In this column, Coursera President and Co-founder Daphne Koller asks Riley Newman, Airbnb's head of data science, to share his thoughts on the interdisciplinary values of understanding and utilizing data. Newman helped assemble coursework for Coursera's data science specialization with Duke University.
With 10 years of service in the U.S. Coast Guard, Riley Newman isn't the first person you might think of as a data geek. But for the head of data science at Airbnb, whose background also includes a Master's in economics and a stint at the United Nations, taking an unconventional career path has proved to be an invaluable asset.
Among the first ten hires at Airbnb in 2010, Newman has played a critical role in both helping the company become the global juggernaut that it is today and developing its well-known culture of creativity and collaboration. He's also a generous mentor who, along with a few colleagues, contributed real-world assignments to coursework in Coursera's Data Science Specialization.
Shortly after Airbnb was named Glassdoor's 2016 "Best Place to Work," I sat down with Newman to ask how he built a world-class multidisciplinary team--and why he thinks tomorrow's executives won't succeed without number-crunching skills.
You've said that data science is "the new MBA." Can you explain?
Well, the forces of globalization and technological development are making business increasingly competitive. An entrepreneur in Nairobi can compete with a mainstream American company for a customer in Thailand. That customer will choose the service that's best suited for them, which depends upon the way the competing companies use their data. So it will be harder and harder for companies--and executives--to succeed without an understanding of how to use data. The truth is, the gap between curating the information necessary to drive decisions and actually making the decisions is very small.
At the highest level, what do you see as a data scientist's biggest job responsibility?
It's a term (and industry) that's still being defined. Some focus on analysis to drive business decisions, others on constructing products that let users help each other (think: machine learning). Here at Airbnb, we do both. The idea is that as our community of guests and hosts grows, and we become more dependent upon data to stay connected with them--so understanding and responding to their needs effectively requires increasingly sophisticated methods.
And what does that look like on a day-to-day basis at Airbnb?
Every endeavor in the company now has a cross-functional leadership team that includes a senior data scientist. The idea is to ensure that data influences strategy in the right way--and that projects begin with concrete goals and metrics, rather than tactics. Then we conduct research to identify the biggest opportunities to achieve that goal, choose a tactic (often a data product), then launch a controlled experiment. If the results are positive, we scale up our work to the entire community of Airbnb guests and hosts; if not, we iterate. It's a process we've found useful not only on the product team but across the business --marketing, operations, etc.
Unlike many data scientists, you don't have a computer science degree. Has that affected your approach to hiring?
Probably. We have more social scientists than other data science teams, which works for us because Airbnb is a social business. More importantly, the field of data science is still young, so there isn't yet a consensus on which academic backdrop is optimal. Recruiting from a wide range of backgrounds has enabled us to scale our team with the company, and the breadth of backgrounds brings a breadth of perspectives, which enhances our creativity.
What are the essential skills you think data scientists today really need to be successful?
I think there's a misperception that you have to have a PhD in math or statistics to be a data scientist. That level of technicality is certainly useful, but I think you can go a long way if you understand three things: how data is collected, stored and retrieved; how to manipulate data using analytical tools like R and Python; and how to tease apart correlation and causation. Beyond those basic skills, the divide between good and great data scientists is most visible in their soft skills: communication, creativity, and curiosity. If you have these soft skills and are open to continually learning the technical skills, you'll be able to evolve with the industry.
And what skills do you see the best junior hires developing over time?
Data science is as much art as science. I think of the science as a prerequisite, while the art is something you learn on the job. In my experience, the art refers to the soft skills, and also relates to judgment. When is simplicity more useful than complicated methods? What is the highest-leverage question to investigate? And how much faith is appropriate for your underlying data? Judgment typically comes with expertise, which just takes time.