How Ad Targeting Reinforces Bias

It’s a harrowing time.

First, we were hit with the COVID pandemic, and in recent weeks, protests and discussions about racial inequities in America have taken the main stage.

Businesses are questioning their operating practices (or sometimes being called out for them and canceled) and uncovering how unconscious biases might play a role in everything they do—from recruitment to product development to marketing. The world of digital media and advertising has long been fraught with its own set of diversity challenges, but maybe now is a time where actionable plans and strategies can be put in place to bring about actual change.

Earlier this week, in our Wrapper newsletter, I brought up the topic of ad targeting bias, nudging the industry to start thinking about a solution. On a visit to Twitter later that day, I discovered that I wasn’t alone.

Right in my feed, Gizmodo’s enterprise reporter, Shoshana Wodinsky asked the question: “are there any concrete digital ad rules wrt racially targeting/tracking a given user? outside of fb’s 2018 settlement, i’m not seeing…………….. much of anything.” To which she added, “genuinely asking here, bc after doing 20 minutes of digging into the ways data orgs infer (or explicitly pull) racial data from……………………. all of us, i’m kinda getting genuinely concerned.”

Her question prompted deep conversations about weblining and digital redlining, which relate to how personalization can lead to discrimination and bias in terms of exactly which advertising is presented to an individual based on demographics such as race, gender or location. It’s not something that’s illegal, but it’s closely related to the practice of redlining—explicitly denying services based on race or zone pricing—which most certainly is.

These are some instances where personalization can go all the way wrong.

AdMonsters friend, Aram Zucker-Scharff, Ad Engineering Director for RED at the Washington Post, wrote one of the most informed and thoughtful responses to Wodinsky’s question. It was so good, we asked him if we could post his Twitter thread in its entirety, here it is:

Hopefully, you made it all the way through Zucker-Scharff’s full thread. It’s quite the worthy read.

Wodinsky and I also had another conversation that sprang from her original question about racial ad targeting. It’s clear that targeting to the right person at the right time can lead to some terribly bad practices and although they’re not illegal, they can lead to some of the worst cases of injustice.

Here’s a couple of follow-up questions for the ad-tech community at large: Can we have targeted advertising that is also fair and unbiased? And will it require a set of regulations to get us there?