“Data science” has become more and more of a trending topic in HR circles lately—what does it actually mean? In relation to HR functions, data science is the application of data mining and data analytic techniques to people-related data.
Admittedly, that definition itself could lead to some head-scratching, so here’s another way to sum it up: HR managers and executives use data science to more effectively manage employees to ultimately reach business goals more quickly and with higher productivity.
Though it is still in the process of being more widely accepted, understood, and used, the HR metric trend seems to be here to stay as an HR-related business strategy. According to a report by Deloitte in 2015, 35% of companies surveyed said they were actively developing HR analytics-based strategies.
To help companies in the process of deciding whether to implement a strategy around HR data science mining, we’ve pulled together some HR data science basics, as well as some caveats to be aware of with each.
Employee Productivity Numbers
Calculating this metric is a simple equation of total revenue divided by total number of employees. Why monitor this particular figure? Paying attention to this trend year-over-year can be useful when comparing against other companies in the same industry.
Don’t: Use these numbers as a ranking system for employee efficiency.
Do: Pull insights from these metrics to determine whether a new strategy is needed to help improve overall employee productivity levels.
Cost-Per-Hire Metrics
A more commonly-used HR metric, this number is calculated by factoring in all costs related to the hiring process, including time spent by hiring managers (and their salaries), job board postings, relocation expenses, etc.
Don’t: Consider cost-per-hire numbers on their own. Some potential hires will inevitably be more expensive to recruit, especially those filling more high-profile roles.
Do: Keep these metrics in mind for budget planning. Just remember to think big picture—at the end of the day, hiring comes down to quality over cost.
Retirement Prediction Formula
Data science is also being used to predict which employees will be considering retirement next, using historical data and complex algorithms. These estimations used to be based on age and tenure, but experienced data scientists are now taking into account recent changes in role, pay level, and incentive eligibility, etc., when calculating retirement predications.
Don’t: Use these predictions to neglect employees deemed as likely being close to retirement.
Do: Use retirement prediction data to start planning ahead for the roles that will likely become open positions in the near future.
Turnover Cost Calculations
This formula is fairly straightforward—it calculates what it costs a company every time an employee leaves, based on spend related to hiring and training each new replacement. According to a study by the Center for American Progress, the cost of losing an employee can cost up to 213% of the salary of a highly trained position.
Don’t: Retain employees just to avoid how much it will cost to acquire replacements.
Do: Be intentional about regularly measuring employee satisfaction to hopefully minimize turnover over time.
While the use of HR data science mining is becoming an increasingly important tool for companies, it’s important to remember that these calculations should help inform decisions made by management, but not definitively make them. Be intentional about which people-related data will be used for which purposes, and then use it to help improve employee management and performance—not to weed out the weakest link.
 Matt Thomas is the President of Indianapolis-based WorkSmart Systems, Inc., which he founded in 1998. He is active with the National Association of Professional Employer Organizations (NAPEO), and has dedicated more than 20 years to the Professional Employer Organizations (PEO) industry dating back to his early career with industry leaders ADP and NovaCare Employee Services. WorkSmart Systems Inc. is a leading PEO, which serves over 200 client companies with employees in 37 states. |