Harnessing the power of talent management data
As talent management evolves into the behemoth of data that it is today, AI is becoming an even more important tool in helping employers approach employee development.
“Talent management has evolved over decades and decades of time to the state that it is today,” said Eric Sydell, executive vice president of innovation at Modern Hire, said Thursday, “and it’s a pretty confusing space where you’ve got applicant tracking systems managing the transactional flow of candidates through the pipeline.”
Sydell, speaking at the virtual HR Technology Conference & Expo, said the emerging concept of deep talent is a state of knowing and understanding how human capital quantitatively integrates and drives performance at your organization.
“The idea is that talent management can become a much more scientific, rigorous, efficient and, to some extent, an automated type of practice,” he noted.
When building a strategy to report talent analytics, there are a number of layers Sydell said HR leaders should keep in mind:
- Level 1, Reactive: Operational reporting on basic metrics
- Level 2, Proactive: More advanced reporting, including some benchmarking and using data to make some decisions
- Level 3, Strategic: Using more sophisticated analytics and trying to understand the cause of what might occur in the hiring process
- Level 4, Predictive Analytics: This is traditionally considered the top level, where you’re really understanding whether your hiring data predicts outcomes that matter
- Level 5, Deep Talent: Where you really begin to automate that process with predictive analytics. Automatically collecting data, housing it in a common place to be analyzed by algorithms and building that system out so you can have AI-driven insights
Josh Allen, director of global selection and assessment strategy at Walmart, added that there are ways you can structure data early to use it later on.
“Really, what we [as HR professionals] would like to move to is intentional data collection,” he said. “How many companies know their associates’ aspirations? Do you know where they want to go?”
He pointed to resumes, as an example. “Things like resumes, [they] could be a source of data further down … but it’s rarely captured in a way that you can use later.”
So, what are some of the barriers?
Structure and linking data, for one, Allen noted.
“I think the issue that largely goes unnoticed in HR research is the lack of clear criteria in variables,” he noted. “We don’t often know what we’re trying to predict in HR data.”
“We don’t often know what we’re trying to predict in HR data.” Josh Allen
That analysis piece is not as hard since so much of that can be automated at scale, Sydell added.
“The hard part is the manual part of getting the data together,” he said. “You’ve got to put in place the tools to collect good data now so later they can be studied and you can make sense of them.”
And AI is enabling greater predictive accuracy, he added.
“And I think the biggest benefit here is in diversity, equity and inclusion: using data to mitigate bias and improve decision-making,” Walmart’s Allen said.
But, these systems can’t go unchecked, he warned. If we put in biased data, we’re going to get a biased algorithm, he said. “We have to understand how these are working and not just push the problems downstream.”
“I really think the biggest unlock is in improving our DEI; I think it can also allow a roadmap for development,” he said. “If we know ‘Here are the things you need to be successful,’ we can be transparent with this and set up people properly with development plans.”
“I agree,” Sydell noted. “I think, early on, a lot of the AI created bias, and the reason that happened is they trained their AI on biased training sets.”
Now that companies have learned more about the proper way to build AI, you see it less and less, he added.