How Big Data is Changing the Recruiting Game

Big data — and its potential to help companies predict which employees are likely to flourish or flail, whether as new hires or future leaders — is all the rage among HR departments today. For some, the conversation is in its infancy, say experts who have been privy to these discussions ("What, exactly, does big data mean? Do we have it? How can we get it?"). For others, the talk has evolved into, "How do we get top executives to buy-in the same way they've embraced big data in sales and marketing?"

Still others — a scant 4 percent, according to 2013 data from Bersin by Deloitte — are actually using big data and predictive analytics in ways that truly impact their workforces. The vast majority of these companies are large, multinational organizations with an employee base where turnover can run high, and they're using big data to help spot the best recruits. (Think industries like retail or hospitality, which depend on armies of in-store salespeople or operate massive call centers, or high-tech and financial services, where data and analytics are already part of the companies’ DNA.)

What's most exciting about data-driven insights, these companies are finding, is that they're often surprising or counterintuitive — such as the discovery one company made that Ivy League graduates who once toiled in minimum wage jobs perform better at work than those who didn't. John Sullivan, a Silicon Valley talent management expert, estimates that these Moneyball-style hiring decisions are about 25 percent more reliable than those based on human intuition.

Xerox estimates it spends $5,000 to train every new call-center rep. After collecting and analyzing performance data on these early hires, the company discovered that some assumptions it had been making about job recruits were false. For example, candidates with experience in call centers or other relevant positions cost more, but they didn't perform any better than those without experience. More surprisingly: Workers who are active on up to four social networks were more likely to stay in their jobs.

Sears has also stepped up its screening game. The giant retailer hires up to 160,000 new sales representatives per year from an applicant pool of about 6 million, according to a report last summer in Briefings magazine. To better identify the right workers, Sears now has applicants complete a video game-like test that includes simulated interactions with a variety of customer types, from the overly demanding to the indecisive.

A few years ago, Wells Fargo set out to do a better job of sussing out qualified candidates who are also likely to stick around. By analyzing data on current workers and devising personality tests that aren't easily manipulated, the banking behemoth discovered that tellers and other front-facing workers with accounting degrees were top performers, but didn't last long in their jobs, according to a report in BAI Banking Strategies.

The most effective analytics, says Michael Housman, chief analytics officer here at Cornerstone OnDemand, rely on millions — even tens of millions — of data points. For example, companies will collect data on current employees' job performance every single day they've worked on the company clock, and combine that with results of pre-employment tests and macroeconomic trends on job growth or population shifts. Software algorithms then identify meaningful patterns that can be used to assess potential new hires. Applicants, in turn, are given a score indicating whether the company should move forward based on myriad factors, including career history and personality testing.

The analysis, of course, is easier said than done. Companies need to have a sufficiently large pool of employees and data collected over time in order to gain reliable insights. This means that predictive analytics, for the most part, doesn't work for smaller companies. "Creating a personality test and testing it against 200 or 300 high performers isn’t really big data analysis," says Housman. "That’s the way things have always been done." Similarly, it can be costly and time-consuming to track down historical data that's rarely consistent and can be carefully guarded by individual departments.

"Everyone is talking about big data," Housman reminds us. "But it’s still really hard to do."

Photo: Shutterstock