
What we encourage brands to do is measure first. In marketing, there’s always been so many sayings, like “You treasure what you measure.” “What’s measured gets managed.” I think it’s true in this case. The more you measure it, the more you can improve.
The bar keeps moving, right? Not every brand can be Dove, who’s on a 20-year journey of inclusivity. Consumer expectations are always changing, and you want to make sure that you understand where you are sitting, compared to those expectations in your peer set. That can change a lot from month to month, year to year.
What kind of advice would you give to a brand that wants to start measuring? How should they go about doing it?
There’s some publicly available resources. I also consult with the Unstereotype Alliance with the UN, and there we have creative guidance called the three Ps. When you’re creating an ad, you look for the perspective, the personality and the person that you’re putting into each ad to make sure it’s unstereotyped. There’s also tools from different resources, like Kantar or the ANA, that allow you to measure your advertising as well.