It's Almost 2023—Why Is Bias in Ad Tech Still an Issue?

Representation may have improved, but there is more work to do

Next week is Women’s Equality Day, which commemorates the adoption of the 19th Amendment and the beginning of women’s right to vote. As a Korean American woman, a mother and a leader in tech, building toward a more fair and equal future for all is not only important to me—it’s personal.

And one  initiative that’s particularly meaningful to me is the ongoing work to reduce the risk of bias in advertising.

Having worked in the media and marketing industry for most of my career, I understand the importance of this work. But I was surprised to discover a crucial nuance upon digging deeper into it.

For marketers, promoting diversity and reducing bias are goals that go hand in hand but are not one in the same. Even for the most well-intentioned marketers, the latter can be far more difficult to identify and address. But it’s no less important. 

Despite admirable progress in terms of representation—both in the people shown in ads as well as those creating and managing campaigns behind the scenes—the algorithms that direct the deployment of ads may potentially contain unwanted and unfair biases that could impact how the technology selects which audience gets which ads. It’s almost 2023. With accelerated advancements in technology, how is this still possible?

Proving the existence of bias in advertising

The moral and social imperatives of mitigating the risk of bias in advertising technology are more than enough reason on their own to take these measures. Advertising makes a major impact on our culture, consumer attitudes and access to economic opportunities.

But the implications of mitigating bias extend beyond social responsibility. Brands may also positively impact campaign performance by increasing ad distribution to extend audience reach and gain new insights.

IBM announced a research initiative 12 months ago to better understand the existence of bias in advertising technology and its impact on outcomes by working with teams at the Ad Council and Mindshare to study their campaigns.

In the latest work with Mindshare, researchers found that the tech used by a CPG brand was—unbeknownst to the company—directing online ads mostly to women because female buyers were more common. The team determined that fairness could be increased by simply including more male buyers.

As a result, the brand identified previously unknown male buyers and learned more about this new audience. This demonstrates that performance and fairness can coexist and be mutually beneficial as well, enabling a deeper understanding of previously underserved audiences that can inform future ad strategies.

If we can uncover this type of bias in a single campaign, improve fairness, and maintain or even increase performance, imagine the impact marketers could create with a groundswell across the entire industry.

Coming together to increase fairness in advertising

The advertising industry is in a period of reinvention as it moves away from cookies and traditional identifiers. The solutions and methodologies currently being developed can set the standard for years to come, which can create an immediate opportunity and obligation for all of us to account for bias.

IBM recently announced The Advertising Toolkit for AI Fairness 360, an open source collection of fairness tools, metrics and algorithms that any brand, marketer, agency and tech platform can use—at no charge. These tools help teams to identify unintended bias in campaigns or technologies, evolve strategies to mitigate these impacts and establish better industry standards.

In June, IBM was one of several companies that lead the launch of the
Advertising Fairness Pledge
to raise industry awareness and encourage everyone to commit to educating themselves on bias in advertising, exploring how it impacts campaigns and advocating for change. Organizations like Delta Air Lines, WPP, Kellogg’s, Dentsu International, The 4A’s, IAB and the Ad Council have taken the pledge and publicly committed to improving advertising fairness.

But this is just the beginning. The entire industry will need to come together with everyone playing an important role. Here’s our vision for a better future:

  • Brands will monitor whether their technologies for reaching audiences could be perpetuating bias. 
  • Agencies that create and execute brand campaigns will build, use and recommend products and processes that mitigate bias. 
  • Ad-tech providers will develop platforms and tools that do not perpetuate bias at scale, especially when used broadly across the industry.
  • Industry associations will set broad standards for the industry at large and craft initiatives that bring awareness to these issues.

A call to the industry

Today I invite you to take the Advertising Fairness Pledge and share it with your professional networks and internal teams. This pledge provides a simple way to publicly demonstrate your belief that marketers must address bias in advertising and show your commitment to building a better future.

For those who want to go a step further, start exploring the Toolkit. You can access it here at no charge along with additional guidance on how to get started and share results.

This issue is particularly close to my heart. I’m committed to doing all I can to promote fairness in advertising now and in the future. IBM is providing the industry with tools, thought leadership, and guidance—but we can’t do this alone.

It’s time to do this together.

Crystal Park (she/her) is the head of b-to-b marketing at IBM Watson Advertising and Weather. She leads the team that elevates stature and brand perception and drives in-market demand through product go-to-market, executive thought leadership, industry partnerships, events and content.