Less Noise but More Money in Data Science

The outlook for data scientists: less hype, more hiring.

The exuberance surrounding big data has passed its peak and is trending down, the technology research firm Gartner declared last August in its annual “hype cycle” report on perceptions of technology.

Perhaps, but it remains a rising market for data scientists. Salaries rose 8 percent on average in the last year, with bonuses adding $56,000, according to a salary and employment survey released on Tuesday by Burtch Works, a recruiter of professionals with quantitative skills.

The median base salary for entry-level data scientists was $91,000 nationally and $110,000 in Silicon Valley, the survey found. For experienced data scientists, leading teams of 10 or more, salaries are up to $250,000. The report was based on telephone interviews with 371 data scientists

The survey also points to a shift in demand for data scientists. The share of workers with deep data skills who are employed by start-ups fell by more than half, to 14 percent this year from 29 percent last year. Linda Burtch, managing director of the recruiting firm that bears her name, said the change reflected how much companies in mainstream industries like retailing, consumer products, insurance, health care and manufacturing were building up data science capabilities.

The appeal for data scientists, Ms. Burtch said, is “you get to work on really good problems, and these established companies can offer them a decent work-life balance.”

Khalid Ahmed, 25, is finishing up his Ph.D. at the University of Michigan. His field of computational physics and materials science, Mr. Ahmed said, has increasingly become a data science — and he has veered in that direction. He was recruited by several companies including start-ups. One analyzed credit-card data to target marketing and advertising campaigns. Instead, Mr. Ahmed chose to join Ford Motor as a data engineer in its hybrid car program.

With start-ups, Mr. Ahmed explained, the work often involves producing data-derived knowledge that is then sold to a larger company. But at a large manufacturer like Ford, he said, “I’ll be influencing the design of the product itself.”

In Charlotte, N.C., Ahmer Inam, 39, recently left Sonic Automotive, a large auto retailer with 100 dealerships in 14 states, where he was director of predictive analytics. He enjoyed the work, but ultimately he decided he wanted to move beyond tailored marketing. “Our work was designed to seduce a customer toward a particular vehicle and determine the optimal price,” he said.

When he decided to leave, Mr. Inam had several offers from traditional companies and start-ups, and he chose the consulting firm PricewaterhouseCoopers, which he joined last month. Mr. Inam was attracted by the prospect of working in a variety of industries and the opportunity to prod organizations to adopt the tools and methods of data science. The consulting firm, he said, also encourages its professionals to use their skills on nonprofit community projects. At PricewaterhouseCoopers, Mr. Inam said, “I can have a work-life balance that allows me to engage in analytics for social good initiatives.”

There is an apparent contradiction between the buoyant job market for big data practitioners and Gartner’s judgment that, on the perception scale, big data has moved from high expectations to what Gartner calls the “trough of disillusionment.” But, in fact, it fits a familiar pattern of technology absorption and use. Significant new technologies always take time to move into the mainstream as people and organizations learn to exploit them. It takes years.

The classic study of the phenomenon, “The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox,” by Paul David, an economic historian at Stanford University, was published in 1990. In it, Mr. David noted, the electric motor was introduced in the early 1880s, but its real payoff in productivity was not evident until the 1920s. It took that long for businesses to reorganize work around the industrial production line, the efficiency breakthrough of its day, made possible by the electric motor.

Similarly, it took a while for personal computers and the Internet to deliver big gains. And so too for big data, which harnesses computing, modern digital data and the software tools of artificial intelligence.

A report this week from Forrester Research described the challenge ahead. “Businesses are drowning in data but starving for insights,” the report began. “Worse, they have no systematic way to turn data into action.”

The Gartner hype cycle acknowledges that enthusiasm precedes mainstream moneymaking. The graphic representation of the cycle looks like an exaggerated, lazy wave. After the disillusion phase comes the “slope of enlightenment” and then the “plateau of productivity.” As Merv Adrian, an analyst at Gartner, put it, “If you want to see the revenue side of the hype cycle, you basically just flip it.”