People analytics is an evidence and data-driven approach to making people decisions within business. It has been an important part of strategic human resource management from the inception of the first employee survey and the rise of the scientific management movement decades ago. While early work sought to understand what happened and why, advances in thinking and methodology have opened the door to effectively predicting and influencing future behavior of people and organizations. This is the future, and in many ways the current reality, of people analytics.
Predictive analytics in HR represents the next natural step in the evolution of integrating data into strategic HR management, promising the ability to look into the future and see what’s around the next corner. In fact, the practice is booming. This article will examine some of the ways predictive analytics is currently being used in HR, and developments on the horizon.
The Promise Of Predictive Analytics In HR
Though predictive analytics holds the promise of foretelling the future, it starts with a thorough examination and understanding of past behavior. Building a predictive model begins with looking in the rearview mirror to ask “What happened?”, “Where did it happen and to whom?” and “Why did it happen?” Understanding those connections and their implications allows data analysts to move past purely quantitative data and get into qualitative reasoning about what will likely happen again in the future. This future-oriented focus creates significant value and attracts attention beyond HR, as leveraging this data from the past allows leaders to anticipate what’s just over the horizon.
Essentially, predictive analytics provides an answer to the question “How do we make something happen?” That something may be recruiting the right people, increasing productivity, reducing voluntary attrition, or some other goal. Gathering and analyzing data allows analysts and HR to offer solutions as to what to do and where to focus, so leaders can strategically position the organization for success. This is the end goal of a mature predictive analytics system.
It’s very important for organizations to be specific and intentional in their approach to predictive analytics. The best practice is to begin with a management question or dilemma, such as retention/attrition or vulnerability to union organizing. The next step is to apply research questions to that dilemma:
- Where are we at risk?
- Why are we at risk?
- Does the risk arise from internal or external factors?
- What can and should be done in response to the risk?
These questions and their answers provide the framework for predictive analytics.
Applying Predictive Analytics To Retention & Attrition
One of the areas where predictive analytics can have a major impact on the bottom line is in reducing voluntary attrition. High turnover costs businesses billions each year in recruiting, hiring, onboarding, and training. Aside from monetary concerns, companies are better off when they can retain employees and the organizational knowledge they possess.
Predictive power about attrition can be gained by using survey data collected six months to one year in the past, and creating a post-hoc demographic of employees who left the organization voluntarily. Analysis of this demographic will reveal where those employees sat in the organization and if there are pockets of high turnover. Did employees at a certain job level, in a certain job type, or in a specific location or business unit leave the organization at a higher rate than others?
This information is particularly important because it helps pinpoint the issues driving employees out of the organization. For example, one organization we worked with repeatedly talked about its 200% turnover rate—a number that seems to indicate an average tenure of six months for every employee. Further analysis instead revealed there were just a handful of job profiles where employees left after a month on the job; at the corporate level, there was virtually no attrition. Company leaders’ description of the problem was coloring their perception. Given a more realistic view, leaders were able to manage the organization differently, putting a different focus on the stable population while looking to make improvements in the problem areas.
Improvements needed for problem areas are identified by loading the post-hoc demographic of employees who departed, and asking questions about where and why there is risk.These can be answered by comparing the employee experience of employees who left versus those who remained with the company. In comparing the experiences of employees who left versus those who stayed, differences in experiences can be understood and an action plan to address the barriers to keeping people in high-attrition job types, locations, or business units can be formulated. Issues that may be revealed can include:
- Dissatisfaction with manager
- Dissatisfaction with work/life balance
- Negative perceptions about career opportunities
Categorizing barriers to retention allows data analysts to build profiles and personas of employees most at risk for attrition, which can be overlaid onto current data to identify those who most closely match the profile. This can flag the groups likely to have disproportionately high turnover. Organizational, personal, and experiential data can be used to create a risk index and predict the number of employees who will leave with 80–85% confidence.
Using Predictive Analytics In Recruiting
Profiles and personas can also be used in recruiting, but in this case, they are used to predict which applicants are most likely to succeed in a specific job or role. For recruitment purposes, analysts construct profiles of highly engaged top performers to see what was different in their selection and onboarding performance. From a predictive analytics perspective, analysts are looking for positive flags, to uncover a disposition or personality profile that is disproportionately effective and impactful in the job role. Candidates for that job role are then compared against this profile to see how closely they match it.
When using predictive analytics in recruiting, organizations need to be careful with regard to the Equal Opportunity Employment Commission (EEOC) and Americans With Disabilities Act (ADA) guidelines, and work to eliminate the opportunity for adverse impact on protected groups. For example, data may reveal that the organization has difficulty retaining employees with a specific ethnic background, but if that is built into the model, it could reduce hiring from that population, negatively affect diversity within the organization, and would be a potential violation of the EEOC guidelines. This is an area where HR teams are wise to involve their legal council to review best practices.
Organizations often conduct candidate assessments, but they should also be careful about the specific assessments they use. If, for example, the assessment is able to detect mental health issues such as depression or bipolar disorder, it can run afoul of the ADA. Keep these implications and potential pitfalls in mind when designing and using candidate assessments.
Ethical Considerations For Use of Predictive Analytics In HRAs noted above, there can be ethical implications in the use of predictive analytics. It’s important to keep in mind that just because a thing can be done, it doesn’t mean that it should be done. Using data to predict future behavior starts to become an issue if it crosses over into attempts to predict individual personal decisions. Predictive analytics can do a good job of predicting and identifying where interventions should be focused, but it shouldn’t stray into territory of becoming so targeted that it singles out individuals and assumes that accurate predictions can be made about what they will do next. As predictive analytics technology becomes more powerful and finds wider use, it should not cross the line into using algorithms to make autonomous decisions. One of the challenges for predictive analytics in the future is to make sure that it never crosses over into Minority Report territory—a very real risk as models become ever more accurate in predicting future events. |
What’s next in predictive analytics for HR?
Predictive analytics models are continuing to evolve and advance as more machine learning and artificial intelligence are integrated into the process. The challenge is in testing and retesting these models against myriad demographic and other data. Organizational network analysis (ONA) is also starting to be incorporated into some models in order to understand organizational commitment and attachment for predictive purposes.
The power of predictive analytics comes from the ability to influence outcomes. In a very real sense, predictive analytics can change fate. People and organizations are dynamic, so there is opportunity to change outcomes. Predictive analytics demonstrates that outcomes are not inevitable; they can be influenced to desirable ends. (Tweet this!) That, ultimately, is where the future of predictive analytics lies.
Would your business benefit from having a crystal ball?
Imagine what your organization could accomplish if it had the ability to predict the future. Predictive analytics can’t give you the winning lottery numbers—but it can give you the ability to make the right hiring decisions and reduce turnover. The Perceptyx platform collects the data you need to create desirable outcomes. Paired with customized survey design and analysis, our platform can help you solve your biggest business problems—and help identify changes you need to make for the most desirable outcomes. Contact Perceptyx today.