Our company is celebrating the second anniversary of one of our most successful programs, the IBM Data Science Elite Team. Since its launch, the team has grown from only five data scientists serving a few clients to nearly 100 members and more than 130 clients around the world.
These elite data scientists form small teams that work directly with companies to help push their first AI models into production. Since the program started in early 2018, we’ve helped more than 100 companies kick-start their AI efforts in a range of industries, including JPMorgan Chase, Lufthansa and Siemens. We’re proud to often hear happy clients say things like, “How did you put this team together? They’re awesome!”
But you know what makes us just as proud? When clients comment admiringly on the diverse makeup of our elite team.
Worldwide, we have female representation that is about 75 percent higher than the industry norm in data science. We are also proud of the racial and LGBTQ diversity we have been able to build within our teams, of whom approximately 20 percent have PhDs and 60 percent have master’s degrees.
Clients remark on the diverse makeup of our teams not because they’re surprised that skilled data scientists come from a broad range of backgrounds. It’s because clients so seldom encounter such diversity in our highly technical field. In fact, the lack of diversity is a widely recognized shortcoming in the data science profession, particularly in the specialty of artificial intelligence, in which human bias can have such a negative influence on the algorithms on which AI depends.
Consider just one measure: gender.
According to a study of data scientist salaries conducted by Burtch Works, the executive recruiting firm, only 15 percent of data scientists are female. The percentage of women in data scientist manager roles is even smaller—less than 10 percent. And that is despite a large number of potentially eligible women.
In the U.S. in 2019, for instance, women represented 43 percent of the workforce of scientists and engineers. Additionally, there are female-focused data science groups that have tens of thousands of participants. For example, the Global Women in Data Science (WiDS) Conference attracts more than 100,000 participants each year.
The dearth of diversity is an industry shortcoming we have consciously set out to change through our recruitment and hiring process for the Data Science Elite Team—not only because it’s the right thing to do, but because it leads to better business outcomes.
The Business Value of Diversity
There’s ample evidence that diversity is good for business.
A recent McKinsey report found that companies ranked in the top quartile for racial and ethnic diversity were 35 percent more likely to have financial returns above their respective national industry medians. The report found that companies in the top quartile for gender diversity were 15 percent more likely to have financial returns above their respective national industry medians.
Diverse teams are more likely to make more accurate fact-based decisions and are more innovative, as reported in the Harvard Business Review. They can also make companies more productive, according to another HBR piece.
But diversity in hiring doesn’t happen by simply wishing it so. We’ve learned that from experience.
Seth recalls an episode in 2013, long before the creation of our Data Science Elite Team program, when he was simply trying to find one good data scientist to add to his staff. Not only was he having trouble finding qualified candidates, but there was zero diversity in the candidate pool. When he voiced his frustration with his wife, she made an astute observation: Seth had written a job description containing more than two dozen qualifications. It was, she said, too much a wish list. Most of the criteria were not skills necessarily required for success.
They worked together to winnow the job description to a short menu of desired attributes. When this new “help wanted” posting went out, Seth suddenly had twice the number of qualified interview candidates, including women, people of color and other underrepresented groups. In the end he hired a highly qualified candidate who also happened to be Latino, adding new skills and racial diversity to the team.
And so, when it came time to begin assembling the IBM Data Science Elite Team two years ago, in a partnership with Susara, Seth could draw upon those hard-won insights.
Applying Diversity to the IBM Data Science Elite Team
We knew from the start that we wanted our elite team to be a client-facing organization, one that could help customers identify and plan AI data science projects that could create immediate business value. To do that, in a variety of business settings around the world, we wanted to hire not only a highly skilled team but a diverse one.
Our approach was essentially to reevaluate the entire hiring cycle, from writing the job descriptions to making the candidate an offer. Here’s how we outlined the approach:
- Job Descriptions: Our first step was to create job postings and descriptions deliberately designed to not discourage underrepresented groups from applying. An HBR article by Tara Sophia Mohr demonstrated that men are twice as likely as women to apply to a job posting that they are not 100 percent qualified for. Based on this study and past experience, we crafted job descriptions seeking variable skills and using gender-neutral words.
- Final Candidate Pool: The next step was setting a rule: No interviews until we had a diverse pool of qualified candidates. Although we set no numeric or demographic requirements for hiring from this pool, we drew confidence in this approach from a 2016 HBR study by Stefanie K. Johnson, David R. Hekman and Elsa T. Chan. Their study found that if there is only one woman or other underrepresented minority in a candidate pool, that sole person essentially has no chance of being hired. But when there are at least two women or two members of other minority groups in a hiring pool, regardless of the size of the pool, the odds of hiring a woman is nearly 80 percent greater and the odds of hiring a person of color nearly 200 percent greater.
- Interview Process: We created a highly rigorous interview process consisting of shallow technical skill interviews, deep technical skill interviews, coding challenges and soft-skill interviews. Importantly, we made sure the interview process was carried out by a diverse hiring team.
By following this process, we have built a global team much more diverse than is typically found in data science. And the business success of our Data Science Elite Team model speaks for the results.
Susara, who led the hiring of the European IBM Data Science Elite Team, credits Seth for having insisted upon diversity-focused hiring. So far, nearly half of her 25-member European team are women, and the group comprises 18 different nationalities and speaks more than two dozen languages.
While we believe we can be more diverse, we are very close.
Our goal was to create a diverse team of highly technical and highly talented individuals, not only because it’s the right thing to do, but because it drives better business outcomes. There’s more to be done, and we’re constantly working on improvements. But we’re thrilled with the results thus far.