"From our clients, we know that many of them have a percentage of hires from referrals that's over 50 percent. But if you're already looking at a big disparity in your workforce -- if 80 percent of those referrals are white men or Asian men, then there's your problem," Snyder says.
One way to address this is to specifically ask your existing talent to go through their networks more thoughtfully, says Michelle McHargue, talent partner with Cowboy Ventures, a venture capital firm that focuses on startups. Admittedly, McHarque says, it's sometimes easier for the companies she works with since they're still in the seed stage and can actively plan for and integrate diversity from the start, without having to undo bad habits that exist in older, more established organizations.
"We do what we call 'sourcing jams.' We ask everyone to go through their networks and instead of recommending someone who's like them, find women and candidates from underrepresented groups to put forward. We're asking folks to delve into potential candidate pools they'd not considered," McHarque says.
And make sure this step is done at the beginning of the sourcing process, McHarque adds. You don't want to start with a large, homogeneous group of candidates and then suddenly introduce women and minority candidates toward the end, when they'd be at a disadvantage, she says.
Watch your language
Writing a job description is usually a dreaded chore, and most organizations want to take the path of least resistance; that means copying and pasting from competitors, from templates and from previous internal role descriptions without much thought toward how candidates will perceive these, says Textio's Snyder.
"From our data, we know that there are so many unconscious signifiers in job descriptions, and women and minority candidates are so much more sensitive to the nuances of language. For example, we know that if more than 50 percent of your job description content is presented as a bulleted list, the number of women who'll apply plummets. If it's less than 25 percent, the number of men applicants plummets. Neither scenario is good," she says.
Snyder adds that specific terms like "rock star," "ninja" and "badass" are shown to discourage women from applying. Those are fairly obvious. But other linguistic signals are more subtle.
"If you're looking for someone to 'manage' a team, you'll attract more men. If you say you want to 'build' a team, more women will apply. But if you say you want someone to 'lead' a team, you'll get about half men and half women. The language you use is extremely important," Snyder says.
Other language to avoid: military and sports analogies, which tend to turn off any applicant for whom those comparisons don't resonate -- and that's a lot of them, Snyder says.
Sign up for Computerworld eNewsletters.