Data Science & HR: Supporting better hiring, promotion, and retention decision

Recruiting of data scientists is a tough endeavour. HR departments oftentimes struggle to adequately pre-screen candidates, as their background is rather diverse and there is “no ivy league” for data science. Interestingly, though, even HR departments themselves might have a clear incentive to hire for smart and driven data scientists. Because they can significantly contribute to accelerate an overdue transformation with the corporate HR function – towards a more data-driven, rigorous improvement of workforce performance. While HR remains a ‘people business’, data science will not revolutionize the function, but contribute sophisticated support mechanisms, which can help HR executives to make better, more informed hiring, promotion and retention decisions.

Better hiring decisions

In large corporations, 300-400 applications per job opening on entry level are not uncommon. Pre-screening for the adequate candidates is thus a key activity. HR specialists oftentimes apply certain proxies to efficiently scan the incoming applications: university, grades, experience in the sector. Most management consultancies, known for their rigorous assessment centers, filter based on target universities. If you’ve gotten your degree from another university, you probably only have a chance if you have an exceptional background and high profile work experience. Is that a good practice? Probably not, but at least it is an easy to operationalize framework and gives an easy heuristic to those HR specialists, doing the initial screening.

Nevertheless, there is large potential to improve the processes with data science. As application records are archived and there are track records of successful candidates, which have moved on in the company, the characteristics of those candidates can be distilled. This, taken as such, would already help in the pre-screening process. However, given the existence of large databases of former applicants, it will be possible to rank the incoming applications immediately along two dimension: (1) regarding their suitability given the profile of previous successful candidates (2) regarding their relative position in the incoming applicant group. Such a system would automate a large part of the screening process, and identify suitable candidates outside of static, experience-based heuristics.

Better promotion decisions

Within the organization, the focus of the human resources department is primarily on talent and performance management. Naturally, activities are of a different kind but once again it essentially comes down to enhancing the drivers of performance. What are the fundamental characteristics of high performers? How can workforce performance be predicted, thus high performers be identified early on and fast-tracked?

Until up to a few years ago, these questions have, to some degree, been primarily answered by qualitative assessments, interviews with managers and 360 degree feedback. Now, with technological advances and data science, all that unstructured feedback, collected and archived throughout the years could actually turn out to be a goldmine. When put into context with actual promotion decisions, various new analytical approaches can be introduced. Natural language processing approaches are able to sense sentiments and trends in assessments. This could help to establish automated “early warning systems”, which could serve as a quantitative perspective on promotion decisions. Reliably predicting the performance of employees – not just depending on gut feeling – will be one of the core activities of HR departments in the upcoming years. And bringing as much data to the table as possible, to allow for fine-grained prediction should thus be a strategic objective for companies.

Better retention decisions

Just as identifying high performers is a key activity, retaining employees is just as important. Some employees, no matter what level, fulfill crucial activities and finding adequate replacement could be a tough undertaking (too much domain knowledge, experience as a distinguisher, tough competition for skill on the market). In order to retain people within the organization, those crucial people need to be identified and – most importantly – reasons for any change in their satisfaction with their position need to be monitored.

How could that kind of change in satisfaction be spotted? Generally, prediction models about workforce attrition could form the basis of such an endeavor. How long are people in certain positions likely to stay in the company? How dependent is this on their socio-economic background, their age and education? Once such an understanding has been conceptualized, companies are able to have an overview of those employees, which are likely to leave soon. Once those employees, flagged as crucial, are popping up on the radar, executives will be informed, enabling them to enact counter-measures or whatever is necessary to prevent a departure of the respective employee.

Erasing biases

More data-driven approaches allow companies and HR executives to compose a quantitative mirror for their actions. Blending qualitative and quantitative assessments, supporting human decision-making will be the way forward in HR. It is still a long way down the road, where one could imagine systems of AI, which automatically detect those applicants, which should be invited for an interview. Until then, there will always be some human component.

As a matter of fact, data science not only offers potential for increasing the efficiency of HR processes, but also could contribute to erase biases in recruiting processes. As a matter of fact, human decisions always involve biases, even though we might not even be aware of them. When training algorithms on historic data sets, those biases are likely to be passed on into code. However, once these biases are “formalized” they are rather easy to detect. Enabling companies to increase awareness about biases in their recruiting and promotion decisions is usually the first steps towards improvements. Data science could thus actually contribute to make HR a more ethically sound function and the organization as such more inclusive and productive at the same time.

Contact the author
Paul von Bünau
+49 (30) 814 513-14

Leave a Comment

Your email address will not be published. Required fields are marked *