New York-based Pymetrics is betting that it can minimize bias and improve the performance of candidate-screening by identifying the traits that contribute to a person’s success in a role and applying them to your incoming job applications.
Everyone talks about bias in recruiting, and more HR experts are looking to technology to be at least part of the solution. Some talk about the blind hiring model used by many orchestras as the solution. Call us old-fashioned, but we can’t imagine managers meeting new employees on their first day of work. That may be an extreme and unlikely outcome, but one thing about organizations: They love to automate.
A number of companies have developed products to attack the issue. Textio, for example, analyzes the hiring outcomes of million job postings and uses the results to predict the performance of the listing you’re working on and suggest improvements.
The company says its customers recruit 23 percent more women than the average employer—which obviously helps thwart social biases—but also 25 percent more candidates qualified enough for an interview. It does this, the company says, by analyzing the job postings in its data set identify “the meaningful language patterns that cause some posts to succeed where others fail,” and using its findings to help users craft more effective postings.
This kind of approach seems to be the basis for many HCM technology vendors: Use analytics and coaching to help recruiters craft job postings and other career-marketing materials to minimize bias, whether it’s about age, sex, race or job skills.
Pymetrics’ Different Approach
Founded in 2012 by two MIT graduates—neuroscientist Frida Polli and data scientist Julie Yoo—Pymetrics uses neuroscience-based games (“talent assessments, some observers prefer to call them) to determine the traits of a role’s optimum performers. It then uses its analytics to uncover candidates who match them.
It works like this: An employer hires Pymetrics to “create bespoke profiles of success,” in the words of CEO Polli. Basically, the employer has successful workers run through Pymetrics’ set of games to establish what traits are common to each role. Those are then used to screen both internal and external candidates during the first-pass review of applicants, where Polli says, “a lot of bad decisions are made.”
For example, the company’s web site says planning efficiency, the ability to read facial expressions quickly and being disciplined in thought when deciding to spend money in new situations are all distinguishing traits shared by strategic planners. Being somewhat easily distracted, not unreasonably chasing success against the odds and depending on more than facial expressions to read emotions are traits shared by writers. Being emotionally influenced by a situation’s context, fast planning and being able to learn under high-risk conditions are traits shared by HR professionals and recruiters. (At first we thought that being easily distracted was counterintuitive as a writer’s trait, but then we thought about how often curiosity can lead journalists away from one path and onto another one that will more efficiently help them reach their goal.)
Correcting Mistakes Up Front
As Pymetrics sees it, many failed hires could be avoided if inefficiencies and biases were removed early in the process. According to Polli, today’s average job attracts about 250 resumes and recruiters spend six seconds giving each one an initial scan. Not only is the process inefficient, it allows each reviewer’s biases to influence the decision on who does or doesn’t make the cut. The result is that 30 to 50 percent of new hires don’t work out, Polli says. That means the job must be posted again once the employee departs, another 250 resumes come in, are subjected to the same fast, furious and biased review, and the employer faces the same odds of an unsuccessful outcome.
“We think you can have the best impact by making sure recruiters see the candidates who are best-aligned with their company’s profiles of success,” Polli says. “Remember, we don’t look at resumes—we look at the traits that contribute to success in different functions at each company.” Pymetrics’ approach, she contends, “ends up being more predictive than a regular resume review.”
With 30 employees in place, and backed by Series A funding of more than $6 million, the company plans to open offices in the UK and the Asia-Pacific region in the near term. Noting that Pymetrics operates in “a space a lot of people are interested in right now,” Polli sees the need to build up her sales team to accelerate growth.
Pymetrics had a big win in 2016 when the consumer-products giant Unilever incorporated its technology into its global (?) hiring process. Rather than recruit in old-fashioned ways–attending college career fairs and the like–the company began using Pymetrics to conduct initial screenings. Candidates who make it through the screen are then video-interviewed using HireVue, whose technology analyzes keyword use, body language and other factors to generate a report for the hiring manager.
According to Business Insider, between July 2016 and June 2017 the new approach doubled the number of job applications in North America within their first 90 days of posting, saw a “significant” increase in the hiring of diverse candidates and universities represented, cut the average hiring time from four months to four weeks, and decreased the time spent on resume reviews by 75 percent.
That last point is significant to recruiters. As headhunters repeatedly say, they’d prefer to spend more time getting to know candidates personally than sorting through resumes since that allows them to make better matches. By eliminating the chaff at the top of the funnel, tools like Pymetrics allows them to look more carefully at a candidate’s work and social media profiles—whether they’re on LinkedIn, GitHub or somewhere else—develop longer-term relationships and, in general, do a more time- and cost-effective job.
One last note: Pymetrics aims to do more than streamline a company’s recruiting process. Polli says its tools can be used to help organizations identify promising internal candidates, for one thing, and it envisions building out profiles to help employees plan for new careers or job paths if they need or want to make a change.
For candidates, the technology can “put you on path you haven’t thought of before,” Polli says. For example, many traits of attorneys and software engineers align, in that “both are detail-oriented and read a lot of complex documents.” By building out a database of such traits and aligned roles, Pymetrics could build a trait-based job classification system. To that end, the company is compiling its own data set as well as incentivizing its clients to share theirs.
- Hiring Managers Want More Data-Driven Recruiting
- Restless Bandit Wants to Automate ‘Talent Rediscovery’
- Q&A: Integrating Assessment Tools, ATS Allows For Faster, Better Hiring