Predictive Monthly: Using Algorithms To Add Science To Human Judgement In HR

Pierre Leroux

If you’ve ever headed a non-revenue generating support function, you’ve likely felt the relentless pressure to justify its existence. HR executives are no stranger to this pressure: In their traditional roles of hiring, firing, and compensation, HR frequently struggles to demonstrate its contribution to the bottom line. However, new predictive analytics software is helping them improve hiring and predict critical business problems related to attrition, skills, and customer satisfaction. Armed with this insight, HR executives are now carving out a more strategic role for themselves within the corporation.

Building the science behind HR decisions

Predictive analytics software works by using algorithms to find patterns in large volumes of data, which then enable predictions to be made about the future. By using machine learning, the software looks at a particular outcome and uncovers the factors behind it (which could include thousands of possible causes – too complex for a human brain to handle). Examples could be the factors causing attrition, the contributors to strong employee engagement, or indicators of leadership potential.

Making the right hiring decisions

Good recruitment is all about hiring people that are a good fit for the job and who will be successful in the role over the long term. Longevity is a very valid concern given the potential cost of a bad hire – estimated to be as much as three times an employee’s salary. The traditional information collected in an HR recruitment exercise – qualifications, past performance, references, and interview interactions – is not necessarily enough for a strong indication of future success in a role. So, predictive analytics is now being used to identify the characteristics and commonalities that are driving top performance in existing employees and providing more data points with which to assess job applicants.

A leading American bank is putting this into practice by running algorithms on digital resumes to identify candidates with the bank’s preferred traits. It is also using the same technology to ensure human interviewers don’t overlook strong applicants.

While this approach can unearth new predictors of a candidate’s success in a role, it can also, unfortunately, throw up superficial commonalities that have little to do with performance. The key here is for HR leaders to ensure they understand how the factors that are being considered were selected. Statistical tests can also be run to determine whether an algorithm is screening out particular demographic groups.

Retention and engagement

In its recent Human Capital Trends 2015 report, Deloitte revealed that 87% of business leaders are highly concerned about retention and engagement. This comes as no surprise when disengaged employees are costing the U.S. economy $500 billion per year in lost productivity.

Losing employees is both costly and disrupting, so attrition is a major focus for most corporations. HR needs to answer key questions such as: Who is going to leave and why? What can be done to keep them? Algorithms can help by identifying and quantifying leading indicators of attrition among employees. Armed with this information, managers can then take preemptive action to retain employees. In addition to spotting potential leavers, predictive models are also helping HR teams predict vacancies and leadership needs, identify skills shortages, and mitigate the risk linked to seasonal absences.

Retention and engagement are two sides of the same coin, so, unsurprisingly, predictive modeling is also helping to uncover critical engagement factors that drive performance in an organization. HR leaders are now able to prioritize initiatives that improve employee motivation and retention.

In transportation, where annual driver turnover can be higher than 100%, a leading trucking company is using predictive analytics software to better understand its workforce, identify the factors behind turnover, and anticipate drivers at risk of leaving. In doing so, it has reduced driver turnover by 15%.

Linking learning to ROI

A question never far from the mind of any learning and development executive is: Which training activities and technologies provide the best value? With training costs under scrutiny, it’s crucial to show that training initiatives are driving business outcomes. Many organizations are now using predictive analytics to find correlations in their HR data that show which training programs give the strongest ROI.

These techniques can also help to tailor training plans for specific employees by predicting which development approach would work best for an individual and which competencies they would find easiest to learn. A global technology company is using predictive software in this way to maximize the effectiveness of training for its call center agents in order to improve its customers’ experiences.

Success in HR is down to making the right decisions and plans about important people matters. Predictive analytics is an indispensable new tool for HR executives that want to sharpen their game by adding science to human judgement.

Read more about success with algorithms in Algorithms: The New Means of Production.


Pierre Leroux

About Pierre Leroux

Pierre Leroux is the Director of Predictive Analytics Product Marketing at SAP. His areas of specialty include Data Discovery, Business Intelligence, Cloud applications, Customer Relationship Management (CRM), and ERP.