Certain job skills have the potential to shake up industries—and understanding them is critical to managing an organization’s talent.

In a new research report for IBM and the Business-Higher Education Forum, Burning Glass Technologies developed a matrix for identifying these “disruptive skills” and getting ahead of them in the hiring market.

As new skills arise, businesses often scramble to find employees with the needed know-how. In cases like this, there isn’t an established training pipeline (college degree programs, boot camps, certificates) so it’s hard for employers to find those workers—or to know if those candidates really know what they’re doing.

Data analytics is one timely example. Back in 2011, in the early era of Big Data, college programs on data analytics were an outgrowth of statistics programs, and were likely to produce the kinds of “quants” who work on Wall Street.

Since then, demand for analytics skills has both increased and changed dramatically. The report projects that there will be 2.7 million data analytics job openings by 2020, up 15% from today. The most significant increases won’t be for data scientists or data engineers, but in data-driven decision making. In other words, the real demand now isn’t just for statistical analysis, but for applying data analysis to business problems.

In our report, we identified four different types of skill sets, based on how fast demand is growing and how much trouble employers have finding them. To map them out we created a matrix of skills:

Disruptive skills matrix

Stabilizers are the skills most employers don’t have to worry about: they’re fairly common in the workforce, there are enough training programs available, and being slow to fill them doesn’t jeopardize the company. In the data analytics world, SQL is a good example of a stabilizer skill.

Escalators aren’t that difficult to find, but they’re growing fast, so there’s a need to step up training and get more candidates in the pipeline. A lot of practical common applications, like Google Analytics and Tableau, fall into this category.

Challenger skills are in demand and costly to find, but they aren’t groundbreaking, either. There are usually established training programs. Failing to find challenger skills could leave important tasks undone. Among data analytics skills, Data Management, Data Governance, and Database Schemas are all Challengers.

Disruptors are skills that are costly to hire, don’t have a lot of training options, and are critical to a company’s growth. Not only are they hard to find, but if you can’t find them, you’re in trouble. Apache Hadoop, Machine Learning, and Data Engineering fall into this category.

All of these skills can pose a problem in hiring, but Disruptors are the real issue. These skills are in demand in a broad context of a fast-growing field without a lot of training programs in place. Frequently, these skills are a force in creating “hybrid jobs” that mix skill sets from different occupations.

Being able to sort out these skills in a particular industry allows managers to plan more effectively and to avoid playing catch-up. This is why talent management is playing a greater role in strategic planning. No manager likes being surprised.