Replacing employees who leave their jobs is costly and time consuming for businesses. Companies have tried many strategies to address employee retention over the years, with varying results. Now new research suggests that mathematical models may hold the key to understanding which employees will remain in their jobs.

Algoriths, mathematical formulas that can process data and report results based on pre-defined parameters, are being used by recruiters and personnel staffing companies to identify workers who are more likely to fit with hiring companies. In the hiring field, some companies are now specializing in tailoring hiring algorithms to companies who enlist their help in hiring, according to Bloomberg.

For example, the hiring-algorithm company Pegged was brought on by the giant hospital chain Adventist Healthcare to help identify candidates who better fit the company’s culture. Adventist had been hiring experienced candidates with top credentials, yet turnover was still above the market average.

To construct their hiring algorithms, Pegged uses data from public records, including from Google searches, information from candidates’ resumes, and data from candidates’ job applications and interviews. The algorithms can include anything from how long a potential employee spends on a webpage to measuring the candidate’s keystrokes. Once Pegged has narrowed the field of potential employees down to those most likely to succeed in a position, hiring managers make their final selcctions.

The Pegged models for Adventist took some time to pay off, but eventually turnover at the company went down by as much as 50 percent. Pegged says that the median effect on retention from its hiring assistance is a 38 percent jump in retention over a 6- to 12-month period. In addition to assessing candidates, Pegged reviews personnel who have recently left or been fired from a company, to help determine what kinds of employees have success in the company.