When Experts Pivot, They Pay a Price

For some of us, the rapid rise of artificial intelligence might not seem the least bit worrisome. We’re learning about it and testing it. And we’re slowly but surely integrating the new technology into our lives and work. In other words, we’re adapting—just as we did with the emergence of the internet and smartphones and mRNA vaccines.

Yet pivoting to new territory can actually be more taxing than we realize, prompting questions both for individuals and for organizations, says Benjamin Jones, a professor of strategy at the Kellogg School, where he codirects the Ryan Institute on Complexity.

“At the individual level, ‘How will it go when I tackle a new area—do great things happen, do I fail, or is it the same as before?’” he says. “From an organizational or societal level, ‘How can we pivot our human resources and expertise to engage a new area?’”

Jones collaborated with Kellogg’s Ryan Hill and Dashun Wang, PhD student Xizhao Wang, Yian Yin of Cornell University, and Carolyn Stein of the University of California, Berkeley, to explore these questions in the context of science research and technological inventions.

They developed a method to determine what happens when scientists shift gears and research topics beyond their typical area of focus, or when inventors and organizations create products outside of their wheelhouse.

After applying their method to millions of research papers and technology patents over a roughly five-decade period, the team found that shifting directions significantly diminished the impact of the resulting papers and patents. This outcome, which the team termed the “pivot penalty,” not only affected nearly all fields of research and classes of patents but also became more severe over time.

“The pivot penalty that we document might come as a surprise to many people,” says Hill, an assistant professor of strategy. “We often think that outsiders could bring new insights to a topic that might facilitate novel breakthroughs. Instead, we find that there are real barriers to success, and we try to study those forces from many angles in the paper.”

“Exploration has always been an important part of scientific work,” adds Wang, “but our data show that when researchers move too far from their core expertise, they face steep penalties.” Wang is a professor of management and organizations and the Chair of Technology at Kellogg, where he also directs the Center for Science of Science and Innovation (CSSI) and the Northwestern Innovation Institute, and codirects the Ryan Institute on Complexity.

The pivot penalty

In studying the sciences, Jones, Hill, Wang, and colleagues examined 26 million research papers from 1970–2015 across 154 fields.

They determined the focus of a paper based on the type of research journals it cited in its list of references. Prior research published in American Economic Review, for instance, would be categorized as economics, whereas research published in American Political Science Review would be political science. Then they quantified the pivot of a paper on a 0-to-1 scale, based on how closely the categories of its references aligned with those in the same researcher’s prior papers.

“If the distribution of journals you cite in a new paper is exactly the same as the distribution in your prior work, that’s a pivot of zero,” Jones says. “But let’s imagine that none of the journals you cite in your new paper are in any of your prior work. You’re doing something completely different; that’s a pivot of one.”

The team also measured the impact of each paper based on whether it was in the top 5 percent of most-cited papers within its field in the year of its publication. Then they input this whole framework into an algorithm to calculate the pivot size and impact of all papers.

They found that the greater the pivot of a paper—or the further a researcher moved away from the area they had previously researched—the less likely it was to be a high-impact paper.

“The further the pivot, the worse it seems to go,” Jones says. “It’s not that you can’t enter a new area and hit a home run, but there’s just a far, far lower chance of that happening.”

To be specific, papers requiring the least amount of pivoting became high-impact papers 7.4 percent of the time, compared with 2.2 percent of the time for papers requiring the most pivoting. The papers with the biggest pivot were also 43 percent less likely to be cited by patented inventions and 35 percent less likely to go from a preprint paper to a journal publication.

Within a given researcher’s portfolio, the paper requiring the smallest pivot was 40 percent more likely to be a high-impact paper than the rest of the researcher’s work. In contrast, the paper with the biggest pivot was 36 percent less likely to be a high-impact paper.

Ultimately, the pivot penalty persisted regardless of factors such as the researcher’s career stage or individual productivity, project team size, use of new coauthors, and funding.

A central tension

A simple response to these findings might be to just stop pivoting. After all, as science and technology continue to advance, it is becoming increasingly difficult for researchers and inventors to keep up with the growing pool of information. This burden of knowledge compels many of them to narrow their area of focus within their already-specialized field.

“As people are getting more specialized, the disadvantage of moving—the penalty for pivoting—is getting worse,” Jones says. “And the advantage of staying in your area is growing.”

This finding also applies to the development of new technology.

The team assessed the impact of 1.8 million technology patents from 1980–2015 across 127 technology classifications. Like in the case of research papers, pivoting had a negative effect on the impact of patents. Patents that required the least pivoting turned out to be high-impact patents 8 percent of the time, compared with 3.8 percent of the time for patents that required the most pivoting.

In addition, pivoting caused a new patent’s market value—as measured by how a company’s stock price responded to the patent—to decrease steeply. Patents that required the most pivoting had a 29 percent lower market value than those that required the least pivoting.

And yet, despite the clear consequences of pivoting, the ever-changing nature of the real world can make it nearly impossible to stop pivoting altogether. External events, such as the Covid-19 pandemic or the rise of AI, continue to draw people’s attention, if not demand a response.

“If you’re doing a huge pivot to Covid or AI, it’s likely going to go badly,” Jones says. “But at the same time, it’s attracting you because there’s this demand premium and it’s potentially very important. It’s going to be a slow start, but you might need to pivot anyway.”

“Understanding these trade-offs is crucial,” Wang adds, “if we want to build a research ecosystem that is both resilient and responsive to emerging challenges.”

Tackling bigger questions

The need to pivot—and how best to approach it—is not unique to science and technology. It’s a dilemma that organizations, governments, and society as a whole face all the time.

Sometimes organizations stick to their guns and choose not to adapt at all. In other situations, they opt for a full pivot, dropping their current projects and shifting their priorities. Still other organizations have turned to acquisitions, buying out smaller companies at the forefront of a particular area to obtain not just their intellectual property but also their talent.

In other words, “they get into a new area by aggregating expertise—hiring or collaborating with people who are experts in that area,” Jones says. “Drawing in those experts can limit the collective pivot penalty and unleash greater advantage.”

Organizations across many industries have been operating this way for years. Academic institutions bring together students, scientists, and other specialists from a wide range of backgrounds to lead its research. And businesses often call on a varied pool of consultants to help guide big decisions.

The ability to pivot well is particularly critical when emergencies force people to adapt unexpectedly.

Such was the case with the Covid-19 pandemic. If structural biologists, immunologists, and virologists were not already well-versed in coronaviruses, and if vaccine makers had not already dedicated years of research and funding to developing mRNA vaccines (which at the time might have seemed like a failed technology), then the pandemic could have lasted even longer than it did. “We’re very lucky, in a sense, that with Covid we had the pre-positioning of important types of human capital,” Jones says.

Whether or not individuals or organizations are currently dealing with an emergency, “the apparatus of research and development in the U.S., and in the world, needs to spread out,” he says. “We need to have people who are experts in different areas so that we are prepared to handle emergent challenges and can tackle bigger questions.”