Using Predictive Analytics in Talent Management

While interest in using more-rigorous, data-based analytic approaches to HR decision-making is growing, real progress has been limited, in part because of organizations’ misguided assumptions. Assumptions regarding costs, technology, methods and skills have prevented organizations from building and deploying predictive solutions to fully leverage valuable historic data that already exist in their systems, said a PwC Saratoga expert in a recent webcast.

Predictive analytics as defined by PwC Saratoga refers to a set of processes, tools and methodologies that can help organizations analyze large data sets to forecast future events. The objective: Make the best use of both public and private historical data to arrive at better business decisions.

Companies can use predictive analytics to increase their effectiveness in predicting attrition risks, hire quality, employee performance and leadership potential, as well as in generating optimal hiring profiles, said Ranjan Dutta, Ph.D., a director in PwC Saratoga’s human capital metrics/benchmarking practice, in a March 20, 2013, webcast titled “Dispelling the Myths of Predictive Analytics.”

But he said HR professionals often fear that their department isn’t mature enough or doesn’t capture enough data to use predictive analytics. There also are fears that it would require making big technology investments or hiring statisticians. Still other HR leaders think they can buy an advanced human resource or business intelligence system and let the system do all the work.

What’s needed is “a paradigm shift in HR thinking,” Dutta told SHRM Online. “You are moving away from being a reporter of past events to being a more strategic business partner.”

Analytic Maturity

When it comes to workforce data, companies move through a multilevel analytics maturity cycle, Dutta explained during the webcast. At level 1, HR can answer the “what happened” questions, such as “How many employees voluntarily quit in June?” or “How many new hires did the finance department have in November?”

At level 2, they can not only answer what happened but also extract key data to compute standardized metrics “that allow for systematic tracking and comparison with peers,” he said.

They can begin to answer the “why” and “how” questions at level 3, such as why high-performers quit and how many new hires are needed to make projected revenue growth.

Most HR organizations have yet to reach the highest-impact maturity level (level 4), from which they can build, deploy and maintain predictive solutions seamlessly or build predictive indicators into their reporting dashboards, wrote Dutta in an article titled “HR’s Game-Changer: Predictive Analytics,” featured in the summer 2012 issue of the PwC Saratoga publication HR Innovations. In fact, only 12 percent of the 383 U.S. organizations surveyed have attained level 3 or higher, according to a recent PwC Saratoga study.

But they should keep this model in mind when thinking about what they need in terms of people, technology and processes “if they want to move beyond their current level of maturity to their desired state.”

Acquiring Data, Technology

Predictive modeling isn’t about having the most data; it’s about testing the “right” hypotheses or theories with the “right” data, Dutta contends. What’s more, organizations can build and deploy predictive solutions without making any deep investment to get their data into a single repository, he added. The data is most likely already there in a transactional system; core HRIS, performance management and talent management systems; and employee surveys.

Companies shouldn’t expect an out-of-the-box advanced HR/business intelligence solution to do all the work, he cautioned.

“You can’t buy your way into predictive analytics without making necessary investments in the skills and capabilities of your people,” Dutta said, because out-of-the-box solutions aren’t useful “unless you’ve got people with the right skill sets already in place to drive value from technology.”

Dutta said such capabilities might already exist in business operations, marketing and finance, and some organizations may even have analytics centers of excellence. These centers work, he said, provided they are distinct from the IT function.

“IT wants to own analytics—they think of it as a natural extension to what they do—but analytics is better off as a central but independent function, or situated within the business functions themselves,” Dutta told SHRM Online.

Analytics Competence

There are three main components to workforce analytics, according to Dutta:

• Data accessing and managing.
• Analysis that includes both descriptive and predictive analytics.
• “Reporting and taking action,” in which insights captured from both the descriptive and predictive analyses are presented back for general business consumption.

“Descriptive analytics is about slicing and dicing data and discovering patterns,” Dutta said, adding that typical business intelligence tools can help with this. “Predictive analytics is hypotheses-driven advanced statistical modeling to predict future events.”

While knowledge of statistical modeling is needed to build a predictive model, it’s most important to work with someone who is capable of interpreting the results and linking them to actionable workforce decisions.

Hiring statisticians isn’t the answer. “Organizations need to be more deliberate and more strategic,” Dutta said.

He said there are three types of experts who can help the HR function build sustainable workforce analytics capability:

• People with data manipulation skills.
• People with advanced statistical modeling skills.
• People who can interpret and effectively communicate “data-heavy” results to business partners.

“The path to the highest maturity level, where predictive solutions are part of your organizational workflow, involves considerable investment beyond technology, extending to the skill sets you hire for, the processes you build, and the technology and tools that enable those people and processes,” Dutta said.

Pamela Babcock is a freelance writer based in the New York City area.