In HR Tech, All that Glitters is not Gold

After a week of bold pronouncements about things to come, one respected observer wrapped up this year’s HR Technology Conference & Exposition with a sober warning: New software isn’t always the answer.

Peter Cappelli, a professor of management at the University of Pennsylvania’s Wharton School and veteran thought leader in human resources, urged executives to think critically about the practical benefits of shiny new HR technology.

The multibillion-dollar industry offers dazzling possibilities through applications of cutting-edge techniques such as artificial intelligence. But that doesn’t mean new software will always produce results worth the cost and effort, Cappelli warned.

“All kinds of things are possible,” he said in a closing keynote address in Chicago on Friday. “That doesn’t mean they’ll take over.”

Many in the business world believe the pace of technological change is faster than ever, and accelerating. That makes buzzword-laden vendor sales pitches tempting because they offer a chance to catch the edge of the next big wave. But Cappelli believes we overestimate the pace of real change.

To illustrate the point, he noted that offices are not much different today than they were a half-century ago. Word processing, for example, has been in offices since the 1980s. A visitor from that decade to a modern office “wouldn’t be surprised by that much,” Cappelli said.

He said experts in the subject generally agree that the pace of technological change in recent decades is slower than in the 1960s – with transistors, for example, and sweeping advances in chemistry. The pace was even quicker in the 1910s, with telephones, radios and automobiles transforming business in fundamental ways. “Now that was dramatic technological change,” he said. The social-media revolution of recent years, by contrast, “didn’t change the way people live.”

Many innovations fail to catch on because they turn out to be too expensive or too hard to use. One example: VCRs. Once they were expected to eliminate TV commercials because viewers would skip over commercials. But most consumers could never figure out how to program their machines, Cappelli noted.

That’s not to say that software hasn’t transformed business, of course. Certain innovations have been particularly important to HR: file-sharing technologies that allow outsourcing, for example. Others include job boards and their successors, enterprise resource planning programs and LinkedIn as a tool to find candidates.

But Cappelli urged caution in adopting platforms that promise dramatic results from “big data,” “machine learning”or “predictive analytics.” Those techniques have value in some settings, Cappelli said. But deriving valuable insights from a complex analysis of HR data often will cost too much and take too long. In the end “you might find something – you might not.”

Cappelli’s prescription is for HR leaders to focus first on what problem they are trying to solve, and get the data needed for a straightforward solution. Instead of focusing on employee engagement under the assumption that engagement improves performance, for example, employers might be better advised to just study what characterizes high-performing workers, he said.

Solutions that allow organizations to standardize, organize and simplify their data often make sense, Cappelli said. One example: Switching from a rating-based performance management system to one based on frequent manager check-ins cries out for technology to properly organize records of those conversations.

Cappelli also thinks dashboards are valuable – digital tools to measure what is happening with a workforce in real time. They can give HR leaders early warning of trouble or important trends.

In the end, Cappelli says, employers need to take a back-to-basics approach before splurging on advanced analytic capacities with uncertain potential.

“If you want to spend some money, you want to spend it first figuring out your own data, figuring out who is a good employee,” he said. “If you don’t have that, ‘machine learning’ isn’t going to do anything for you.”