4 Steps for Employers to Develop a Reliable People Analytics Process

 

Lisa Nagele-Piazza, J.D., SHRM-SCP By Lisa Nagele-Piazza, J.D., SHRM-SCP November 12, 2019
LIKE SAVE

NEW ORLEANS—Artificial intelligence (AI) can help employers streamline their hiring processes, improve diversity and inclusion efforts, and monitor their practices for pay disparities and other workplace biases. But people analytics can be scary—and ineffective—if employers don't know how to use it properly.

Research shows that people don't understand or don't trust data analytics, said James Banks Jr., general counsel for the Society for Human Resource Management, during a Nov. 8 session at the American Bar Association's 13th Annual Labor and Employment Law Conference. 

For example, only 35 percent of senior executives from around the globe who were surveyed by KPMG, a Dutch professional-services firm, said they have a high level of trust in their organization's use of data and analytics. Sixty-five percent said they have reservations, limited trust or active distrust regarding the data and analytics they produce.

"Oftentimes in a new area there's quite a bit of 'snake oil' being sold," said Peter Boumgarden, Ph.D., a professor at Washington University in St. Louis. Employers need to figure out what claims from vendors are actually credible and what people analytics can really do for their company, he noted.

Boumgarden introduced the following four-part process to help employers evaluate the effectiveness of their people analytics procedures.

1. Model

Start by evaluating the organization's goals. "What exactly are you trying to measure?" Boumgarden asked. "What's your approach to modifying the organization?" For example, is the employer looking to improve diversity and inclusion through its hiring and retention efforts?

Employers need to consider whether they want to look at individual engagement or more collectively at the average engagement across the organization. "In general, the bigger the aggregate is, the easier it actually can be to model those things with accuracy," he said.

Banks added that employers need to think about how they will pull data. A lot of data are managed by the HR department and housed in a human resource information system, but some important data may be stored in a separate, stand-alone applicant tracking system or third-party payroll system.

"If it's all separate, then the analytics are really limited," Banks said. "The better corporate operations really focus on trying to integrate all the data."

2. Measure

Measuring human behavior accurately is really tricky, Boumgarden noted. "What's interesting about the people analytics space is … we're starting to address more unobtrusive measures," he said. So, for example, instead of asking conference attendees to respond to survey questions, analysts may look at attendee reactions to the conference in e-mails and on social media.

After determining what the organization wants to measure, consider how to capture the most relevant metrics, Boumgarden said. Does meeting attendance actually measure engagement, or does analyzing attendees' smartphone use during a meeting better show whether they were listening?

"It's not enough to talk about the numbers," Banks noted. Employers need to transform data into a story about their company.

3. Modify

Consider what's working and what isn't. Employers should develop a method to test changes, Boumgarden said. How can the organization modify its systems to achieve its most critical objectives?

Samantha Grant, a management attorney with Sheppard Mullin in Los Angeles, noted that HR professionals tend to support the use of AI to improve hiring processes, diversity and inclusion efforts, pay equity, and other aspects of employment—"though sometimes lawyers are afraid that the data will be used in a lawsuit," she added.

James Finberg, an employee-side attorney with Altshuler Berzon in San Francisco, said employers need to ensure that the algorithm isn't biased. A seemingly neutral system may have a disparate impact based on protected characteristics, such as national origin, sex and race.

"You have to test to see if it has disparate impact, and if it does you have to fix it," he said.

[SHRM members-only online discussion platform: SHRM Connect]

4. Mitigate

Legality and employee perception are employers' biggest fears when using AI, Boumgarden noted.

There is also a chance that the statistics will be skewed any time people are measured—especially if they're aware that they are being measured, he said. For example, if workers know their employer is monitoring their Twitter comments to gauge their job satisfaction, workers might tweet more positive statements and discuss their criticisms offline.

Employers want to use technology, but they want to do it in a way that eliminates biases in hiring and promotional decisions and helps them become more efficient, especially when they're getting thousands of resumes, Grant said. "The biggest concern is doing it right."

Data collection can sometimes feel like Big Brother to employees, Banks added, but it can be very valuable if it's aligned with the organization's strategy and communicated properly to employees. He said company leaders need to reach out to employees to say, "We hear you. You matter to us."

LIKE SAVE

Job Finder

Find an HR Job Near You
Search Jobs

Earn a SHRM Talent Acquisition Specialty Credential.

Do you have what it takes to win the war for talent? Find out.

Do you have what it takes to win the war for talent? Find out.

LEARN MORE

SPONSOR OFFERS

Find the Right Vendor for Your HR Needs

SHRM’s HR Vendor Directory contains over 10,000 companies

Search & Connect

HR Daily Newsletter

News, trends and analysis, as well as breaking news alerts, to help HR professionals do their jobs better each business day.
temp_image