Amongst the shiny visions of driverless futures and bitcoin fueled disruption, the needle of “diversity and inclusion” in tech has moved unacceptably slowly. While most people agree that lack of diversity is a problem, no one knows what diversity metrics look like across most companies. This makes it especially hard for those seeking employment; one jobseeker told us, “scrolling through LinkedIn to find other black people isn’t sustainable.”
After talking to 50+ HR leaders and techies, we decided to create DearTechPeople to unearth the data behind diversity in tech starting with a race/gender ranking of 100 top tech companies. We are excited to build upon previous work such as Tracy Chou's Women in Tech list, Project Include, Diversitylist, and Opendiversitydata.org.
Choosing Companies: We chose 100 tech companies, focusing on the "darlings" - the ones that repeatedly show up on top lists (Breakout List, Career Launching Companies, LinkedIn Top Startups). All companies we chose have between 50 and 3000 employees.
Collecting Profiles: We analyzed thousands of profiles pulled from online sources, such as LinkedIn, and consolidated diversity numbers at a company level.
Determining Gender & Race: We employed a combination of name analyzers, facial recognition technology, and manual identification through Mechanical Turk to determine an individual employee’s race and gender.
Defining Leadership Roles: There’s very little standardization regarding what constitutes “leadership”. Some companies (like Github) indicate that leadership is anyone who is a manager, while others (Slack) define it as Directors and above. We’ve chosen to classify leadership as anyone who is a VP and above. This includes the following titles:
CEO, CTO, CMO, COO, CFO, CPO, Founder, President, VP, Head of Design, Head of Engineering, Head of Sales, Head of Product, Chief Security Officer, Chief Architect, and Chief Customer Officer
Defining Technical Roles: We defined technical to be anyone who conducts a job that is technical in nature, regardless of whether they are specifically in the engineering department. As such, we’ve classified the following roles as “technical”:
Software Engineer, Engineering Manager, Analytics, Data Scientist, Machine Learning, Front-end/Backend/Web Developer, IT, System Architect, Programmer, Data Center Manager, Database Administrator, Release Manager, Information Security Specialist, Software Tester
Categories: We opted to use Latinx over Latino for a more inclusive term. This category encompasses both those who identify as Latinx or Hispanic. Currently, we are unable to categorize based on sexuality or identities outside the gender binary. Unfortunately, this information is simply not available via our public datasets, nor could it be extrapolated with the tools available. Similarly, we are unable to provide data on Native American representation at this time. We hope that as Dear Tech People becomes more sophisticated, we’ll be able to provide more data in these respects.
Unidentified Profiles: Those who were unidentified in gender, are classified as with a light grey color and marked “unidentified.” Those who were unidentified for race are counted under the “other” category in orange.
Geography: For those companies above 1,000 people, we looked at only at US-based employees. Otherwise, the ranking takes into account all employees globally.
We compared our data with pre-existing, company-released diversity reports to ensure our data collection and analysis process was sound. These companies included publicly available reports released by Clover Health, Buffer, Clever, among others, as well as some companies with private data not yet released. Below is an example comparison between our data and BuzzFeed’s diversity data released for 2017.
| BuzzFeed 2017 Diversity Breakdown | |||||||
|---|---|---|---|---|---|---|---|
| Male | Female | White | Asian | Black | Latinx | Other | Undeclared |
| 45% | 55% | 64% | 13% | 7% | 8% | 7% | 2% | Dear Tech People BuzzFeed Breakdown |
| Male + unidentified | Female | White | Asian | Black | Latinx | Other + unidentified | |
| 46% | 54% | 67% | 12% | 8% | 10% | 4% |
| BuzzFeed 2017 Diversity Breakdown | |
|---|---|
| Male | Female |
| 45% | 55% | Dear Tech People BuzzFeed Breakdown |
| Male + unidentified | Female |
| 46% | 54% |
| BuzzFeed 2017 Diversity Breakdown | |||||
|---|---|---|---|---|---|
| White | Asian | Black | Latinx | Other | Undeclared |
| 64% | 13% | 7% | 8% | 7% | 2% | Dear Tech People BuzzFeed Breakdown |
| White | Asian | Black | Latinx | Other + unidentified | |
| 67% | 12% | 8% | 10% | 4% |
Verified Companies: Those companies listed as verified have checked our data against their own internal data. They submit official numbers based on Dear Tech People standardized definitions. If you are interested in becoming a verified company, please let us know.
Disclaimer: The information we share on Dear Tech People is based off of publicly available information like LinkedIn profiles. As a result, the raw data (from LinkedIn for example) may not be be 100% accurate or clean. Gig economy companies are likely to be less clean, given the sheer number of contractors that are listed on LinkedIn. We have done our best to filter through those contractors across all companies. We do not assume liability for the perpetual up-to-dateness or completeness of publicly available data.
Ranking Types: We have three different rankings. First, we have our official, overall ranking, which holistically evaluates companies based on a composite score. We also have separate leadership and technical rankings. All three rankings take both underrepresented demographics and a Simpson’s diversity index calculation into account.
Simpson’s Diversity Index: This is a measure of diversity that calculates the number and relative abundance of demographics present. Diversity goes up as a result of an increase in demographic richness and evenness. The Simpson’s Diversity Index is one of several variables used in the ranking. This measure is commonly used in biodiversity research and amongst universities to measure diversity of applicants and students.
Composite Score: We created the overall rankings using a composite diversity and inclusion score assigned to each company. These scores are based on a calculation that weighs gender and race-based diversity across all types of roles, as well as the Simpson’s Diversity Index. More specifically, we looked at overall gender diversity, overall race diversity, diverse representation in leadership roles, diverse representation in technical roles, and intersectionality of race and gender.
Ranking Benchmarks: To further contextualize our rankings, we have also included where the American population, the public (non-standardized) Google diversity report, and sector averages fall. We often found patterns to companies that perform well--consumer companies consistently rank higher than enterprise companies. For example, these benchmarks show what security companies are out-performing the security sector average, which tends to rank significantly lower than other sectors.
Leadership and Technical Ranking: The leadership ranking counts women and non-White men as diverse leaders. The technical ranking counts women and non-White or Asian men as diverse in technical roles. Given the percentage of diverse representation out of total roles, we have ranked both the leadership and technical roles.
DearTechPeople is a work in progress. We’re constantly thinking of ways we can improve and what features we might want to include to move the needle forward.
1) Data Quality: Online profiles are not 100% representative and our process for identifying demographics has its limitations. We'd like to improve our data quality by partnering with companies and reporting on actual numbers. Moving forward, we hope to see more companies release their own standardized reports on DTP in which employees self-report their identity and categorize their job title (technical/leadership).
2) Broader Diversity: Gender and Race are just the tip of the iceberg. We'd like to include many other forms of diversity such as age, sexual orientation, ability, a complete gender spectrum, and more inclusive racial categories.
3) Best Practices: Beyond the metrics, we'd like to share data on policies, benefits, and programs at various companies to better define best practices in creating a diverse and inclusive work environment.
Verified Companies: A big part of the Dear Tech People vision is for more companies to participate in our verified companies program and commit to data transparency and diversity. Verified companies get a verified check mark on the front page, meaning they have self reported diversity data to Dear Tech People standards and definitions. As more companies get verified, we see the potential for more sophisticated data in terms of those who may identify as transgender, outside the gender binary, two or more races, Native American, and other complex intersectional identities.
Feature Requests: We want to hear from you, the reader! What product features do you want to see in future versions? How can we better incentivize companies to invest in inclusion initiatives? What companies should we add to the list next? Diversity and inclusion is a collaborative effort. Let us know your thoughts!
Many thanks,
DTP Team
Dhruv Maheshwari
Adina Luo
Wyatt Shapiro
Questions? Press? Want to help? Contact us deartechpeople@gmail.com