D + E + I Done Right

Introduction

I thought to myself, “Hey you know what you should do? You should write a post that everyone will hate.”

So, understatement of the year, DEI is a bit of a hot topic right now. I think that the current public discourse is harming DEI, but the reason that the current public discourse is as it is, is because most DEI implementations are done very, very poorly pretty much everywhere.

How am I doing so far?

Many of you will want to quickly dismiss this post because it’s written by the Mayor of the Majority Group (me). I’m going to ask you to evaluate the ideas and content, and not their source.

So what purpose does this post serve? I want to offer a road map of sorts to doing D+E+I impactfully. Imagine D+E+I done in a way in which underrepresented folks are able to see measurable impact, while majority group folks see what’s happening and actually kinda think it makes sense. Imagine no shit improving a company while improving your representation (D). Imagine (measurably) no gender pay gap, (measurably) equitable promotion rates, and (measurably) fair performance reviews (E), and imagine measurably improving Inclusion (I), all with common sense business reasons to go alongside our justice reasons.

We might say that our working vision is: “Representation (D) that matches the world around us, inclusion (I) that far exceeds it, alongside equitable outcomes and opportunities (E) for every last employee”

The Metaphor

Many of us have had Neapolitan ice cream — three separate flavors, chocolate, vanilla, strawberry, all in one container. Almost none of us has had a Neapolitan milkshake. Why?

Because mixing the flavors is not really the point.

D+E+I has three distinct components (That’s why I write it this way, using the plus signs). Each must be defined carefully and measured differently. Like the ice cream, three distinct flavors, but all in the same container.

Typically, companies, pundits, and DEI practitioners alike lazily synonymize and mix the terms, which practically means they neither carefully define nor carefully measure each of D+E+I. This is probably the biggest foundational problem with D+E+I work right now.

The outcome is high visibility activity that leads to nowhere special. The D+E+I practitioner is frustrated by the lack of support in the company, but doesn’t realize that without careful definition and measurement it’s difficult for the typical manager out in the company to latch on, and folks out in the company see what my friend and colleague Sarah Marrs sees. She wrote this as a comment on LinkedIn when I was teasing this post. I said “I’m writing a post that everyone will hate. Anyone want to guess the topic?”

Sarah is a thoughtful supporter of D+E+I, and yet this is what she sees, and so do I. And let’s be honest, most people see the same thing, but are simply unwilling to speak up, because there are massive repercussions to doing so.

I have another colleague — a proud member of multiple distinct protected classes, and a very strong analytical bent. I was talking to her about new roles, and suggested she try to include D+E+I as a part of a new gig; I knew she could do the work because I’d once seen her do it. What she said to me was surprising: “I don’t want it.” The reason? While playing a major role in making measurable (not anecdotal) progress along each of D+E+I, her colleagues vocally punished her for not doing enough, or simply because of their own ignorance about what was happening or what should be happening. Thankless and painful. Nothing is ever good enough.

We shouldn’t punish people for trying to make this better and for offering, gulp, diverse ideas on the topic.

The Roadmap: D+E+I Defined & Measured

Diversity

Diversity = Representation.

Focusing on diversity is a way to potentially find otherwise undiscovered talent. I thought Mark Cuban’s first point here hit the nail on the head.

“Best possible” vs. “First possible”

A common clap-back against “diversity” is “I believe we should just hire the best possible person for the job.” Before I tell you how disingenuous this is, please raise your hand if you’ve ever said this. I said this years ago, before I sat down and really thought about it.

Here’s the insight: Almost any reasonable accounting within a company will reveal that the company does not hire “best possible,” but it instead hires “first possible.”

There are times when speed is, in fact, the most important variable. Just please understand that speed, or “first possible” often means

1. unconsciously hiring people that look like the incumbents and
2. not necessarily hiring the “best possible” people

If a company wants to say “we’re moving fast and need to hire the first qualified person,” I’m ok with a company making that specific decision, but let’s be really clear, that you are almost certainly not hiring the best possible people. I’m asking you to be honest about that. Let me give you a feel for what I mean through example:

Two examples from a B2B SAAS company (all approximate numbers):

1. This company studied qualified black college graduates as candidates for its lower level US-based roles and compared that proportion to its actual proportion in these roles. Qualified in this case was defined roughly as STEM majors from top tier universities. Let’s say that about 10% of the total qualified college graduates in the US were black grads. Inside the company, however, black employees made up fewer than 2% of the population for those roles. (The numbers are somewhat imprecise, but the orders of magnitude are accurate). This is not indicative that this company and all its “Get the best possible!!” hiring managers had scoured the earth to find the “best possible” people; interestingly, the company was hitting its overall hiring targets quite nicely. But they were hiring the “first possible” people.
2. Same company: women represented, say, 40% of the total employee base (fascinating this was relative to a user base that was over 50% women). However, when they evaluated how many women were at the management level, this number fell to 20%, and at the executive level to 10%. (Numbers again imprecise, but I promise, directionally right.) This, again, does not seem to indicate systematically finding “the best” person because if that were the case, you might expect those numbers to be at least closer to equivalence because men are not inherently better managers than women.

These are practical outcomes of unconscious “first possible” hiring, which prioritizes speed. Importantly, in the two cases above, the people who were hired were also qualified. These examples above suggest, however, that there are cases where the company is likely systematically missing out on the “best possible” candidate, because speed-to-hire can simultaneously deliver qualified candidates, but at the same time be antithetical to “best possible” hiring.

The firm would need to decide what it cares most about, and then we would set up a measurable approach to achieving those outcomes.

First possible ≠ best possible.

The Business Case

• We might care about this because we believe the third party research, which tells us diverse teams get better business outcomes.
• We might also care because we believe it’s the right thing to do.
• We might care about this because we believe “quick and qualified” is not as good as “slower and awesome.”
• We must ensure that all hires are at the end based on merit; not actually hard to do, but it’s the scarlet letter of diversity hiring: a feeling that unqualified folks have been handed positions for which they were not qualified. My hunch is that for just about any role in a company today, there are enough, in absolute terms, qualified people to do that job from just about any underrepresented group. Those folks, by definition, are fewer in number, and therefore relatively hard to find.

Your firm gets to say what it cares about and why, but whatever the firm chooses, leadership must explicitly and unequivocally own that choice the to the board, customers, and employees.

Measurement

If there’s a notion that representation matters, then representation should be measured at the firm census level, at the management level, and at the executive level, perhaps even board level (one day).

If we believe we have representation gaps, those can be solved with three or four simple process changes

1. Clear goals — we must articulate clear goals for the representation we want. These should never be quotas. Quota = tokenism. Please don’t use that word.
2. Rubric-based hiring — especially for the culture fit interview; not controlling for bias is more likely to result in a majority group hire;
3. Introduce diversity analytics in our various hiring funnels. Insisting on diverse funnels helps ensure opportunities are exposed to folks we might not normally expose them to. The ATS Greenhouse, has an excellent tool for this.
4. “First possible ≠ best possible” — First possible will almost always be a majority group person. Best possible might take more time, but urgency bias should not force our hands away from diverse teams. Let’s be adults and note that occasionally “qualified and fast” is the right way to go for our business. Let’s also be adults and not lie to ourselves with our “best possible!” clap-backs.

Equity

Equity = Equitable treatment in our talent processes.

We should not have, for example, a gender pay gap. We should not expect men, say, to get systematically higher performance ratings. We should not expect women, say, to have systematically higher promotion rates.

Equitable outcomes in our talent processes are a little tricky to navigate, but for companies doing something around D+E+I, they are usually of a size at which they have sufficient scale to expect equitable treatment across all processes. Please remember that equitable is not the same as equal, which is why it’s ok to say we expect equitable outcomes. In the gender pay example below, you’ll see what I mean.

The Business Case

We care about equity not only because we care about justice and fairness, 💡but also because we want our incentives to be rational 💡and entice folks to perform because they believe that the incentives are fairly devised. That’s it. That’s the entire business case.

To make it a bit less abstract, let’s go ahead and grab a lightning bolt. Let’s take gender pay. Importantly a gender pay gap (If you prefer, we can call this inequitable pay practices explained by gender or we can call this discriminatory pay practice, because I guess gender pay gap has become some kind of a dog whistle for folks) does not get measured using average pay, though this is a common (and very, very wrong) practice. Why? Because it is possible to have an internal representation problem — ie higher concentration of men at higher levels in the hierarchy and higher concentrations of women at lower levels in the company, which will skew average pay in favor of men, and may very well not indicate a gender pay gap. It is arguably a representation (Diversity = representation) problem, but not necessarily indicative of a gender pay problem.

Whether there is a gender pay gap must be evaluated at the job+level couplet. For example, you would not expect to see systematically higher pay for male level 5 software engineers than female level 5 software engineers. You must also evaluate both cash compensation and equity compensation. In my experience, 9 times out of 10, when a disgruntled employee thought they were on the short end of the comp stick after having spoken to their colleagues, they had failed to discuss equity compensation, and were, in fact, paid equitably given the total picture.

You can also expect some differences based on performance and tenure. Careful with performance as a pay differentiator, because again, you would not expect, say, male employees in certain jobs to systematically outperform female employees in the same roles.

So to evaluate a gender pay gap, you should:

• Evaluate at the job+level couplet, not average pay
• Evaluate full comp picture, not just cash
• Control for performance
• Control for tenure

This comes all the way back around to say that it might be the case that a male L5 software engineer is paid more than a female level 5 software engineer, and that, though clearly not equal, is equitable, if he, for example, is a top performer and has longer tenure. Equal and equitable outcomes are not the same.

As a credibility building side note, my team at Qualtrics did this work, and we had a third party come in to evaluate our systems and findings, and they gave us permission to publicly say what we already measured, which was that “We do not have a gender pay gap.” Imagine that. A tech company that can say that publicly.

Measurement

Here’s a conceptual model for how the various talent processes interact with each other and ultimately affect the primary economic relationship between employee and the firm: Pay.

• The offer is the first place that compensation is affected. It affects pay directly now (black or white arrow), and over time because the offer will stipulate the level (top gray or red arrow).
• Level is / should be closely correlated with pay (black or white arrow) — During hiring levels should be determined using a 5 part rubric (not a formula, and there is not one): (the following numbered list is supposed to be nested under the bulleted “Level” but I can’t figure out how to do that on Medium. The bulleted list picks back up after #5, and restarts with “Promotions”)

1. Level budgeted for — if you’re hiring for an SDR (the most junior sales role), and a guy applies with 30 years of experience, he should expect to be paid entry level salary
2. Knowledge, Skills, and Abilities — capabilities that suggest one can do the job. Separate and distinct from years of experience. Maybe a younger employee who’d successfully founded a company accumulated lots of K,S,A, with relatively fewer years of experience.
3. Years of experience — Imagine a teacher effecting a late career transition. They might have lots of years of experience, but less K,S, A. Those years could be inherently valuable, but also maybe not.
4. Internal equity — Be very careful about bringing someone onto the team at a higher level than incumbents doing the same work.
5. Competitive considerations — this one should be used very infrequently, and I’ll argue only for senior folks. Sometimes you have to pay to compete. Don’t blow up your carefully devised comp structure at VP and below to compete. SVP and up, you might need to be more flexible.

• Promotions change your level (right hand side gray arrow), which ostensibly changes your pay (black arrow)
• Performance Management should be the primary input in determining whether someone gets promoted (bottom gray arrow), and in the case of, say, bonus structure tied to performance, it will directly affect pay. (black arrow)

These systems must be constantly tuned and improved

• Design them to not only help us achieve our intended business outcomes, but also to be fair and intentional around controlling for bias.
• Train managers to be able to own these conversations with their employees, including less than desirable outcomes (ie “I wanted to give Tom a high rating, but he didn’t calibrate” or “I wanted to make your promotion pay higher by they wouldn’t give me the budget”), to execute these processes in partnership with the People Team to help ensure fair, low bias outcomes.
• Execute — we have to run the processes! We can get a little scale by running most of them at the same time. Managers will be the primary execution channel.
• Measure — we must measure whether the outcomes we are seeing appear fair. Did men get too large a proportion of the highest ratings? Did men get too large of a proportion of the lowest ratings? (both are likely, btw, given 3rd party research). Are people of color getting promoted at a lesser/greater rate than expected? Are women being systematically offered employment at a lower level than their incumbent peers?

💡 Remember gender, and other pay gaps are not evaluated by “average pay.” The company demography tells us why. If there are higher concentrations of men at higher levels of the company, for example, then average pay will be higher for men. This is a diversity issue, not an equity issue. The gender pay gap is evaluated at the level+job couplet (ie IC5 engineer or IC3 marketer), controlling for performance. The evaluation should include cash and stock. 💡

Inclusion

We can understand Inclusion quite easily and measurably. (most of us have not earned the right to differentiate between inclusion, safety, and belonging. Once you’re measuring and managing inclusion at scale, then you’re ready to introduce complexity. As always be very clear about defining and measuring your new terms, like belonging)

Everyday inclusion might be this simple: “Make sure every voice is heard including your own.” Measurably understanding feelings of inclusion at scale and per manager, though, is somewhat more complex than that.

The Business Case

Please remember this: No matter who constitutes the majority group at a company, division, team, etc, they will very likely, almost entirely by accident, create an exclusive culture for those in the underrepresented groups, unless inclusion is actively measured and managed. It does not matter who constitutes the majority group or underrepresented groups. A company run by, say, black women, will almost certainly (and accidentally) create an exclusive environment for, say, white males. Before you get pissed at that last sentence, let me explain through a personal example.

I had this experience once. First a house-keeping note that works its way into the story. Long before COVID validated me, I was anti-hand shake (and also not a fan of hugging work colleagues). So much unnecessary germ transfer in a handshake, and so I fist bumped. It’s a fun, somewhat cool little way to enthusiastically greet someone, with significantly less germ transfer than a handshake. And to be super clear, I have consciously been, and remain, an equal opportunity fist bumper. When I come at you with the fist, you’re welcome to ask what the hell is happening and you’re welcome not to. All I ask is that you don’t run your jib behind my back about what the fist bump “means” without checking with me first.

Reminder, I am Mayor Majority. I attended an all-day seminar at the Clayman Institute, Stanford’s DEI think tank. There were approximately 150 attendees, and my colleague and I were 2 of probably 30 males, 20 white males, and 10 straight white males. (Approximate numbers). It was a rare time in my life where I could directly empathize with what it feels like to be out of the majority group.

A little background on our keynote speaker. This woman is a baller. She worked with Carol Dweck on Mindset, and was/is a PhD professor at a well known midwestern university. She was a great presenter, absolutely commanded the room, compelling content, knew her stuff. Baller.

We were doing some group exercises, and my group had a question. I volunteered to take the question to the keynote because I was excited to tap into her experience and wisdom. Here’s how it went.

“Excuse me, Dr. [Jones], my name is Russ Laraway, Chief People Officer at Qualtrics, and my group has a question.”

“Yes, you’re the fist bump guy.”

6 words that did a number of things. First this sentence characterized me as some kind of tech bro, which I am not. Second, it made it clear that people were discussing my tendency to fist bump as a greeting, behind my back. Third, these 6 words identified me in a way that I would never identify myself, and that identification was reductive in nature.

I am quite a bit more than “the fist bump guy” in this context. I’m an engaged participant. I’m a vocal advocate for the underrepresented. I’m someone with diverse and important experience to bring to this table. I’m a student.

What this keynote did, almost certainly unconsciously (if this was conscious, shame on her!), was leverage the massive power differential between us in this setting, and diminish me to a tech bro fist bump guy. She made it clear there was this ephemeral tech bro brand zipping around the seminar behind my back.

And I feel really lucky this happened because it directly taught me two major things:

1. For the first time in my senior career, I could feel something that many others feel far more often, and that many of us gaslight them over feeling: Exclusion. I’m not confused how fortunate I am to experience this less frequently than many others
2. It taught me that whoever constructs the majority group, which just so happens to most frequently be white, straight, males, is capable of creating an exclusive environment, and actually almost certainly will, entirely by accident.

Inclusion vs exclusion is a surprisingly powerful workplace force. In David Rock’s book, Your Brain at Work, he articulates 5 social threats common in the workplace. They are acronymized by SCARF:

• Status
• Certainty
• Autonomy
• Relatedness
• Fairness.

These manifest as literal, physical threats, which means that the employee’s brain is chemically operating in a threat state, which then necessarily and definitionally means that that employee’s brain is not in a thinking state.

💡The threats that people feel when they feel excluded tend to be around the RF in SCARF, or “Relatedness” and “Fairness.” 💡

While in a threat state, it’s not feasible for people to do their best work, and they certainly won’t be totally psyched while doing it. This, friends, is biology (Rock gives excellent treatment as to why in Your Brain at Work, and I give an overview in my book, When They Win, You Win, although my overview is not in a D+E+I context — it’s in the context of Improvement Coaching, or having harder conversations.) and this is how feelings of exclusion can erode a high-performance culture, and why it’s justifiable to measure and manage it.

Measurement

Inclusion is the least empirically and objectively measured or evaluated of the three flavors, D+E+I. Generally speaking, it is measured via survey, using concepts that tend to correlate, or explain, employee engagement. Companies like Qualtrics have out of the box Inclusion solutions. Here’s a link. They say it’s a DEI solution, but it’s not. It’s an I solution. D+E are not solved by their solution, but it is excellent, and heavily evidence and research based.

While I was there we ate our own dogfood, and used our own solution to measure inclusion across the whole company. We then did some demographic work to rough-cut our employees into either the majority group and the underrepresented group.

We found a 12 point inclusion gap — majority group folks felt more included than folks in the underrepresented category. The numbers were in the 85% to 73% range, which, I have to be honest, was much, much better than I had hypothesized.

What happened next floored me.

We set a company goal to reduce that gap from 12 to 6 by the end of the following year. There was no incremental investment in HR, no fancy operating plans. The folks on my (amazing) team wrote and uploaded “prescriptive actions” into our product. The product then teed up for each manager what their most leveraged inclusion areas were, and then it delivered them hyper relevant prescriptive actions to take. Since our (amazing) executive team agreed to set a goal against this measurement, and since we were measuring it transparently for the company, our (amazing) managers paid attention to it; they took these micro, prescriptive actions based on their unique team composition, unique culture, and their unique measured problems. It scaled beautifully.

By the end of that year, we did, indeed, cut that gap to 6 points. For the following year we set out to half it again (or more), but I punched out and really don’t know what happened.

Now sometimes, when I relate this case, people hear me saying some wildly arrogant thing like “WE SOLVED INCLUSION!”

No.

Helping people to feel included needs to be measured and worked on with great consistency. It is hard to do, and harder still at scale. Our managers at Qualtrics had been trained to use measurements to evaluate the state of their own leadership (I wrote a book about that!), and so to add another measurement, Inclusion, for them to focus on, located in the exact same tool where they got both their employee engagement and their manager effectiveness scores, turned out to make it rather easy for them to address their unique Inclusion issues.

I really want to highlight that inclusion efforts that are based on anecdotes, squeaky wheels, water cooler talk, or the CEO’s listening posts, are unlikely to systematically impact the company. The beauty of Qualtrics’ solution is that inclusion concepts are correlated with engagement (which predicts business results), and in a world in which it’s unreasonable for a company or manager to boil the ocean, the highest priorities items are teed up, per manager, at scale. Way better than the water cooler.

What Kind of Profile Should Run D+E+I Efforts for a Company?

This is going to be hard for many to hear.

DEI needs to be run in companies by a person/people with a measurement disposition. Majority group folks with a measurement disposition are no less capable than underrepresented folks with a measurement disposition, and of course, vice versa.

The typical anecdotal approach to DEI is failing, relentlessly, because these so-called strategies lack definition, goals, and rigor, and that in turn makes them lip service — high visibility activities that lead to nowhere special. I hear a lot of folks chartered with DEI responsibilities agree with this perspective, but then turn around and blame executive leadership. No, this is the strategy you must bring with you to the table and then drive.

The people who will drive measurable impact will be people who can find the data, free it, manipulate it, analyze it, identify and articulate findings, synthesize those findings, and help others to scale impactful actions.

Now in cases in which there is a large DEI effort, you might want more of a GM type, but that person must demonstrate that their DEI efforts will be built around an analytical approach.

Sometimes folks ask how Employee Resource Groups (ERG) fit into this. I like ERGs and think they are important. First, the most important (by a mile) aspect of the ERG is to create sub-community inside companies. ERGs, though, should not drive D+E+I strategy. If not to drive strategy, then, I think the perspective of the ERG should be used to help interpret measurements. This is exactly the opposite of how things are done in most companies today (and that’s if measurements are even a part of the process).

Strategy should be driven by measurements, though, not anecdotes. I suspect this is another hard truth for many.

A Simple D+E+I Business Case

💡 Inevitably, there will be new business/new logos, for which the contract is contingent upon your company having a good story to tell around its D+E+I practices.💡 I have no firm opinion about whether this is a smart requirement for buyers to emplace upon vendors. The reason I don’t directly endorse this is because what these buyers actually ask for is, well, terrible. Often, for example, prospects/buyers/investors ask to see a “DEI Policy.” What the hell even is that? So these buyers/investors are, in my opinion, asking the worst possible question. It’s very frustrating as someone who clearly takes this topic very seriously and gives it rigorous treatment. It’s lip service and box checking. Pointless.

This happened to us at Qualtrics — lots of buyers as part of their diligence, wanted to know about our “DEI Policy,” and luckily we’d been running a measurable D+E+I strategy in the background, which allowed us to have a surprisingly strong story to tell our prospects who required us to have a plan.

We received exclusively positive feedback, and guess what. We never had a “DEI Policy,” and our D+E+I strategy ran circles around anything else being done by any of those buyers. (Take a bow, team you-know-who-you-are)

Now a word of caution. A system like this works, I know for a fact, but it only works when you have executive and manager buy-in. At Qualtrics, most executives were quite supportive, and others were more in “I won’t get in the way” mode. Most importantly, we had the top guy, Ryan Smith, on board. And by having already established measurement systems for our managers, it wasn’t difficult, even at scale, to introduce these concepts to them, and ask them to get on board.

Now, here’s the thing — this is a system that works at scale. That means that if you were to run a system like this, you are very likely to see company measurements improving. And even in the context of measurable progress and improvement, it will still be the case that someone is on a thoroughly homogenous team, that someone is measurably being treated inequitably, and someone is feeling excluded. Such is the nature of scalable systems. We improve, tune, and reiterate these systems, and you’d think it’s possible to achieve perfection, but it isn’t. The system works, and it will never fully solve for D or E or I.