How To Predict Which Of Your Employees Are About To Quit

You’ve got more data on how your team members are behaving, thinking, and feeling than you probably realize. Here’s how (and why) to tap into it.

How To Predict Which Of Your Employees Are About To Quit
[Photo: shironosov/iStock]

“People analytics” may sound daunting, expensive, and difficult—something the ordinary manager can’t possibly concern herself with even if she’d like to. But the field isn’t necessarily as high-tech as you might imagine.


There’s more untapped data, of some kind or another, floating around your workplace than you probably think. With a little extra effort to spot behavioral patterns, you may be able to get ahead of some of the more common issues, like employee attrition, that can hurt your workplace and your organization’s bottom line. Here’s how.

Phoning It In

Turnover tends to be high at call centers, where many people take jobs temporarily, then quit when once they’ve earned enough to return to school or cover a big expense. Lower attrition means higher performance, so managers are interested in predicting and reducing attrition.

My company helped one call center analyze some basic data that it was already collecting: the length and number of calls operators were taking, and how often those calls got escalated or resolved. At the end of each shift, employees received a “report card” reflecting those data points. Since the call center employees’ compensation was linked directly to that performance data, they were highly incentivized to earn good marks.

But a low overall score wasn’t necessarily a sign that an employee was performing poorly, getting paid less, and therefore planning to bounce. Analysts found two specific factors were much more predictive: increased time spent on calls, and fewer calls ending in resolutions. Those operators were just going through the motions.

So the call center’s managers sent supervisors to meet with each operator within a day of those two indicators popping up. Most, however, hadn’t yet reached a point where they were considering quitting. But they often did reveal job frustrations that were usually easy to address, a like a faulty headset or having to work an undesirable shift. Supervisors were empowered to fix most of these problems, and over the next few months, the call center’s attrition rate fell by half.

Feelings And Actions You’re Not Picking Up On

“Sounds great,” you might be thinking, “but I don’t run a call center.” Even so, you can probably start looking for small, early signs of dissatisfaction that are relatively easy to remedy once you spot them. Here are two:


1. Ask employees how they’re feeling–continuously. Measuring “perceptions” might seem impossible, but it’s not. To collect data on something like this, you can use pulse surveys, run focus groups, or take snap polls using common Slack integrations like Polly.

Some large, physical office spaces even go analog and install those sentiment buttons you might have seen in airports or hotels. They’re simple, inexpensive devices that ask a question like, “How was your day?” and provide red (bad), yellow (okay), and green (good) buttons for people to press quickly as they go about their day. Whatever method you use to gather sentiment data, aim for something easy and anonymous, and watch for trends, not absolute values.

2. Look for dips in hours worked or effort spent. A basic place to start is total login time, but unless your office requires workers to “punch in” or “out,” introducing software to monitor exactly who’s sitting in front of their computers when can feel like surveillance. So start with the data you’ve already got on hand but may not be analyzing fully: How much sick leave is being taken this quarter, compared with last quarter or with the same quarter the prior year? How much annual leave is being requested (regardless of what’s actually granted)?

These are usually good indicators of who may be on their way out. Sick days can be requested to attend interviews or to burn up unused leave balances—or maybe that person is just feeling burned out and needs to take some mental heath days to deal with on-the-job stress.

The LinkedIn Trick

There’s a third method, too, that I’ve seen work wonders. A well-known tech firm that recently worked with my company was losing its precious engineers. Recruiters who spent a lot of time looking for coders on LinkedIn were already in the habit of noticing recently updated “Skills” sections, interpreting that as a sign an engineer might be interested in hearing about new opportunities. So it occurred to the tech company to apply this principle in reverse.

The managers realized that their own coders were probably doing the same thing–updating their LinkedIn profiles whenever they were ready to hear from other firms. So the company wrote a simple script to capture the LinkedIn update feed for the profiles of around 2,000 of its top-performing coders. That let managers to react quickly whenever one of those employees added new info. Similar to the call managers, supervisors then swooped in to discuss the career goals and professional-development opportunities with the coders who might be wavering.


As a result, turnover fell, and many of those engineers were moved to assignments or projects that suited their talents and interests much better.

Use Your Data Wisely–And Fast

Whatever patterns you decide to watch, make sure you’re gathering data for two weeks to two months, so you’ll have enough information to perform a reasonable analysis.

But once you do spot a certain trend, don’t wait to act. Start looking for the source of the dissatisfaction in the corner of the company where you’re picking up on it. Maybe a certain team just really needs flex schedules or better recognition, or they feel starved for information. Often the most effective remedies aren’t even monetary. Once you’ve determined a solution, measure its effectiveness to make sure it continues to produce the outcome you’re hoping for.

At the end of the day, most employees all want the same basic things. Done right, people analytics starts from that humane premise and doesn’t reduce people to numbers–it just helps companies understand why certain situations cause people to keep behaving in certain ways. Ideally, it’s good for everyone when there are fewer surprises, and there’s more happiness to go around.

About the author

Anne Loehr is a sought-after keynote speaker and the executive vice president of Center for Human Capital Innovation, which provides leaders with valuable tools, training, analysis, and innovative insights to advance the science of people management and improve organizational performance across all sectors. Follow Anne on Twitter at @anneloehr.