Predicting Employee Flight Risk: My Take

Greta Roberts, CEO, Talent Analytics, Corp.

Greta Roberts | Talent AnalyticsIt’s exciting to watch advances in predictive and prescriptive employee solutions. Workday recently announced the release of an application enabling employers to “identify which employee is likely to quit, and what options need to be considered to retain that person”.

Workday is not the first to announce Flight Risk Scores of current employees. Many top Talent Management solutions have made similar announcements in the past several months. It’s a step in the right direction.

As exciting as these announcements sound, I wanted to tease apart some of what we have learned that matters to businesses and their employee decisions. Perhaps frame how to interpret current innovations coming from the Talent Management industry.

Is “Focusing Only on a Small Subset of Top Performers the Right Focus?”
Businesses and their Talent Management vendors seem to be obsessed about a small subset of top performers they are afraid of losing. Meanwhile thousands of bottom performers are hired, weigh down our organizations, turnover quickly and cost our businesses millions.

Of course businesses need to attack the problem from both sides. The most costly expense to mid to large businesses are frequently high volume, high turnover roles where thousands of employees are hired only to leave quickly or be a bottom performer.  In both instances these new employees incur a massive cost to the business before providing any value to the employer.

Flight Risk Scores for Current Employees is Too Late
They are already hired. They are on-boarded.  They are costing the organization without delivering value. A better approach would be to predict how long a job candidate will stay, before you hire them avoiding hiring those with a high flight risk altogether.

Employee Cost and Performance Curves | Talent Analytics

Figure 1: Turnover Prediction is a Process not a Single Score

Flight Risk is not a Single Score
Employees in a specific role have different flight risks at certain times.  Like saying the probability of machine failure is 78%.  There is a missing piece of important context, which is “time”.

For a machine the probability of failure might be 32% at 6 months, 51% at 1 year and 78% at 2 years and so on.  This Survival Curve gives the most accurate view of Flight Risk and enables the organization to be more sophisticated about their decisions.

Understanding that flight risk is a curve instead of a single point in time allows the organization to pinpoint when intervention is optimal and can save the most employees and the most money.

Talent Analytics, Corp., has a pre-hire predictive recommendation engine, Advisor, with advanced predictive pre-hire analytics. Call +1-617-864-7474 or watch a recent webinar to learn more.

Posted in Tags: , , , , , , , , ,

2 Responses to “Predicting Employee Flight Risk: My Take”

  1. Andy Maddocks Says:

    April 15th, 2015 at 7:43 am

    Thanks for sharing your high level insights. I can assure any new readers of your posts that your conclusions are very data-driven and that you put a lot of care into formulating insights and suggestions.

    As you suggest, not paying attention to the larger middle tier of employees in large organizations could be even more costly in the long run than just worrying about getting and keeping the right ‘top performers.’

    Personally, based on research I’ve both conducted (some years ago) and more recent studies I continue to follow, their may be important individual differences (i.e., motivation and other ‘psychographics’) that could help employers attract and retain strong mid-tier employees that are likely to work well with (or for) top performers. That appears to be an area where many Talent Management solutions are weak, yet it’s a place where careful scientific approaches using large scale data sets could really pay off for large enterprises.

    I’m sure you’re working on that, even if other TM solution providers are not.

    Thanks for your post.

  2. stewart hu Says:

    April 15th, 2015 at 6:12 pm

    Nice post Greta. Another metric that is making a lot buzz in the HR analytics world is the J-score from Joberate, which I joined this week. We build machine learning algorithms to predict job seekers’ real time behavior based on their public digital footprint, taking advantage of a lot leading indicators to make our clients stay ahead of curve. Also our algorithms are constantly evolving to adapt to the changing trends in the job markets.

Leave a Reply