We used to assume that the best salespeople were extroverted, but research shows the biggest predictor for sales success actually is persistence—the courage to keep going despite initially being told no. The findings, based on millions of worker surveys and tests conducted by I.B.M. and reported in The New York Times, highlight a pivotal point in HR. No longer relying solely on gut instincts, human resource managers are turning to Big Data and analytics to hire, fire and promote employees.
Called “workforce science” or “people analytics,” the marriage of HR and analytics presents unprecedented opportunities for employers to gain a better understanding of how their staff operates. “It used to be that only really large organizations like Home Depot were good at analyzing their workforces,” says Matt Stevenson, workforce analytics and planning practice leader for North America at Mercer Workforce Sciences Institute. “But now the technology has become democratized and it’s a lot cheaper for smaller companies.” As more organizations incorporate Big Data and analytics into HR efforts, they'll need to understand the right—and wrong—approaches to workforce science.
Starting points for an analytical approach
The simplest approach to workforce science is what Stevenson and his colleagues refer to as hiring source analytics. Companies recruit using a variety of different methods: referrals, web ads, career fairs, etc. Using various performance review metrics, they can go back and analyze, for example, whether employees hired from a web ad tend to perform better than peers who came from referrals. Do they get promoted faster? Do they stay longer?
It’s important that companies control for certain factors when conducting these analyses. For example, people who apply for jobs through web ads might fill entry level positions, while people who are brought in through recruiters might be more senior. Comparing the job experiences of the two camps would result in a false comparison.
Another way companies incorporate workforce science is by testing people before they start a job. Selection tests are nothing new. The U.S. military used selection tests for soldier placement during WWI. “What’s different now is you can go back and see how well this test actually predicts someone’s performance,” he says. Companies test their existing employees to understand the skills that contribute to top performers' success, and those results help them set a baseline for testing future hires.
For as little as $30 per employee, companies can conduct a very simple online test through various vendors, according to Stevenson. One of the biggest benefits of this type of testing is that managers can look at more than one factor. “It’s no good if you predict who’s going to be the best seller, if you can’t predict whether that person will stay. You can create these selection criteria that ask and test for a bunch of things that you care about,” Stevenson says.
Beware the caveats
Before managers jump on every shiny data science model for hiring, firing and promoting, they should understand that the discipline is not without flaws. Wharton researcher Peter Cappelli told The Economist about a case “where the software rejected every one of many good applicants for a job because the firm in question had specified that they must have held a particular job title—one that existed at no other company.”
Overreliance on algorithms means that employers risk overlooking some factors in their decisions. “We’re getting to the point where some of our hiring managers don’t even want to interview anymore,” Teri Morse, vice president for recruiting at Xerox Services, tells The Atlantic. According to the article, “They just want to hire the people with the highest scores.”
In the case of selection tests, Stevenson cautions that human resources managers aren’t getting the full picture: they never get to see how people who they didn’t hire perform. “Your selection test may predict how well people will do in your company based upon the people you already have. But what if the people you have aren’t as good as other people?” he says.
The future of workforce sciences
Workforce sciences are more relevant in some industries than others. A company that designs nuclear power plants, for instance, is looking for incredibly specialized people and will get more value out of testing to find the right match than say, McDonald's, where people join and leave daily, Stevenson says. As the labor market tightens, however, looking at hiring from all angles—including a statistical standpoint—will become more valuable.