5 barriers to creating data-driven HR — and how to overcome them
Savvy HR leaders know that data is the foundation of a killer people strategy. Here’s why that’s hard — and how you can make it easier.
“In the simplest terms, data can help you make better decisions and, thus, better support your organization and the people in it.”
That’s the belief of Alexis Fink, Intel’s general manager of talent management, and a belief that’s shared by many others. Indeed, an increasing number of human resources leaders have come to trust that properly collecting and analyzing data can lead to more effective HR decision-making — that is, workforce-related decisions that result in better outcomes for both employees and the business.
Still, progress has been slow.
“I think we’re still seeing HR organizations not knowing how to address this,” said Dan Staley, a partner at PwC, based in New York, who heads the consulting firm’s U.S. HR technology practice. “It’s more of a hobby or night job for people versus something that’s really deliberate, where people are dedicated to the function and driving analytics in the organization.”
Most companies, he said, “are still in the crawling and walking stage, especially when it comes to [people] analytics.”
Consequently, executives who want a truly data-driven HR function have a lot of work to do. They face obstacles ranging from a lack of understanding of data’s potential to technical hurdles and the complexities that go along with change management.
With that in mind, here are some of the biggest hurdles organizations confront when implementing data-driven HR — and suggestions for getting around them.
1. Misunderstanding data’s basics
Data-driven HR departments recognize that not all data is created equal. Because so much HR data is based on transactions, Fink said, much of the data an HR information system (HRIS) provides isn’t that useful for making meaningful decisions. So, to create a truly data-driven HR approach, companies must invest the time and resources necessary to determine what information will be most useful, then go out and collect it.
Predicting leadership capability, for example, is easier if good leadership assessments are in place instead of simply considering how long an individual has been in the same job or what their past performance review scores were.
“It’s a bit like throwing a dinner party,” Fink said. “The likelihood of success is greater if you plan a menu and get the right ingredients, rather than simply cook up whatever you can find in your pantry and hoping that aggregates into a decent meal.”
2. Not knowing what you don’t know
Organizations can’t simply load a batch of numbers into a system and call it data. True insights require information that’s reasonably accurate, consistent, relevant and easy to get at, Fink said.
More often than not, that kind of data set isn’t in place as companies begin down the path toward data-driven HR. In particular, most organizations begin with their data housed in different systems, essentially siloed and disconnected from one another, said Michael Housman, chief data scientist at RapportBoost.AI, based in Venice, Calif., and a well-known expert in people analytics.
Solving that problem requires you to identify the source of each piece of data, determine whether or not the data’s clean — meaning “accurate and complete” — and whether it can serve as “a single source of truth,” Staley said.
To accomplish this, HR departments must be ready to dedicate resources to their efforts. While some organizations may bring in a data scientist to help them, the true solution “is going to be a combination of not only people who understand data and data science, but also people who understand HR and understand the business,” Housman said.
3. The urge to overreach
Professionals experienced with data-driven HR are nearly unanimous in saying organizations should start small. As Staley put it: “Don’t try to boil the ocean.”
For example, Housman said companies should try to answer a single question using data they already have and that’s easy to get it, not scattered across different systems.
“What you don’t want to do is start a massive data warehousing project to combine all of those different data streams,” he warned. “I always think it’s more effective to get some quick wins under your belt and show the value versus saying, ‘We’re going to sketch out a two- to three-year roadmap, then spend a year just doing data cleaning.'”
Following that course, he said, can lead to “organizational fatigue” before you’re able to demonstrate results.
Another angle of attack, Fink suggested, is to base your first efforts on data you gather fresh.
“It’s very possible to do lots of useful data work that starts from scratch, collecting new data, rather than locating and cleaning up old data,” she said. “In many cases, the most interesting insights aren’t going to come from HRIS data points like time-in-role, anyway.”
For example, a company might survey employees with an eye toward understanding where performance is suffering or attrition risk is high.
4. The dazzle of technology in data-driven HR
While HR technology vendors preach about the power of artificial intelligence and machine learning, workforce data and analytics experts say it’s important to keep technology in context. The point of data-driven HR isn’t to analyze information for its own sake, but to solve real problems.
Before investing in technology, HR must understand the fundamentals of data gathering and data quality. Just as you shouldn’t start with too big a project, you shouldn’t invest in complex and expensive systems before you truly understand your organization’s needs.
“Having sophisticated tools help, but a lot of this you can do on spreadsheets or more basic tools in the beginning,” Staley said.
At the same time, data-driven HR requires an idea of what metrics are important to answering the most pressing and frequent questions, said Alec Levenson, senior research scientist at the Center for Effective Organizations, University of Southern California Marshall School of Business.
As they consider how to incorporate data into HR, he continued, organizations must take into account “the larger context of what people are doing on a daily basis and how their organizations are making decisions.”
Underlying his words is the caution that data for data’s sake is pointless.
“If we’re going to bring this supposedly wonderful HR data to the table, what is it going to tell us?” he pointed out.
5. People don’t trust data
Employees who aren’t comfortable with data often resist taking advantage of it. Starting small and proving what data-driven HR can do, Housman said, is as much about getting organizational buy-in as it is about learning how HR can put data to use.
“You have people with lots and lots of experience who are distrustful of algorithms,” he explained.
For instance, company recruiters may balk when they’re presented with a candidate score that’s been “generated by some machine that hasn’t been sitting in their seat for 20 years.
This means that building a data-driven HR function requires developing a change-management strategy. At every step, involve stakeholders inside and outside of HR, showing them examples of how they can put data to work so their jobs become easier.
“A lot of it comes down to how you can influence people by putting data in front of them and getting them to change some decision or behavior,” Housman said.
“I think of data-driven HR as a mindset — a philosophy about how we make decisions,” Fink said. “Of course, organizations that prioritize data in making decisions might be more likely to invest in technology to make that simpler, so there’s an element of being data-driven and also an element of being technology-enabled.”
However, she continued, it’s also possible for organizations to invest heavily in technology and base their decisions on instinct or habit.
“I’d argue that those are not data-driven organizations,” she said. In Fink’s mind, truly data-driven HR uses data to guide the company to addressing critical retention issues.
That’s a much different approach than letting leaders decide which path to pursue first, “then go fishing through the data” in hopes of finding evidence to back up the decision they’ve already made, she pointed out.
To access the original article, click here.