What happens to all the data HR collects? There’s the demographic information, compensation, headcount, tenure, turnover, leave, and absences of your current workforce. There’s also internal mobility, promotion rates, succession plans, performance reviews, exit interviews, and, increasingly, skills assessments and training records. And piling up at regular intervals are employee sentiment surveys and quarterly pulse data.

According to a report by people analytics platform Visier, HR is investing time and resources interpreting that data, but teams are getting hung up in the foundational stages of analysis; that is, calculating figures like headcount, attrition, and demographic changes. Visier’s VP of product development, Ian Cook, estimates that HR spends about 70% of its analysis time on those low-level calculations, and it’s occupying resources better used for solving business problems affecting the bottom line. HR departments preoccupied with foundational analysis are missing the automation that produces scale. They’re sinking time into updating dashboards and calculating stagnant figures that are quickly outdated.

“A human capital management system, typically, is a system of record people,” said Cook during a From Day One webinar, about how to get started with people analytics. “People analytics is a system of insight.”

Some HR leaders stumble into those systems of insight by necessity. Carla Williams spent 20 years of her career in human resources. Eventually, she found herself having to answer a different kind of question. Her company’s board approved a few million dollars for one-time retention bonuses to keep highly skilled workers. But before they started cutting checks, her CHRO asked: If we hand out bonuses, are we solving the right problem?

“We started an analytics project where we ultimately found that we were solving the wrong problem,” said Williams, who now leads the advisory services practice at Visier. “What we had was a career problem, not a compensation problem. Based on that, we reallocated the money and made changes that actually helped retain our high-potential employees.”

Plenty of organizations do this, throwing money at personnel problems only to see poor results or chase down the same problem elsewhere. “On the surface, it might seem as though the problem is that we need to pay them more money, but when you pull back and look at the information that underlies it, then you might see a very different story.” Williams said. It became her mission to build a people analytics function to ask–and therefore answer–better questions.

People Analytics Isn’t a Side Project, It’s a Business Tool

If firms aren’t using people analytics like they could be, it’s usually for lack of understanding what it is, who is capable of it, and what it can accomplish.

Williams and Cook hear from HR leaders who think it takes a PhD or a data scientist working in the department to do people analytics, or that it’s better left to consultants who parachute in, but they’ve found that HR practitioners themselves are best equipped to solve their own problems.“Context is so incredibly important, and that’s why it can’t be done in a back room by somebody who doesn’t talk to people,” Williams said. “You need to really understand the business. The most important characteristic of someone who can be really great at discovering insights is experience and curiosity and having the right context. That doesn’t require a data scientist, it requires being able to ask the right questions.”

Asking smart questions doesn’t impart some higher-level math skills, so HR teams are using sophisticated analysis tools to solve problems. Tools like artificial intelligence, for example, which Cook and his team build. “Generative AI allows our clients to remove all of those barriers to access,” he said. Maybe you want to know whether you’re at risk for turnover in the next year, you might ask, Is my team engaged? Are they underpaid? “A generative system can take natural language and turn it into question sets that the technology can understand, so you can then get back the answers you want.”

“Analytics should follow a talent strategy that is built off a business strategy,” Williams added. The organizations with flourishing people analytics programs start with a business problem, then go to HR because they have the greatest understanding of the company. That’s the best way to build an analytics practice: with real problems the business needs to solve.

If an executive says turnover is too high and HR needs to stand up a retention program, the analyst intervenes to diagnose the problem before prescribing a remedy: Why do you think retention is too high? What does “too high” mean to you? Is turnover high across the organization or in a single department? Is this a new problem or long-term? How do we compare to others in our industry?

HR departments tend to run universal programs–performance management, engagement surveys, or compensation cycles, for example–but Cook says there’s a better way. “You serve nobody by trying to serve everybody,” he said, noting that people analytics forces the business to focus on specific problems, not vague ideas of what might be going wrong.

“It’s much better to focus narrowly on a specific area in the business, solving a business challenge with a key set of data in a meaningful way and then scale from there,” he said. “It’s OK to ignore 90% of the business and make 10% really successful. That will demonstrate capability, and it will likely lead to further investment.”

Editor’s note: From Day One thanks our partner, Visier, for sponsoring this webinar.

Emily McCrary-Ruiz-Esparza is a freelance journalist and From Day One contributing editor who writes about work, the job market, and women’s experiences in the workplace. Her work has appeared in the Economist, the BBC, The Washington Post, Quartz, Fast Company, and Digiday’s Worklife.