To help improve HR decision-making, identify future skills gaps and reduce attrition, payroll processor ADP turned to predictive analytics.
The Project: Deploy predictive analytics internally at ADP to improve human resource decision-making, identify future talent gaps and reduce attrition.
The Business Case: As the largest payroll services provider in the world, with $11.3 billion in revenue, ADP was sitting on a gold mine of employee information -- data on 33 million workers, to be exact.
As a provider of benefits administration, time-and-attendance tracking, and other HR-management services, "we have a broad swath of human capital management information," says Mike Capone, ADP's senior vice president and CIO. "We asked how we could leverage that in a way that no one else could."
With the post-recession war for talent heating up--not only within ADP's four walls but also among its 620,000 clients -- and the emergence of increasingly sophisticated tools for parsing unstructured data, the time was right to explore predictive analytics for HR. "Unemployment numbers may be high," Capone says, "but the reality is that the really great people are hard to find."
First Steps: ADP had been performing rearview-mirror analytics for years, but in 2012 it began the shift from "here's the information in a nice, pretty dashboard" to exploring what-if scenarios, says Capone. ADP's own HR professionals wanted information to help it tackle issues related to talent retention and an aging workforce.
The company created two tools: First, a retention risk indicator, which analyzed activity like increased absenteeism and exercising stock options to identify high-performance employees who might be leaving.
Second, a demographic visualizer that predicted what the company's workforce might look like in five years. IT built the tools internally, based on IBM Cognos software, working hand-in-hand with HR to validate data and test the tools.
The biggest challenge for the 60-year-old company was data cleanliness. "We've built this data up over a long period of time," Capone explains. "The transactional data was intended to produce a certain outcome, and it wasn't always consistent. We spent a lot of time--more than we anticipated--on mapping and transformation. It was painful but necessary."
What They Discovered: HR had sensed that there was a looming retirement problem, and predictive analytics confirmed it. "We hadn't quite anticipated how extreme it would be when the stock market came back and employees' 401(k)s started looking better," Capone says. "We were able to analyze that to quickly set up some training programs to get the next generation of workers up to speed."
The system pinpointed certain skills ADP was likely to lose, like mainframe support, which is critical to its tax services business. HR was able to focus on entry-level workers who might view mainframe work as a foot in the door at the company, and consider offshore workers who could perform those tasks.
The retention risk indicator served as an early warning system for managers who might not have realized that a key employee was likely to leave. They got alerted when an employee hit a certain threshold of activity and were able to have conversations with those employees about what might make them stay -- for example, more competitive compensation and benefits.
"Most turned out to have been already looking around," Capone says. "These are people we would have hated to have lost." One such critical employee that ADP was able to retain: a data scientist.
Read more about business intelligence (bi) in CIO's Business Intelligence (BI) Drilldown.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.