After weeks of scandal, Uber released its first report Tuesday detailing the gender breakdown and racial makeup of its employees. Though welcomed by diversity advocates, the revelations weren’t that surprising: Within engineering, only 15 percent are women and 6 percent are black, Latino or multiracial, which makes Uber not that different from other tech companies, where those groups make up a tiny fraction of the technical workforce.
In fact, the more illuminating piece of data about diversity at Uber isn’t in its report at all. Retention — the percentage of each demographic group that stays at the company each year — reveals more about the hospitality of the workplace culture than the percentage of white and Asian men there overall. Most companies, including Uber and its peers in Silicon Valley, track it. They just don’t make it public.
Retention data is a proxy for job satisfaction, and tech companies won’t be able to build the diverse workforces they claim to be working toward unless they can hold onto the small fraction of female and minority employees they successfully recruit. Since 2014, demographics at Apple Inc., Facebook Inc., Google, Twitter Inc. and others have stayed roughly the same: Women make up about 17 percent of technical workers; only about 6 percent are black or Latino.
Across industries, few companies release employee retention data, but tech firms have consistently held themselves up as meritocracies, based on data-focused decision-making. At the same time, they’ve assembled homogeneous employee bases, and come under fire for tolerating behavior and attitudes that make minorities feel unwelcome. As far as data and transparency, they have mostly stuck to releasing annual demographic breakdowns and the occasional tidbit about the race and gender of new hires.
“People are looking for ways to tell a good story around their diversity and inclusion efforts,” said Ellen Pao, co-founder of Project Include, which advocates for more diversity in Silicon Valley. “The fact that tech companies aren’t sharing their retention numbers indicates those numbers likely aren’t good.”
A handful of industry-wide studies support Pao’s suspicion, at least as far as women are concerned. A 2008 survey of women in tech found that female workers in the industry leave their jobs at twice the rate of men. A 2013 study found that after about 12 years, 50 percent of women in STEM fields — mostly in computing and engineering — changed their professions, compared with 20 percent of professional women in non-STEM fields. The studies didn’t ask the same questions of other underrepresented groups, though a new study expected in April from the Kapor Center for Social Impact will.
Meanwhile, retention numbers are trickling out of the Valley’s biggest companies. In a recent essay, former Uber Technologies Inc. software engineer Susan Fowler described how her request to transfer to another team was denied, and she couldn’t figure out why. Later, she learned, her manager was more concerned with maintaining the retention rate on his team.
“It turned out that keeping me on the team made my manager look good,” Fowler wrote. “I overheard him boasting to the rest of the team that even though the rest of the teams were losing their women engineers left and right, he still had some on his team.”
Fowler took it upon herself to estimate the retention rate for her department, which had gone from about 25 percent female to 6 percent in a year, according to her post. “Women were transferring out of the organization, and those who couldn’t transfer were quitting or preparing to quit,” she wrote. “There were two major reasons for this: there was the organizational chaos, and there was also the sexism within the organization.” Fowler left Uber in December. Uber declined to comment on its retention rates.
Uber is now investigating Fowler’s claims, and the company promised to release findings from that investigation next month. But it had long been tight-lipped about any diversity-related data. Even recruiters tasked with hiring more people from underrepresented groups were denied access to information about the company’s makeup, Bloomberg reported last week.
Tech companies all collect data on retention, said Joelle Emerson, CEO of Paradigm, which consults with Silicon Valley firms about diversity. “So it’s just about analyzing and reviewing this information regularly.”
So far, only Intel Corp. has released this data. The company had long tracked its retention rate internally, and it wasn’t good: Underrepresented races and women were more likely to leave the company than white or Asian men. But after Intel started to release retention data in 2015, it set a goal for the next year to keep exit rates even for all groups and reached it in 2016. Next year’s goal is to retain underrepresented groups at a higher rate than their counterparts.
Intel Chief Diversity Officer Danielle Brown acknowledges that publishing the info can be scary. “Are there risks? Sure,” she said. “No one wants to put out a bad retention number that makes it look like there’s a problem.”
Collecting and publicizing retention data is important because tech companies are often more swayed by hard numbers than personal anecdotes, said Erica Joy Baker, another member of Project Include who left her job as a software engineer at Alphabet Inc.’s Google in 2015. The year before, she had written publicly about the challenge of being one of the few black women at Google.
In 2015 Baker made a public call on Twitter, using the hashtag #RealDiversityNumbers, for tech companies to release retention rates and other metrics instead of just demographic breakdowns. “People leaving the company is the strongest signal that something has gone wrong,” she said.
Around the same time Baker left Google, other women who had departed the company shared their experiences of casual sexism and sexual harassment on Twitter. But it wasn’t the most effective medium. “Had that been presented in data form, or if the company had had to report that one year, some percentage of their women in engineering left, I think they would have been more incentivized to work on that,” Baker said. Google declined to comment.
Some turnover can be healthy, and exit rates can vary widely between companies and from year to year. Retention metrics can also hide perverse incentives — employees who don’t leave because they can’t find a better job, or those who stick around longer than usual to take advantage of stock options. Because so few companies make the data public, there isn’t really a known standard, said Emerson, the diversity consultant. No matter what the numbers are, she added, companies should aim to not see big disparities in the retention rate across demographic groups.
Nevertheless, the retention metric remains a critical indicator of culture inside a business that isn’t shown by hiring data, said Freada Kapor Klein, co-chair at the Kapor Center. Companies are reluctant to release it for now, but they might be convinced if it starts to hinder them in one important way: hiring.
Also Read: As criticism mounts, Uber seeks chief operating officer to temper CEO
“Any woman engineer, any engineer of color who is looking for a respectful work environment and has some options about where they might go, ought to be asking companies that they’re interviewing with for retention data,” Kapor Klein said. “When highly sought-after talent begins to demand the information, companies have to pay attention.”