Submitter
Dan Pikelny, Vice President and Chief Analytics Officer
Company
Navistar
Industry
Manufacturing
Business Challenge
Navistar has been in a turnaround effort for several years and relies on its workforce to accomplish it. Of course, as in many organizations, the financial challenges have been accompanied by staff cuts and lack of variable pay, leading employees to look for alternate employers. While both leadership and the human resources organization have worked to create an effective employee value proposition, the challenge is where to direct their limited resources.
Analytics Solution
The analytics team developed a predictive model for human resources for employee retention. The team had an opportunity to use many types of data, including information such as websites viewed from work or arrival time in the morning, but decided to work exclusively with data that (1) employees would not feel compromised by in order to ensure transparency and (2) the organization could impact. Navistar developed an algorithm to predict which employees were likely to leave the company in the next 12 months, calculating a probability of voluntary departure for each employee. While testing more than 60 explanatory variables, the ultimate model included tenure, compensation, demographics, networking effect (contagious quits with the manager/business team), and proximity of workplace to home. The model achieved nearly 80 percent on an ROC curve and was also validated on data that was held out from the algorithm building.
Impact
By using the model, the human resources team identified the employees in each function who are at highest risk and is working with managers to focus on the highest performance and highest potential team members. The goal is to cut the voluntary departure rate for identified employees through proactive programs such as flexible work arrangements, moving employees across managers, and selective progressions. The result is a faster turnaround for Navistar with lower turnover costs. In addition, the predictive model now also informs hiring decisions to help identify candidates most likely to become long-term valued team members. The analytic work has helped the human resources organization align leaders on how to use employee data as an additional tool.