At Uber, discrimination was baked into the system, a new suit alleges, pointing to systemic bias against women in the ride-sharing company’s performance evaluations.
To assess and promote — or fire — workers, Uber Technologies Inc. used a practice known as stack ranking, a zero-sum evaluation that pits employees against one another. According to the lawsuit, filed in October by a former engineer at the company, Roxana Del Toro Lopez, stack rankings unfairly and disproportionately hurt women.
Microsoft Corp. and Goldman Sachs Group Inc. have faced similar legal challenges; both firms and, more recently, Uber, have abandoned the practice.
It’s still common in big U.S. companies. Almost one-third of Fortune 1000 companies use stack rankings, according to a survey by research firm CEB Inc. The process forces managers to measure their employees against one another, so people with similar skills can’t be rated equally — even if their performance is equivalent. At Uber, women regularly got penalized by this system, the suit alleges, which resulted in lower pay and fewer promotions.
“The problem with forced ranking is that it forces managers to make artificial distinctions between employees that don’t reflect performance,” said Jahan Sagafi, the lawyer representing Del Toro Lopez.
Uber declined to comment on the lawsuit. The company overhauled its employee ranking system earlier this year, part of a broad reorganization influenced by widespread reports of bad behavior and sexism at the company. The company still uses performance evaluations, but instead of rating and ranking employees, managers provide feedback and set goals.
As part of the overhaul, Uber also surveyed employees about what they did and didn’t like about the old system. Many said the ratings were too subjective and asked for more feedback.
Academic researchers have found that performance rating systems like stack rankings play to managers’ unconscious — and conscious — biases. Reviewing a decade of performance reviews at a “large professional services firm,” Paola Cecchi-Dimeglio, a senior research fellow at Harvard Law School, found that women were 1.4 times more likely than men to receive critical feedback in highly subjective categories.
For example, in one pair of reviews a female employee was described as having “analysis paralysis.” A man with the same behavior was praised for his careful thoughtfulness. “There is a lot of bias in the system, more than in the people,” Cecchi-Dimeglio said.
In a 2015 lawsuit against Microsoft, Katherine Moussouris alleged that forced rankings “systematically undervalue female technical employees relative to their similarly situated male peers.” A gender discrimination lawsuit at Goldman Sachs in 2010 also faulted the review process, contending that male managers favored male employees, disproportionately placing them in the top rung of performance.
“We strongly disagree with the contentions in the case because data and other information is mischaracterized,” Microsoft said in an email. It is defending the case in court. Goldman Sachs declined to comment.
Beth Steinberg, who has worked for two decades in human resources in tech companies, says she saw the biases in the system unfold in real time. When she worked at Hewlett-Packard, she said, “My observation was whoever was the best orator got their people ranked higher.”
In organizations where leaders and managers are mostly male — which is almost everywhere — men tend to fight for other men. “I was horrified even in a company as good at it as HP,” she added. The company says it has changed its performance evaluation system.
In the last few years, about 6 percent of Fortune 500 companies, including Accenture, Amazon, and even GE, have abandoned stack rankings, according to CEB. Many have replaced employee ratings and annual reviews with regular feedback, a process that Cecchi-Dimeglio has found leads to more objective reviews.
If the performance evaluations are biased, and raises and bonuses are based on performance, it will hurt women’s compensation, said Anna Tavis, a clinical associate professor of human resources development at the New York University School of Professional Studies.
“This pay-for-performance system is fundamentally flawed,” Tavis said in an interview. “The performance element is very subjective.”