Workers are quitting their jobs at record rates, thanks to a tight labor market and high confidence that moving on will also mean moving up, particularly in pay.
Besides leaving in search of more money, recent job-quitters have also cited toxic work environments and limited room for growth as major reasons for putting in their notice.
For employers, this kind of insight might only become clear during an employee's exit interview, if at all. But with the help of artificial intelligence and machine learning, researchers have developed an algorithm that may be a better predictor of when a worker is at risk of quitting.
That could be a good thing — for employers and employees alike. By understanding who's at risk of leaving, companies may be able to pinpoint the main reasons why employees are seeking other opportunities. If they're willing to invest in retention, they'll avoid the time and financial cost of replacing workers. The flip side, of course, is that employers have to put measures in place to improve employee satisfaction, whether that's in the form of better compensation, workplace culture or career advancement.
Here's how digging into the data could help businesses and workers thrive.
In a recent article for Harvard Business Review, Professors Brooks Holtom of Georgetown University and David Allen of Texas Christian University describe the results of their latest research. Using big data and machine-learning algorithms, the two developed a real-time indicator to measure two main indicators that an employee is about to quit.
The first was "turnover shocks," which are events that prompt workers to consider leaving an organization. This could be a change in leadership or major acquisition, for example, and was measured with events including news articles about a company, changes in stock value and legal action taken against the firm.
Researchers also measured "job embeddedness," or how deeply connected a worker felt to their organization, based on publicly available data like number of past jobs, employment anniversary and tenure, skills, education, gender and geography.
When put to the test, the algorithm identified that those marked as "most likely" to be receptive to a new opportunity were, in fact, 63% more likely to be in a new job by the end of the three-month study period.
Knowing that HR can use this information to determine who's a flight risk might feel like an overstep. There's a financial benefit to re-engaging dissatisfied employees rather than having to hire someone new. By some estimates, the cost of replacing a highly skilled professional could be one to two times their salary, in terms of time and money spent recruiting, selecting and training a new hire, Holtom tells CNBC Make It.
"There are hard costs, like training and selection, as well as soft costs, like reduced customer service or inability to deliver in the short-term when someone leaves," he says.
In a worst-case scenario, 50% of HR managers say they have open positions they can't fill, and extended job vacancies are costing companies $800,000 annually, says Michelle Armer, Chief People Officer at CareerBuilder.
But beyond a company's desire to protect the bottom line, workers could stand to benefit from a more focused approach to employee engagement — so long as employers use machine learning results in a productive way.
"The very best organizations develop a listening culture," Holtom says. "They want to hear from their employees what's going well, what's not going well, and they want to do all that they can to improve conditions for their employees that are cost-effective. AI is complemented by a thoughtful dialogue and one-on-ones or other types of engagement with employees.
"Even if you can predict who's leaving, it still requires you to respond thoughtfully," he says.
Collecting and analyzing these data points can yield better results than gathering answers from what can be time-consuming company-wide surveys. With surveys, for example, it's possible companies aren't asking the right questions, or employees who do opt in to provide feedback aren't being as candid as they could be.
By using available data, "The potential benefit to employees is that, over time, sophisticated managers and employers learn more about what their people really value — not just say they value — and are able to create a more positive and perhaps more personalized work environment," Allen says.
Of course, there are downsides to letting machines predict employee engagement.
"My major concerns are around privacy boundaries," Holtom says. In the study, researchers worked with a talent intelligence firm to gather only publicly available data, including social media feeds. "At some level, that might start to become invasive. Pulling information from Instagram, Snapchat or Facebook — information that people post publicly whether it's controlled or password protected or not — there's risk for employees."
Algorithms could become outdated, be applied out of context or rely on attributes such as protected class characteristics, Allen adds, which could create a host of other issues.
It's crucial to take privacy concerns into consideration, especially as information-gathering in hiring and retention becomes more commonplace.
"The nature of HR and talent acquisition are constantly evolving as we find ways to streamline processes and create better experiences for both employers and job seekers," Armer says, adding that CareerBuilder itself uses AI, semantic search and machine learning to help job seekers create their resumes, match job seekers with open roles and aid other elements of the search and hiring process.
In other words, the researchers' quitting algorithm is just one example of how companies are investing to leverage technology and change the way they source and hire — hopefully, in a more efficient and constructive way that stands to benefit all involved parties.
"The genie is out of the bottle, and I think this is the direction most organizations are going to head," Holtom says. "They're going to collect more data and they're going to analyze it more thoughtfully, more carefully, to increase the odds that they can attract and keep good people."
Like this story? Subscribe to CNBC Make It on YouTube!