Putting predictive analytics to work in hiring

Finance departments that leverage the right applicant data can streamline the interview process and pick better candidates.
By Lou Carlozo

For all the speculation over how automation and artificial intelligence will impact finance jobs, this much is true: In the here and now, machines are already helping to hire the right people.  

"The use of predictive analytics in the recruiting and hiring space has exploded over the past couple of years," said Eric Knudsen, manager, people analytics at Namely, an HR platform for small to medium-size businesses.

As the name implies, predictive analytics involves using advanced data analysis techniques to make predictions about the future and may involve advanced technologies such as artificial intelligence and machine learning to refine those predictions. In the context of hiring, more companies are turning to predictive analytics because it allows them to identify fixed variables to measure job candidates (such as leadership positions held or years of experience in the same position with another company). Then, résumé and questionnaire data provided by applicants are used to identify correlations and matches.

Thanks to predictive analytics, employers that once relied on observation and interview techniques can now approach the hiring process in a much more systematic fashion by putting data to the task. In one scenario, a job applicant who takes a multiquestion assessment can be scored in real time; if he or she scores high, an immediate invitation for an interview is sent out.

Or in another case, certain keywords such as "project manager," "MBA," or "CPA" could trigger a filter that sends the résumé to the top of the applicant pile, saving time for HR departments while providing more assurance that good candidates aren't overlooked in a large applicant pool.

Many companies also customize their predictive analytics programs by using attribute and performance data collected from the company's top performers. From this, they can create a profile that represents an ideal match. Candidates round out the picture by supplying screening assessment data that help to create a psychological/emotional profile, score leadership and collaboration skills, and/or determine a degree of fit within the company culture. 

For the companies that use a predictive model, it boils down to the no-nonsense notion of taking the guesswork out of hiring decisions.

"Predictive analytics identifies, in detail, the behavioral talents that fundamentally drive performance," said Hugh Massie, founder and president of Atlanta-based DNA Behavior International. "This enables the recruiter to consider fit for role, fit to line management, fit under pressure, and gaps that need to be addressed with training."

It can also work well in the opposite direction. Massie noted, "Recruiters and hirers who use predictive analysis can identify potential mismatches, hence avoiding workplace disruption and making more accurate hiring decisions."

And if you're looking to shuffle the deck with current employees, "it also helps management make more calculated and transparent decisions in moving people between roles — it can be done on fit — which is good for the organization and clear to the employee or candidate as to why," Massie said.

Yet there are pitfalls. They include a reliance on algorithms that create valid statistical results but don't mean anything in practical application — what Brent Holland, executive vice president at FurstPerson Inc., calls "dustbowl empiricism."

"Garbage in, garbage out," said Holland, whose company creates custom, technology-based talent assessment and hiring solutions. "One of the greatest risks to any analytics program concerns the quality of data feeding the algorithm. Bad data will ultimately lead to erroneous and possibly disastrous results."

For example, an imprecise analytics program that correlates between number of jobs held and experience might identify posts such as "intern" or "clerical assistant" in its scoring — even though they might be irrelevant to the job requirements.

"Additionally, more attention has recently been directed at the tendency of predictive algorithms to perpetuate bias, especially in legally sensitive arenas like recruiting and hiring," Knudsen said. "The overarching premise of this claim is that if humans build the algorithms, we essentially teach them our biases in the process."

One way to avoid these traps is to customize analytics through careful research. Here are three more action steps for companies to take when adopting a predictive analytics platform:

  • Account for personality traits. To succeed, a predictive program must look at more than just years of experience, promotions, and titles. "It can provide objective insights into key personality traits intrinsically related to workplace performance," Massie said. "Using that information as the foundation for the interview then increases the recruiter's ability to get to the candidate's real strengths and struggles for performance in the role." In other words, predictive analytics, while useful, is still a foundation for human decision-making and action, not a substitute for it.
  • Ensure connection between HR systems. "One of the biggest obstacles preventing organizations from getting maximum benefit from their predictive analytics is that the various systems they use don't talk to each other," Holland said. "Companies today often utilize multiple systems, such as applicant tracking system, assessments, learning management, performance management, and the HRIS [human resource information system], and when these systems operate as silos, the company will be unable to gain the complete picture of its talent or identify ways to improve."
  • Hire an expert. "This could be an internal expert in predictive analytics and/or machine learning," Knudsen said. "If your predictive analytics needs are more specific or perhaps related to a one-time project, seek out consulting firms that specialize in your needs. Ensure that as a part of your engagement, the firm is able to walk you through their philosophy and methods."
  • And if that setup works out, you could well use it to hire your next predictive expert.

Lou Carlozo is a freelance writer based in Chicago. To comment on this article or to suggest an idea for another article, contact Chris Baysden, associate director–content development,

Where to find May’s flipbook issue

The Journal of Accountancy is now completely digital. 





Leases standard: Tackling implementation — and beyond

The new accounting standard provides greater transparency but requires wide-ranging data gathering. Learn more by downloading this comprehensive report.