Diversity and inclusion are hot topics in tech, and a lot of research has shown it’s not only the right thing to do but also a good business decision.

A 2015 McKinsey report found that companies with ethnic and racial diversity in management were 35% more likely to have financial returns above their industry mean, and those with gender diversity were 15% more likely to have returns above the industry mean.

When companies lack diversity or don’t create inclusive environments, it often shows in their products.

They can miss big groups of their users like Apple did in their 2014 “comprehensive” health tracking app, which failed to track women's periods. Or they can shut users out of their services, like Domino’s pizza chain’s inaccessible website, which they got sued over. In some cases their products end up amplifying prejudiced stereotypes.

A good example is Snapchat’s “beauty” filter which applied white features (e.g., lighter skin, narrow nose, etc..) to people’s pictures, implying that white is the beauty standard.

And when it comes to artificial intelligence (AI) a whole set of issues is surfacing due to lack of diversity.

Take this sobering example from Compass, a case management and decision support tool used by U.S. courts to assess the likelihood of defendants becoming re-offenders.

The software estimates the possibilities based on offenders’ responses to 137 survey questions.

Image source: “Machine Bias” by ProPublica

A ProPublica investigation found that Black offenders were almost twice as likely as white offenders to be labeled as high-risk but not actually re-offend.

Compass also produced the opposite results with white offenders: They were more likely to be labeled as lower-risk compared to Black offenders, despite their criminal history displaying a higher probability to re-offend.

So what’s happening here?

Input is biased so results are biased. Diversity was not considered when designing the survey or during the development process, resulting in a racist system.

What Can We Do As Product Managers?

No one wants to create biased products or miss a big part of their audience. But that will very likely happen when we don’t consider diversity in our process. There are few things product managers can do to mitigate bias.

Image source: Aseel Hamarneh

First, start with yourself and acknowledge you’re biased

None of us sets off to discriminate, but we all grew up in a biased society, so whether we like it or not, some of that bias gets internalized.

This is referred to as unconscious bias, which forms involuntarily from our experiences and exposure to media, and in many cases, goes against the beliefs and values we hold consciously.

Because we are unaware of our unconscious beliefs, they sometimes end up driving our behaviors. Research has shown that most of our decisions regarding people are heavily influenced by our biases and that our assessments of others are not as objective as we think.

Think of five people you trust/admire most at work. On your list, add next to each person their gender, race, ethnic background, English as a first or additional language, rough age group, sexual orientation, and whether they are able-bodied or have a disability.

Examine your list. How diverse is it? If it’s not, why? How come the people you admire most are all similar? Could it be bias? (Exercise adapted from Circle of Trust exercise by Include-Empower.com)

One common reaction when we recognize our own bias is to judge ourselves. But that’s not helpful. It’s more productive to pause on judgement and reserve the energy to acknowledge bias and disrupt it.

Challenge your assumptions and ask questions like these: Would I have interrupted my colleague if they were a man? Would I have supported them in their promotion if they were a different race?

Again, don’t judge your answers. The point is to be mindful of those thoughts and not let them drive your actions. If you pretend they’re not there, you’re more likely to act on them.

Second, look at the teams around you

How diverse are they? It’s likely that they’re not. So next time you’re involved in hiring, make diversity a priority. Having a diverse team equals diverse opinions and more creative solutions. There is a lot of research on removing bias from the hiring process and improving diversity.

There are things you can do to write inclusive ads, reach out to diverse networks, and remove bias from interviews. Here is a good article from HBR to get you started.

Third, address bias in your product

1. Examine your default user(s): Your priorities might look different if you deliberately consider diverse users. If for example, Google maps explored the needs of their female users or those with limited mobility, then perhaps safe routes or ones with step-free access would’ve been addressed early.
2. Include different voices in the development process: Think about product decisions and how they’re made. List the voices represented in the room and make sure it’s a safe environment for everyone to contribute (aka create psychological safety). If you’re lacking diverse voices in the team, find alternate ways to include diversity, such as co-design sessions.
3. And when your team comes up with solutions, take the time to examine the unintended consequences of these ideas. Check if you are negatively impacting some groups and look into ways to mitigate the risk early. An awful example of these consequences comes from abusive partners who weaponized smart home devices and used them to monitor and intimidate their partners.

I put together this table to summarize the questions you can ask at different stages of the product development process:

Discussing diversity can be uncomfortable, as it challenges our perception of ourselves. But it’s part of our role as PMs to provide a safe space for our teams to discuss challenging topics and reflect on what they mean for our work.