Tracking compensation and promotion inequity
Plenty of tech companies are attempting to make their pipeline of candidates more diverse. But an organization won’t find much success recruiting a more diverse group of employees unless its leaders are aware of their existing internal inclusion and equity issues. Unless leadership has already started to tackle these issues, it’s likely that these new hires will enter into an environment that they won’t want to stick around in for long.
One of my suggestions is to calculate whether you compensate and promote people fairly, which requires some level of manual analysis. It also takes a lot of work to make this math repeatable, so you can check in on your organization’s progress over time. And if you’re doing this analysis for the first time, you probably won’t have statistically significant results, because you may not have enough folks from underrepresented communities yet.
Below, I detail some of my experience doing this work. But first, check out this excellent Project Include resource with TONS of great info about what to measure, the importance of the privacy of demographic data, and tons more actionable tips on measuring your company’s progress. I’m not an HR professional, nor am I an expert like the Project Include folks, so please take this post as just a starting point for your own analysis.
Salary
Calculate the percentage of total people in the department that are members of underrepresented groups, and the percentage of total yearly compensation paid to them (inspired by Cate Huston). Be sure to look at this holistically—include bonuses, vesting of stock options, etc.—and follow Project Include’s Guidelines on demographic categories.
| Year | % people in department are (category) | % total compensation to (category) |
|---|---|---|
| 2015 | 5% | 3% |
| 2016 | 15% | 20% |
| 2017 | 25% | 20% |
Read Jason Wong’s post Bootstrapping Inclusion: Education & Effecting Behavioral Change to read about how he structures Compensation Calibration meetings. Also check out this example template for tracking and measuring compensation changes, based on Jason’s original comp calibration template.
I often see members of underrepresented groups hired at more “junior” levels in an organization1, which means that total dollars for URMs will skew even lower compared to the total department compensation. When possible, look at the percentage of URMs at each level, in addition to the compensation analysis.
| Level | % people are (category) | % total compensation to (category) |
|---|---|---|
| Engineer | 60% | 80% |
| Senior Engineer | 40% | 40% |
| Staff Engineer | 40% | 20% |
| Manager | 50% | 50% |
| Senior Manager | 20% | 30% |
In this made-up example, 40% of staff engineers are a member of an underrepresented group, but those people are only paid 20% of the total yearly compensation given to that group. They’re likely underpaid relative to their peers; time to address this under-compensation!
Resist the urge to inspect on a case-by-case basis
In my experience, when questioned about these disparities, managers will defend why a URM is compensated less than their peers. Due to unconscious bias and trends like marginalized groups receiving much less helpful and actionable feedback, people who are otherwise very in-tune to D&I topics like the pay gap will still look for other reasons why someone is under-compensated2.
Correcting wage inequity requires constant effort and vigilance, and sweeping fixes. Keep pointing to your high-level analysis and remind people that this is a systemic issue that needs to be corrected across the board.
Making the corrections
A natural reflex from HR may be to incrementally bring someone’s base compensation up over time by making changes more frequently than once or twice a year. I haven’t seen this be effective. While you are slowly bringing someone’s compensation up to match their peers’, their peers are getting promoted and additionally compensated, too. In my experience, I’ve found it best to push for immediate, full corrections to comp to bring people in line with their peers before the next review cycle.
When you make a double-digit correction to someone’s base salary, you’ll need to prepare for what could be an uncomfortable conversation with that person. You could couch it in “you’re doing a great job, so here’s a pay bump!”, or you could choose to be transparent about the reason, like “we are bringing your compensation to be in line with your peers”3. It’s up to you (and your HR team) to figure out how to have that conversation in a respectful manner.
Retention
I’ve found it personally helpful to compare the makeup of my engineering team to the makeup of my outside community over time, in addition to whether I am retaining members of underrepresented groups.
I live in Brooklyn, and these are Brooklyn’s demographic stats from the American Community Survey (ACS) in 2009:
| White | 46.6% |
| Black or African American | 34.2% |
| Hispanic or Latino | 19.6% |
| Asian | 9.5% |
| American Indian and Alaska Native | 0.3% |
| Native Hawaiian and Other Pacific Islander | 0.1% |
| Multiracial | 1.4% |
I also hire remote employees outside of Brooklyn, but nonetheless, I’ve found it personally useful to have a benchmark in my head of what the outside community represents. I don’t want to compare my organization to other tech organizations because, frankly, the bar is far too low.
In terms of retention, figure out if members of underrepresented groups are sticking around in your company’s environment at the same rate as the overall population. What’s the median lengths of tenure for URMs? How do these numbers compare to the overall tenure of your employees? Measure start date to departure date, and again, look at the categories that Project Include lists out. A made-up example:
| Category | Median Tenure |
|---|---|
| Overall | 200 days |
| Women and Nonbinary People | 140 days |
| Non-White Race/Ethnicity | 150 days |
Also capture the overall makeup of the organization every six months or so to see how much the makeup of the group is shifting (or not). Don’t just count the number of hires from underrepresented groups you’ve made; you may be losing non-white non-men at a faster rate than you’re hiring them.
More made-up examples measuring the diversity of an Engineering org over time:
| Q4 2016 | Q1 2017 | Q2 2017 | |
|---|---|---|---|
| African-American/Black | 1% | 2% | 3% |
| East Asian | 1% | 2% | 3% |
| Hispanic/Latinx | 1% | 2% | 3% |
| Middle Eastern | 1% | 2% | 3% |
| Native American/Alaska Native/First Nations | 1% | 2% | 3% |
| Pacific Islander | 1% | 2% | 3% |
| South Asian | 1% | 2% | 3% |
| Southeast Asian | 1% | 2% | 3% |
| White | 94% | 84% | 76% |
| Men | 80% | 70% | 60% |
| Women | 19% | 28% | 37% |
| Other gender identity | 1% | 2% | 3% |
Promotion
Measure start date to date of promotion for folks in each level, and compare over time. You’ll need to get a bit fancy with your math if people have been promoted multiple times since they were hired. Another made-up example, with a placeholder for the categories mentioned in the Project Include list:
| Level | (category) promotion velocity | Org-wide promotion velocity | Difference |
|---|---|---|---|
| 1 to 2 | 12 months | 18 months | 50% faster |
| 2 to 3 | 24 months | 24 months | Equal |
| 3 to 4 | 36 months | 28 months | 22% slower |
Obviously a lot of this is chicken/egg - if you’re not retaining URMs, how will you be able to promote them? And if you’re not promoting them fairly, then you won’t retain them either.
At Etsy early in my tenure there, I found that men in Engineering were being promoted much more quickly than people who did not identify as men. In fact, there were no women in the staff engineering cohort at the time. To tackle this, a buddy and I led trainings for staff engineers and engineering directors about sponsorship, and requested unconscious bias trainings for these groups too.
I continued to do this assessment, and though there were some seasons in which managers promoted women and nonbinary people slower than men, we caught up a bunch of people and ended up closer to a more equal pace. Though there will always be more work to do to make the senior engineering cohorts look like the overall engineering demographic, I’m proud of the work we did towards this goal.
Choose your words carefully
When describing the statistically significant difference in rates of promotion to those leadership groups, I chose my words carefully. Rather than “women and nonbinary people get promoted” or “earn promotions”, I used “we promote women and nonbinary people more slowly”. Because, after all, it is the group of managers who are doing the promoting at an unfair rate, rather than women and nonbinary people not earning the promotions as quickly.
The same applies for other inequity you find: “we compensate people of color 10% less than white people” rather than “they earn”, “we retain people with disabilities” rather than “they stay”. Though subtle, I’m hopeful that it reminds leadership that the onus is on them to both correct these issues as they’re surfaced, and continue to track and correct these issues going forward.
1 Such as when an organization does a lot of hiring out of bootcamps, or mislevel new hires due to unconscious bias. There's also that statistic that hits too close to home for me: after 10 years of work experience, 41% of women in tech leave the industry, compared with 17% of men. So you'll likely find less women available to hire for more senior roles.
2 Example: I had heard in one company that HR and the leadership team had done a round of compensation corrections, but one of my direct reports hadn't had her wage corrected to bring her in line with what her peers at the same level were making. I had been working fixing the gap with HR for about a year, making incremental improvements, as she had not received any raises in the years she had worked there (under two different previous managers).
When I asked a VP about why she wasn't included in the corrections, he said he remembered that she had had some performance issues years ago and he hadn't thought to ask me if things had gotten better. Instead, he simply excluded her from the compensation corrections. This is the depth that our unconscious biases can come into play when looking at individual cases, even while well-intentioned folks are actively trying to fix these systemic issues.
3 I've been on the receiving end of this conversation multiple times. The one that I appreciated the most was when my manager was frank with me. Yes, it sucked, and yes, I carried some baggage around afterward about my previous manager, but overall I'm so appreciative that this person was honest with me. If you're reluctant to have this conversation transparently, remember: your direct report isn't dumb. Trust me, I've talked to LOTS of white women and people of color about guessing the unspoken reason behind a major compensation bump.
New: The Complete Demystifying Management Video Program
I've transformed my renowned in-person workshop for managers into this self-paced virtual program. It's accessible, actionable, and will level you up each step of the way. Learn more!