HR Management & Compliance, Technology

Flight Risk Models: How to Avoid Analysis Paralysis and Take Action

Flight risk models have gained in popularity and can certainly be useful as these models boast the ability to predict which employees will stay and which will leave. The caveat is that a flight risk model alone is a tool, not an approach. And the key to success is to act on the data at the local level as swiftly as possible to mitigate turnover.

Source: ProStockStudio / shutterstock

There is no perfect flight risk model and proactively getting out in front of an organization’s turnover is much more important than over-tweaking the model and suffering from a case of analysis paralysis.

To get to a place of action, human resource (HR) leaders must first understand what a flight risk model is, best practices, and dangers.

What is a Flight Risk Model?

A flight risk model is an advanced statistical model that predicts likely future turnover. The model incorporates as much data as possible relevant to what might make an employee voluntarily leave an organization. Organizations can subsequently take action to reduce turnover in the future, based on the insights uncovered by the model. The model should be simple, easy to read, and highly actionable.

Flight risk models can reveal many insights:

  • The causes/drivers of turnover;
  • Specific at-risk areas of the organization (e.g., leaders, jobs, functions);
  • Potential actions the organization or leaders can take to reduce turnover; and
  • The cost of turnover.

Here is an example of a flight risk model:

Flight Risk Model: Considerations and Watch-Outs

An organization must first make sure that it takes an analytics approach that will deliver credible output. Relying on correlations won’t cut it. Logistic regression, random forest, decision trees, and other more rigorous methods will make sure that the key drivers will truly impact turnover. Weak analysis is rampant. Many firms are still just using cursory and correlational analyses to uncover drivers of turnover, or they examine only a few people who turned over to create a “profile.”

Next, an organization must ensure it’s able to show the dollar impact. Leaders want to see the impact of a flight risk model. The statistics may be interesting, but understanding not only the cost of turnover but also the downstream impact of turnover (how much is turnover directly impacting sales, customer satisfaction, etc.?) will drive action from the flight risk model. In addition to these dangers, oftentimes flight risk models end up being information that is examined by a few leaders or just by HR and no real action comes from them (i.e., analysis paralysis)—more on this below.

Beware of social media-scraping. This piggybacks on the weak analysis in that firms will claim that they “know” the couple of factors that drive turnover (e.g., updating a LinkedIn photo). The danger in this is not the data but the effects on employees’ trust. Will the organization let employees know that it is doing flight risk modeling at all—even if it is only with internal data? Is leadership willing to tell the employees this?

If organizations are allowing frontline managers to see flight risk information on individual employees, they can react in different ways and, unfortunately, these reactions might be inappropriate. For example:

  • A manager may get angry at an employee for being a flight risk and take him or her off the high-potential (HiPo) employee list or pull him or her out of developmental programs.
  • A manager may begin to treat the employee negatively because he or she feels betrayed by the employee for being a potential flight risk.
  • A manager may be unnecessarily generous to an employee and give raises or make immediate promises to a quality employee who might be a potential flight risk.

It’s also important to get clarity from the very beginning of the flight risk project of who will see the data (HR only? All leaders?).

Analysis Paralysis vs. Action

This is where most flight risk models come apart. The models are overanalyzed and agonized over, but then nothing happens. Based on the findings, several actions can be taken. A great first step is conducting stay interviews within groups with high flight risk. Stay interviews are a retention tool in which a manager conducts a one-on-one interview with employees in these high-risk areas to uncover what he or she does and doesn’t like that may lead to turnover. Additional follow-up actions in high flight risk areas include the following:

  • Online Training. Conduct training for leaders focused on factors that have been linked to flight risk. For example, if teamwork and career development were found to impact turnover, online training modules can provide learning and action lessons to drive behavioral change.
  • 360 Leader Assessments. For leaders of areas of high risk, launch 360 assessments that focus on competencies and skills to mitigate turnover and provide a better experience to their employees.
  • Pulse Surveys. Use these to gauge employee experiences linked to turnover to drive action planning as well as to check in on the progress of action plans that are currently in place.
  • Hiring Process. Evaluate the hiring process to examine if new hires are good fits for the organization. Make sure to have the proper structure and rigor in the process (e.g., behavioral-based interviews, role plays, fit assessments, cognitive ability assessments). In other words, stop turnover before it starts.
  • Onboarding and Quality of Hire Surveys. Measure and connect the new hire experience to early turnover and flight risk. Turnover is typically focused within the first year, so improving the initial experience with the organization and identifying hires who need additional attention is key.

Below are specific actions that can be taken based on the outcome of the flight risk model diagnosis:

When done correctly, flight risk models are essential in HR’s toolbox for adding value and bottom-line impact to the organization. In the end, a model can provide valuable insights into the causes and costs of turnover, as well as identify specific organizational areas with the greatest risk of turnover. However, to be successful, an organization must have confidence in the model itself (why rigorous analytics are necessary) and be diligent in its plans regarding what will be done with the information.

Shane Douthitt, PhD, is the Cofounder and Managing Partner of Strategic Management Decisions (SMD). More than 25 years in the areas of measurement, talent management, executive assessment and coaching, and organizational development at global Fortune 50 companies provided  Douthitt with invaluable experience, but more importantly, fueled a  desire to infuse innovation into a field (HR) that had been stagnant entirely too long. Focused on removing the divide between HR and return on investment (ROI), he can be described as an out-of-the-box thinker, analytics expert, and technology enabler.