ZeroedIn is providing critical intelligence to frontline managers in the form of predictive talent indicators. One of those indicators predicts flight risks – valued employees who leave a company voluntarily. With the insights from ZeroedIn, Human Resources (HR) professionals can identify potential flight risks and save their company enormous amounts of time and money in talent replacement costs.
Extracting Critical Commonalities From Historical Data
ZeroedIn relies on a company’s untapped stockpile of employee data. That critical historical data helps our technology identify commonalities among valued employees who quit their jobs. We’re also interested in those who choose to stay. These commonalities and differences are the inputs that our flight risk algorithm uses to learn and identify employees who are most likely to leave your company unexpectedly, and those who are not.
Are those who stay and those who leave polar opposites? No, not usually. It’s more important to identify the similarities among one group and zero in on which of those characteristics are the most critical. Some of the factors we identify as critical to flight risks are personal – things like education, commuting time, use of childcare – but most speak to activities within the company: movement, performance, and employee engagement, for example.
We typically do not factor age, gender or ethnicity into our models, and we also caution a company not to look too closely at these. There are legal issues of discrimination involved and some companies might be inclined to profile one group that tends to stay employed the longest and just hire members of that group (often, “people like me” as explained in our Diversity blog).
Putting Flight Risk Prediction To The Test
In a recent study with a 56,000+ employee healthcare service provider where 4,680 employees separated during the year (8% turnover) and 51,348 did not separate, ZeroedIn’s flight risk algorithm identified employee staying power at a remarkable rate of 99.75%! Details of results are as follows:
- ZeroedIn’s algorithm predicted 1,371 employees correctly as separating (29%).
- ZeroedIn predicted 125 incorrectly as separating who stayed (2.6%).
- ZeroedIn predicted 51,223 correctly as staying (99.75%).
- ZeroedIn predicted 3,315 incorrectly as staying who separated (6.4 %).
Our predictive model produces a probability score for each employee. Using the scores, ZeroedIn assigns ranges such as low, medium, and high probability. Sometimes there are different thresholds within those ranges, creating more defined scores as the algorithm continues to learn from the variables that influence employee flight risk and staying power.
Based on the scoring we provide, HR now has the intelligence to segment the employee population and make good decisions to reduce flight risk among valuable employees. Those employees should get extra attention and effort to encourage them to stay. Other flight risk employees who are not deemed as valuable to the company might be allowed to leave on their own, saving the company the cost of firing them and paying compensation.
A Word Of Caution About Sharing Employee Probability Scores
Scores that identify potential flight-risk employees can be great information for HR. But we want to caution anyone against just handing that list to people managers. As President Franklin Roosevelt said, “Great power involves great responsibility.”
One way that HR can mitigate unintended outcomes is to ask managers, “Who are the most critical employees on your team?” With a list of people in critical roles, HR can identify which are flight risks and work with management to make decisions on how to talk with those employees. In this way, management’s actions are much more targeted and thought-out.
Should You Consider Flight Risk With Job Candidates?
ZeroedIn doesn’t apply flight-risk scores to job candidates – we take a different approach to identifying the perfect match for those critical open job requisitions. Given that we’ve identified factors that lead to flight risk though, there is value in recruiters watching out and listening for them when trying to attract the best talent. We’ll talk about predictive indicators for recruitment in a future blog post.
