-
Tweet
-
Post
-
Share
-
Save
-
Get PDF
-
Buy Copies
-
Print
Paul Garland
Imagine that you’re looking at your company-issued smartphone and you notice an e-mail from LinkedIn: “These companies are looking for candidates like you!” You aren’t necessarily searching for a job, but you’re always open to opportunities, so out of curiosity, you click on the link. A few minutes later your boss appears at your desk. “We’ve noticed that you’re spending more time on LinkedIn lately, so I wanted to talk with you about your career and whether you’re happy here,” she says. Uh-oh.
It’s an awkward and Big Brother–ish scenario—and it’s not so far-fetched. Attrition has always been expensive for companies, but in many industries the cost of losing good workers is rising, owing to tight labor markets and the increasingly collaborative nature of jobs. (As work becomes more team-focused, seamlessly plugging in new players is more challenging.) Thus companies are intensifying their efforts to predict which workers are at high risk of leaving so that managers can try to stop them. Tactics range from garden-variety electronic surveillance to sophisticated analyses of employees’ social media lives.
Some of this analytical work is generating fresh insights about what impels employees to quit. In general, people leave their jobs because they don’t like their boss, don’t see opportunities for promotion or growth, or are offered a better gig (and often higher pay); these reasons have held steady for years. New research conducted by CEB, a Washington-based best-practice insight and technology company, looks not just at why workers quit but also at when. “We’ve learned that what really affects people is their sense of how they’re doing compared with other people in their peer group, or with where they thought they would be at a certain point in life,” says Brian Kropp, who heads CEB’s HR practice. “We’ve learned to focus on moments that allow people to make these comparisons.”
Some of the discoveries are unsurprising. Work anniversaries (whether of joining the company or of moving into one’s current role) are natural times for reflection, and job-hunting activity jumps by 6% and 9%, respectively, at those points. But other data reveals factors that have nothing directly to do with work. For instance, birthdays—particularly midlife milestones such as turning 40 or 50—can prompt employees to assess their careers and take action if they’re unhappy with the results. (Job hunting jumps 12% just before birthdays.) Large social gatherings of peers, such as class reunions, can also be catalysts—they’re natural occasions for people to measure their progress relative to others’. (Job hunting jumps 16% after reunions.) Kropp says, “The big realization is that it’s not just what happens at work—it’s what happens in someone’s personal life that determines when he or she decides to look for a new job.”
Technology also provides clues about which star employees might be eyeing the exit. Companies can tell whether employees using work computers or phones are spending time on (or even just opening unsolicited e-mails from) career websites, and research shows that more firms are paying attention to these things. Large companies have also begun tracking badge swipes—employees’ use of an ID to enter and exit the building or the parking garage—to identify patterns that suggest a worker may be interviewing for a job. Companies sometimes retain outside firms, such as Joberate, to monitor employees’ social media activity for indications that people are scouting for new options. (Among other things, such firms look at whom employees are connecting with.) Joberate CEO Michael Beygelman compares this emerging science to the way that credit scoring can predict which consumers will fail to repay loans. Although some companies hire Joberate to help them anticipate which individual employees might think of leaving, others use the intelligence to zero in on departments or locations with high “likely to leave” scores so that they can work on team building and overall engagement. One large tech company uses it to target people it might lure away from other firms. Some investors use it to identify companies that may soon face turnover in key positions. “If the CIO and the head of sales are both likely to be job hunting, you have to ask what’s up,” Beygelman says.
Lori Hock, the CEO of Hudson Americas, a recruitment process outsourcing company that uses Joberate, values predictive intelligence because it helps her reduce clients’ attrition—and spot things that may be driving it. “Is it a bad manager?” she says. “Is there a training component? Are we undervaluing certain positions? It gives you a nice opportunity to think about what the trigger might have been—and to ask questions before you lose talent.”
Genevieve Graves studied astrophysics before joining hiQ Labs, a start-up that applies predictive analytics to talent management. She says the fields aren’t as dissimilar as they sound: “Most of the techniques I used as an astronomer—machine learning, scientific computation, large data-management tools—directly translate, but now I study people instead of galaxies.” She spoke with HBR about the emerging science of predicting attrition. Edited excerpts follow.
What does hiQ Labs do?We use public and internal company data to predict turnover risk. We also provide tools to help with skill mapping, succession planning, internal mobility, and career development. It used to be about retention, but now it’s also about getting the most from the employees you have.
What public data do you look at?We read online résumés and profiles. We consider an employee’s social media footprint, which indicates visibility to recruiters. For instance, is an engineer participating in open-source code projects? We look at work histories (to get a sense of how frequently an employee changes jobs) and at the opportunity landscape, meaning how much demand there is for a particular employee’s skills. These things don’t necessarily mean someone is job hunting—they just indicate recruiter attention. People trying to predict attrition often think of “push factors” that make people want to leave their jobs, but public data can point to “pull factors” that indicate recruiters might be wooing someone who’s not actively looking.
What do managers do with the information?Once they know which people to worry about, they can have check-in conversations. Do employees find their work challenging and interesting? Do they see a clear trajectory? (For knowledge workers, attrition usually isn’t driven by compensation.) The data is an early-warning signal that lets managers intervene.
Why do employers find this valuable?Attrition can be expensive, especially with knowledge workers. We focus on industries such as finance, technology, pharma, and biotech, which have high-value contributors. A lot of company knowledge walks out the door when an employee leaves. You can hire somebody with the same skill set, but it takes months for that person to get up to speed. And when managers leave, they may take a whole team with them.
Some firms, such as Credit Suisse, take this tack with employees identified as being at risk of leaving: Internal recruiters cold-call the employees to alert them to openings inside the company. In 2014 the program reduced attrition by 1% and moved 300 employees, many of whom might otherwise have left, into new positions. Credit Suisse estimates that it saved $75 million to $100 million in rehiring and training costs.
Researchers agree that preemptive intervention is a better way to deal with employees’ wandering eyes than waiting for someone to get an offer and then making a counteroffer. CEB’s data shows that 50% of employees who accept a counteroffer leave within 12 months. “It’s almost like when you’re in a relationship and you’ve decided you want to break up, but your partner does something that makes you stick around a little longer,” Kropp says. “Employees who accept a counteroffer are most likely going to quit at some point very soon.”
A version of this article appeared in the September 2016 issue (pp.20–21) of Harvard Business Review.