Employee turnover has always been a pain point for many organizations. Just when you have put in the time and investment to ramp up a new hire to their role, they leave. It seems like a vicious cycle that never ends.
What if you could use data to combat this issue? Could you learn anything from the data collected about the individuals to find a solution? The answer is: it depends. (Just like any other questions my professors in graduate school would say.)
Let's take a look at some case study examples of how organizations can adopt a data-driven approach to retain talent.
Case Study 1
An organization collected thousands of resumes during the hiring process for various entry-level manufacturing positions. They wanted to streamline the process of the resume review as well as devise a standard way to screen candidates based on their probability to turnover. A team reviewed more than 500 resumes and applications for which about 400 individuals have turned over whereas another 100 individuals have not. The review revealed several themes of interest. The team was able to quantify the resume data into something more palatable. A number of different factors were evaluated.
Some example factors include:
Does the employee live within a 30-mile radius of the workplace? (yes or no)
Did they go to trade school? (yes or no)
Were they referred from job centers? (yes or no)
Were they referred from other employees? (yes or no)
How much was their salary expectations? Was it greater than $3,000 per month?
Did they work for the employer previously? (yes or no)
Were the references provided on the resume work related or personal?
Was their last job prior to joining this company in an industrial manufacturing environment? (yes or no)
How many jobs did they have in the past 5 years?
Was the resume sloppy and incomplete? (yes or no)
The data were analyzed to identify factors that would relate to the likelihood of turnover. For instance, the data suggested that employees who lived outside of the 30-mile radius from the workplace were more likely to turnover. This seems to make sense as a longer commute means more expenses and stress for the employee, both of which could lead to turnover. A Turnover Risk Index comprised of multiple factors was developed. A statistical correlation analysis suggested that this index is significantly related to the employment status two years into their employment with this organization.
Additionally, it was estimated if the organization were to use this index to screen out candidates, they could have reduced new hire turnover by 15%. Based on these results, the organization implemented a standardized resume screen to remove candidates with high turnover likelihood early on in the hiring process.
Case Study 2
An insurance company implemented a selection system to select the best candidates into two of their contact centers as customer service agents. Part of the selection system included an online application.
A series of analyses were conducted to identify application questions that were related to employment status. There are some consistent findings across the two call centers. Individuals who had a much higher chance to stay with the company after 90 days had:
1 or 2 years of previous customer service experience
1 or 2 jobs in the past 5 years
No history of quitting a job because of interpersonal conflict at work
Expressed interest in part-time work
Reported not being late to work in the past year.
The results suggested that employing a weighted application blank, where points are given to factors differentiating candidates on their turnover likelihood, could potentially reduce new hire turnover up to 17%. Individuals who quit were two times more likely to fail the weighted application blank than individuals who did not turnover. More so, individuals who were terminated involuntarily were 2.8 times more likely to fail the weighted application blank than who those stayed.
Tying it together
Both organizations were able to leverage the data they had accumulated about their candidates to investigate what the characteristics are of individuals who are a high turnover risk. The success of this approach heavily relies on good, quality data and information that can be quantified for statistical analysis.
Equally important, but perhaps less discussed in the world of big data or data-driven decision making, is the fact that not only is there an empirical relationship established in the data, there should be a good theory behind why certain factors are important considerations. Otherwise, it would just be dustbowl empiricism, and may not generalize to future candidate pools.
Keep in mind, however, sometimes the answer might be in the data that you don't currently have. For example, none of these empirical analyses would reveal that the turnover issue might be stemming from a few toxic managers in the workforce. It's often true that people don't leave because of the organization, but they leave because of bad bosses. An organization would likely benefit from a more holistic review of the situation, such as labor market analysis, exit interviews, etc. to drill down the cause and devise a strategy to retain talent.
Now, it’s time to get started on collecting the data that might help your organization reduce costly new hire turnover.