Applying Big Data to the Workforce Turnover Conundrum
Finally, the art of understanding and addressing workforce turnover is turning into a science. This is great news because turnover – in particular, voluntary turnover – has real consequences for an organization. Besides the obvious loss of employees, more often the good instead of the not-so-good employee, there is impact to productivity levels and the ability to hire other workers.
In a newly released whitepaper, Revelations In Workforce Turnover. A Closer Look Through Predictive Analytics, the ADP Research Institute® (ADP RI) - which identifies and analyzes trends in the labor market to help unlock business potential and enable informed decisions and actions – takes on the challenge of understanding workforce turnover by applying predictive analytics (big data) to its massive database of employee information. The findings not only move key industry benchmarks from speculation to clear fact but also reveal connections among dozens of workplace characteristics that point to the likelihood of turnover. Even for individual companies.
Mapping the Turnover Landscape
The problem with studying turnover is that the data available to analyze has been inconsistent. Based on surveys and interviews of various companies, each input can be an educated guess, a partially measured projection or even a misunderstanding. The margin of error can be substantial. Using a database of actual payroll statistics over time and with a clear set of specific definitions can provide a factual result. ADP can now verify the national average monthly turnover rate (5%) and it can also identify the months when that rate is highest (September) and when it is at its lowest (March).
ADP RI also shows how the rates vary by industry, which has the overall highest (Leisure and Hospitality) and which has the lowest (Manufacturing). These are helpful for an employer to see that their own turnover rate is high (or low) relative to a national average but they can see whether the same situation applies within their own industry. As you might expect, a manufacturer might be comfortable in noting that their turnover is low compared to national rates but when they can see that they are at the highest rate in their industry, concern about competitiveness becomes apparent. The great thing about ADP RI’s study is that there will be more insights specific to subsets of each industry as the data is delved into more deeply.
A Model for Predicting Voluntary Turnover
60% to 70% of turnover is the voluntary kind that employers do not control. This is what creates volatility and uncertainty and what gives many managers a headache. There is little worse than having a prized member of the workforce depart with their training, their skill sets and their institutional knowledge. Regardless of whether or not they go to a direct competitor, they leave a hole in the pattern of production.
In this white paper, ADP RI shares how a sample subset of 1900 firms representing about 7 million employees was used to identify a set of about 40 attributes relevant to voluntary turnover. The attributes – or factors – work in combination with each other and consist of a mix of employee characteristics, industry benchmarks and ratios. They range from the expected things like job level and experience to the lesser expected and sometimes harder to quantify, like commuting distance.
Much of what you would expect and assumptions you have operated on are validated – the lead drivers of voluntary turnover fall in the compensation category. What you might not anticipate are that commute characteristics tend to be more important than experience or tenure.
While the white paper shares some of the factors that contribute to turnover and their relative impact, the ability to do this analysis for individual companies is likely to be of most interest. The contribution of factors and how they combine with other factors varies by industry and by company. The risk for turnover can now be measured and the factors within a company that are most clearly contributing to that risk can be pinpointed. Thus, developing a plan of action need no longer be a guessing game. Recommendations for addressing the issue can point to clear, data-based analysis giving more assurance as to what steps need to be taken.
Jump in and read the full whitepaper - Revelations In Workforce Turnover. A Closer Look Through Predictive Analytics – for more information including turnover probability by industry.
About this report: For this study, ADP RI looked at monthly anonymized payroll data for companies with 25 or more employees for a two-year period from January 2015 through December 2016. This sample of 41,000 companies and 12.5 million employees was used for describing the overall turnover landscape and determining benchmarks by industry. To develop an additional model for predicting voluntary turnover, a subset of the above sample selected 1900 companies with 1000 or more employees. This provided a sample of 7 million employees.