A colleague of mine named Diya was being interviewed about a network she had founded that supports women of color.¹ “Why is it important for women of color to have a community just for themselves?” the interviewer asked. “Are your experiences that different?”

In response, she shared an incident that had happened while picking up her mother from the airport. Diya had met her mother at baggage claim to help with her luggage. As they were crossing the garage to Diya's car, a white woman tapped her on the shoulder, stopped them, and asked what their daily rate was. Diya and her mom stared back, not sure what she was asking.

“For cleaning. Do you do house cleaning?”

Diya asked the woman what made her think they cleaned houses. The woman stammered and walked away.

It wasn't the first time this had happened to her.

Diya doesn't experience life only as a woman or only as an Indian. She lives at the intersection of those two identities. In addition to facing the barriers of her gender and the barriers of her race, she faces barriers that only surface when these two intersect. As a white woman, I will never be mistaken for a house cleaner. I'm fairly sure Indian men are not often mistaken for house cleaners, either.

Frameless

Intersectionality has made its linguistic way out of academia and into the mainstream. When Columbia Law School Professor Kimberlé Crenshaw coined the term 30 years ago, it was within an obscure legal context. Now, her TED Talk on intersectionality has over four million views,² and Hollywood stars are dropping the word in red carpet interviews.

As noted in chapter 4, understanding language around DEI can be overwhelming. This book is designed to provide a guide for anyone, regardless of their level of DEI knowledge, but “intersectionality” is a concept that is important to understand before we start adjusting cultural levers in part 2.

Over the decades, the definition of intersectionality has evolved. In broad terms, it describes how class, race, age, gender, sexual orientation, and other aspects of identity “intersect” with one another or overlap. In her TED Talk, Kimberlé Crenshaw lays out a more impact-focused definition: “If you're standing in the path of multiple forms of exclusion, you're likely to get hit by both.”

To illustrate this point, Crenshaw presents us with the case of Emma DeGraffenreid, an African American woman who had applied for a job with an auto manufacturer but was not hired. She believed her application was rejected because she was a Black woman. A judge had dismissed her claim of race and gender discrimination based on the employer's argument that they did hire African Americans, and they did hire women.

The judge was not willing to acknowledge the nuanced but pivotal points of Emma's argument: The African Americans who were hired, usually for industrial and maintenance jobs, were all men. The women who were hired, usually for secretarial or front-office work, were all white. The overlap of these unwritten rules meant that Emma was faced with twice the discrimination and none of the opportunities. She could not fit the role typically given to African Americans because of her gender, and she could not fit the role typically given to women because of her race.

The antidiscrimination policies that did exist at the time would not protect African American women, because their experiences were not the same as those of white women or African American men. There was no frame of protection for someone like Emma.

Most workplaces continue to have no frame for including women of color and other individuals living at intersections, including introverts like Trevor, whom we met in chapter 4. Unless these individuals are intentionally included, they will be unintentionally ignored.

Minorities in the Margins

On March 24, 2021, my inbox filled with emails announcing that it was, once again, Women's Equal Pay Day. I receive several versions of this email every year, marking the date through which women had to work to make as much as men in the previous year. This subject line, though, wasn't quite accurate.

Asian American and Pacific Islander Women's Equal Pay Day would have been March 9; African American Women's Equal Pay Day would have been August 3; Native American Women's Equal Pay Day would have been September 8; and Latina Equal Pay Day would have been October 2.

Overall, women employed full-time, year-round in the US are paid 82 cents for every dollar paid to white, non-Hispanic men.³ Black women? Sixty-three cents for every dollar. (Chapter 10 includes a more in-depth look at what these numbers mean, and at different definitions of the “pay gap.”) We're able to track the gaps between these specific groups because organizations like the Economic Policy Institute look at the intersection of race and gender. Most employers, though, do not.

When I work with employers to conduct a pay gap audit, their compensation specialists usually come to me with their pay data aggregated in two separate categories: by gender and by race. They're siloed. This means that when I'm looking for gaps, all I can see is the difference between what women make and men make, and the difference between what white employees make and employees of color make. I have no way to compare, from looking at the data, the earnings of women of color to those of white women.

A company I recently worked with proudly turned over their compensation data with an analysis concluding that they had reached pay equity. With proper controls applied, accounting for variables such as education, job difficulty, and tenure, they stated that they were compensating men and women, and white employees and employees of color, with similar pay for similar work. We took the individual employee data, did our own regression analyses, and as long as the categories stayed separate, we came to the same conclusion. It was only when we looked at the intersections that the gaps appeared. Eight out of the ten women of color were making at least 25 percent below what white women made in similar roles, and 17 percent below what men of color were making in similar roles.

Companies that ignore intersections in their data can go for years thinking that they have reached pay parity, when they have not. This hurts both employees and employers. As chapter 10 explores in depth, pay gaps that are not caught early widen over time and cost more to remediate when they are finally discovered.

Pay, though, is just one part of the workplace experience. Employees who experience multiple forms of exclusion are pushed farther into the margins over the course of their careers. Looking at GEN's survey data, Black women are the least likely to report that they have someone at work they can confide in, and most likely to report feeling isolated. Asian women are the most likely to report being asked frequently to help with office “housework” such as making coffee, party planning, or other tasks outside their core job functions. Men of color are most likely to report retaliation or stigmatization when requesting time off for mental health reasons.

These gaps are the symptoms of a DEI mindset that assumes inclusion will trickle down from the “majority minority” to those at the intersections. While inclusion doesn't trickle down, exclusion does.

The Costs of Trickle-Down Exclusion

For years, Google had attempted to address the outsized turnover of women in their workplace. They provided more training and mentorship opportunities, started affinity groups, and publicly recognized the senior Google women who had moved up the ladder, but their attrition numbers didn't budge. Finally, they took another look at their data and aggregated it by one critical factor: parenthood status. They realized they were not losing women—they were losing mothers. Specifically, they were losing new mothers. In response, they tried extending the amount of time off that new mothers could take under their paid leave policy. The rate at which new mothers left dropped by 50 percent, and they are now no more likely to leave than their peers.⁴ The cost of extending leave came nowhere close to the cost of replacing and retraining the talent they had been losing.

Intersectional aggregation doesn't just tell us who is falling through the cracks, it can often tell us why.

In the public survey that GEN uses to determine national benchmarks around employee experience, we ask respondents if they have had access to a range of experiences or opportunities in their workplace. For example: “Has your manager asked you about your career development goals in the last 12 months?”

If I aggregate the responses we've received to this question by race and gender, the results look like this:

This is how most companies sort their survey data (if they include race at all).

Nowhere in that aggregation can we see the following:

If this same data represented an employer's workforce, and they did not aggregate beyond race and gender, they would not know that their managers were less likely to ask mothers about their career goals than fathers, or that employees over the age of 55 were less likely to get asked as well. Not only would these employees be missing out on potentially pivotal conversations, but employers would be missing out on insights that could improve employee retention and engagement.

At this point you may be asking, how granular should we get? Is the “women of color” category specific enough, or should we subdivide the category by race? Should Asian include a separate category for Southeast Asian? Should age be divided into every ten years? Five years?

And what about employees like Trevor, the introvert? As a straight, white male who does not have any caretaking responsibilities and considers himself neurotypical, Trevor is probably not going to check any of the boxes that represent underestimated groups. How do we account for employees who are not going to show up in any demographic grouping, but have often-overlooked and valuable contributions to offer?

Is it really possible to capture everyone?

Beyond Thin-Slicing

Thin-slicing indefinitely is not the point. For many organizations, aggregating data beyond gender just isn't possible. Even for large organizations, at some point there is a limit.

If your organization is not large enough to aggregate your data by intersections of groups, or even by more than gender, there are still solutions, which stem from one underlying principle:

Majority rule is not always the best rule.

When I was working with a Seattle-based tech company, one of the directors came to me with a challenge. Every six months employees got to vote, floor-by-floor, on how a discretionary “Fun Fund” should be spent. The short survey that collected votes offered several Fun Fund options: an Xbox for the floor, lunch delivery for a week, an early end to the workday, a ping-pong table, etc. The director told me that every time, without fail, all the guys got what they wanted, usually something game-oriented: a gaming console, ping-pong table, or set of games.

Men made up about 70 percent of the employees on the floor, and the results of the Fun Fund survey cut almost exactly along gender lines. Women voted for lunch delivery or time off, but several attempted to be “good sports” and join the guys in their gaming. They were told they weren't welcome because they hadn't voted for the games in the first place, so the games didn't “belong” to them. “I'm literally having to kneel down next to these guys' desks and ask them to share their games with the girls,” complained the director. “As long as they're the majority here, nobody else is ever going to get what they want.”

When “majority rules” is the policy, a homogeneous majority that makes up 70 percent of the office doesn't get what they want 70 percent of the time—they'll get what they want 100 percent of the time. When this approach is applied to policies with a more significant impact than the Fun Fund, it creates a culture that over-indexes on what the majority wants. Employers who are trying to attract and retain underrepresented groups will not do so by building a workplace that caters to, and therefore attracts, one overrepresented group, again and again.

While there is no thin-slicing rule of thumb that can guarantee an inclusive workplace, there are guidelines that can integrate intersectionality into your systems. As you kick off implementing cultural levers in part 2, these overarching principles can sustain intersectionality when designing, implementing, and evaluating new workplace policies and mechanics.

Designing Policy: Gathering Intersectional Perspectives

Often, organizations collect input from employees before writing a new policy or procedure. In theory, this is a great practice. The key to addressing more than one group's needs is analyzing this feedback through an intersectional lens. If you're a large company, look at the intersection of groups in your data. What are women of color saying? What are men of color saying? What are mothers saying? People in different age groups? Those who are legacy employees versus those who are still ramping up? Your categories should be large enough, though, that no individual is identified by the group label; if you only have one woman of color in your organization, don't have a “women of color” group.

If your organization is too small to look at these intersections, the following guidelines can help to ensure that all voices are accounted for.

Rotate into 2^nd and 3^rd choices. If a policy or opportunity you're creating receives input on a recurring basis, like the Fun Fund example, rotate which choice gets to “win” each time. Either let second place be the winner on an alternating basis, or, after the first round of voting, send the survey only to those who did not win the first time.

Weight by relevance. If employees are providing input on policy decisions, weight the answers of those who will actually be impacted by the policy. One employer I worked with was trying to determine if they should implement a keep-in-touch policy for employees who went on parental leave, to help them still feel included. Over half of the employees who provided input on this policy were not even considering using parental leave in the future, but their input overshadowed those who would have been affected. The employer was going to take all feedback into account equally, but weighting the responses of the relevant parties led to a policy that worked better for those who took leave, without disrupting their peers.

Provide space for open-ended responses. Trevor the introvert's way of thinking through problems and presenting feedback runs counter to his workplace culture. He likes to consider all possibilities quietly before weighing in, and he often highlights the work of others ahead of his own. When given opportunities that honor how he performs, though, Trevor—and his employer—can flourish.

Employee feedback surveys tend to be multiple choice. While this may be efficient, it also limits the answer choices to the realm of possibilities in the survey creator's mind. Ideas like Trevor's may not have ever gotten enough attention to make it into the options that appear on an employee survey. The creator of the survey simply may not have heard from those who think differently, are introverted, or come from underestimated backgrounds. Open-ended questions provide an opportunity for these groups to give feedback, in their own words, that would not have been represented in the answer choices.

Open-ended questions also allow people to provide input that they wouldn't feel comfortable volunteering otherwise. Veterans of color, for example, report feeling less comfortable speaking up about barriers they face—especially around race—than white veterans or other employees of color. Dr. Lamise Shawahin, a former Veterans Affairs psychologist, explains that the military champions the ideal of colorblindness, which can erase the idea of race-based experiences by emphasizing sameness: “This idea—‘we're all green'—and these types of slogans, they instill the notion that racial differences are not important in the military.”⁵ This learned ethic of prioritizing the group's needs over one's own can persist when veterans return to an office setting. Open-ended questions offer an active invitation for feedback and the anonymity some may need to feel safe, without visibly “stepping out of line.”

Who's in the room? Finally, when collaborating on policy design, pause, look around, and ask, “Who's in the room?” Are those who are actually going to be directly impacted by this policy represented? Who's overrepresented? Similar to weighting survey answers from relevant parties, this step ensures that the people you're attempting to include have their voices represented. If you don't want to single out individuals to provide their feedback, allowing space for anonymous comments on a questionnaire can provide the opportunity to capture all voices in the metaphorical room.

Implementing Policy

Designing policy with an intersectional lens is the first step toward equal treatment. Even if employees may have equal opportunities as written in company policy, accessibility to them can still be unequal. The following guidelines ensure that policies are implemented so that all employees have equal access to resources and opportunities.

Model the behavior yourself. This is especially important for those in leadership positions. You may create policies that do not apply to you personally, but you can make it easier for those who would benefit from the policy to feel comfortable taking advantage of it.

Let's say your company has decided to embrace the use of pronouns in email signatures, but the choice to do so is optional. You may never have experienced someone misgendering you, so you may not bother putting pronouns in your email signature. The fact that you personally have nothing to gain from this policy means you can have even more of an impact by embracing it. This opportunity was created for those who have been misgendered or feel they easily could be. They are the ones who are going to benefit, but they are also the ones who must take on the risk. If only those who “need” this policy are the ones taking advantage of it, it makes them stand out. Being one of the few to embrace pronouns in their signatures can feel like waving a giant flag that says, “Look at this choice I'm making!” It draws almost as much attention as correcting someone who has misgendered them in conversation.

If, however, it is common practice, or at least not a glaring exception, for people to include their pronouns in their email signatures, the person who would benefit feels less like an obvious outlier. Adding the pronoun comes at little risk for most and lessens the risk for those living at intersections.

Ask, “What are the criteria? What is driving our decision-making?” If there's one policy principle I repeat most often, it would be, “When criteria aren't present, bias will fill the void.” This bias may not even have anything to do with an individual's demographic characteristics. Interviewers, for example, are more likely to hire someone they interview in the morning than in the afternoon (provided it's not right before lunch hour). This does not mean that your organization must ban afternoon interviews, but having criteria front and center makes it easier to assess candidates clearly, even when we're battling fatigue or undercover biases. Reviewing GEN's data, we find that organizations with clear criteria for evaluating candidates for promotions have greater percentages of leadership roles filled by women of color, parents, and others who live at intersections.

In the same way that the willpower hacks discussed in the previous chapter prepare the path of least resistance, it is easier to evaluate someone accurately with a checklist in hand. The harder we have to work, the more decision fatigue we take on, and the more vulnerable we become to defaulting to bias-ridden assessments. Preemptively reminding ourselves of what makes someone fit for a role lessens our chance of missing out on a good candidate.

Communicate “why” and communicate universally. Often I work with HR teams who have put admirable time and effort into crafting inclusive policies, but when I survey their employees, many of them do not know that the policies exist. What appear to be opportunity gaps are often communication gaps in disguise.

In one architecture company I surveyed, I noted that only half of their employees knew there was a nursing room located on the third floor. When I aggregated the data, I did not find that women knew and men didn't. Instead, certain business units knew, and others didn't. The difference? Business units with female managers knew there was a nursing room, while business units with male managers did not. For female managers, this policy may have been front-of-mind, while a nursing room may not have stuck in the minds of male managers.

Formalizing the ways that a policy is communicated can override the unconscious motivations that make some more likely to share out a policy than others.

In addition, make sure to state why you've created a policy. The benefits of some of the cultural levers we will recommend adjusting in part 2 are not always obvious. For example, employees may not understand why you're moving the gender checkbox to the end of the application. Communicating why fulfills two purposes: It increases buy-in by showing there is an evidence-backed reason behind the change, making it feel less arbitrary to employees. It also signals to underestimated employees that their employer is genuine in their commitment to inclusion. Broadcasting that you are taking an impact-driven approach to inclusion demonstrates that your strategy will lead to meaningful change.

Evaluating Policy Impact

You've designed your policies, you've implemented them, and time has passed. How do you know if they're working and what success looks like? These next steps can help you measure impact at the intersections.

Who's showing up? After policies are implemented, take note of who is or isn't taking advantage of them. If an opportunity, such as networking events with senior leaders, was meant for all employees, but you are seeing only homogeneous attendance, provide an opportunity for employees to state anonymously what would make them more likely to attend. One employer I worked with took this step and discovered that some employees were not attending networking events because there were no kosher or halal options. Once they communicated this need to caterers, those employees felt more welcome and attended events, where they found mentors!

Evaluate Indicators “Upstream.” Many organizations measure end results, or lag indicators, such as the percentage of people in leadership who are women or people of color. These indicators do not help to identify processes upstream that would lead to different outcomes. Measuring lead indicators—the processes that drive success on lag indicators—will drive inclusive success. An example of a lead indicator would be the survey question I shared earlier: “Has your manager asked you about your career development goals in the last 12 months?” If employers do not ask this question but look only at who has been promoted into leadership and see that women of color are not represented, they will not have any insight into the cause. They may misdiagnose the problem as a professional development issue, provide more training, and wonder why the gaps persist. Tracking indicators upstream lets you address the root causes of gaps, rather than just guessing at them.

(Re)gather intersectional perspectives. Finally, the rules about how you collect data, from the “Designing Policy” section of this chapter, also apply to how you evaluate policies. When gathering and analyzing feedback, make sure that you're looking at intersections as much as possible, and making use of open-ended questions when it's not.

≈

You know now why common approaches to DEI haven't worked, and why we need to change. You've gotten a preview of what it looks like to design equity in, and bias out, without missing individuals who live at the intersections.

It's time to start adjusting cultural levers.

It's time to hire the right person for the job.