Western Digital is pleased to be a supporter of the Global Women in Data Science (WiDS) Podcast. This new series features women leaders across the data science profession, as they share their advice, career highlights, and lessons learned along the way.
Leading a team of over 100 data scientists at one of the world’s most valuable startups wasn’t always in Elena Grewal’s plan. As a kid growing up in New Haven, Connecticut, she recounted her early struggles in public school to be a top student. The turning point in her academic career was in middle school during a meeting with her teacher and parents. “She’s a really nice kid and going to be fine in life, but she’s never going to be a top student,” her teacher had said.
So, the instructor had thought. With her father, a university professor, Elena worked tirelessly to hone her abilities in mathematics and writing. It paid off: her grades soared and she ended up graduating at the top of her class.
Now, just seven years after finishing her Ph.D. in Education from a world-leading research institution, she finds herself as the Head of Data Science at one of Silicon Valley’s well-known unicorn companies: Airbnb. She recently returned to her alma mater to share her journey into data science on the Women in Data Science Podcast. In this article, we list four highlights from her interview.
To hear Elena’s full interview, listen to the podcast episode below!
HOST: Did you see a natural connection between your Ph.D. in Education and your work as Head of Data Science?
ELENA: Yeah, one of the things that I love about data science is that often your job is to make connections. That’s what happened here! I had this background and degree in education. Being in my university and Silicon Valley, I knew many people who were in tech. It was the end of my Ph.D. and I was thinking about going into academia, and I was excited about that.
But, I also thought, “I’m here. It might be interesting to look at these tech companies and maybe do a summer internship.” I got connected to a friend of mine who was in recruiting and I was like, “What should I do?” She said there’s this thing called “data science”, which I’d never heard of before. I thought, “Oh, that’s interesting.” Then, I started talking to people and went to my school’s Career Services…. Their advice was just to talk to people and connect that way instead of sending out my resume and cover letter.
That proved to be really helpful so that I could understand “what is data science” and also get connected to the company I’m at now. I had used the product before. But, it was really the interview process, walking through the door, and meeting the people there that inspired me. They made it clear that my background was a perfect fit for this new field… My training was in statistics and econometrics. I was really doing “data” work and looking at data in the context of education. At the same time, I was applying this lens of “How do I collect the right data? How do I analyze it and make sense of it?”
That skill was exactly what I’d be doing as a data scientist in industry. It was also a little bit of luck, too, to make that connection. At the time, people were confused about my life choice. My parents, in particular, were like, “What is going on? We’ve never even heard of this company. We can’t even pronounce it.”
So, as much as I use data to make decisions, I do believe that there’s data in your gut. And I truly believed that it was going to be a great adventure that I wanted to go for. I couldn’t put my finger on exactly what it was. But, it was kind of a combination of all the inputs and that feeling you have a path forward and it just feels right.
HOST: What are some of the data science challenges that come from operating in so many different countries?
ELENA: Some of the challenges come from all the different ways that you can cut the data. When you launch something, you want to know who is going to be impacted by it and in what way. Different countries have different norms and practices. So, people might respond in different ways to what you’re launching. That’s one challenge: How do you look at the data in the right way and capture those differences?
Another question is around your ability to generate hypotheses, if you don’t understand that other culture and how people from a particular group might be using the product. If you don’t have that knowledge, you might not be able to ask the right question to know how to look at the data.
For the first challenge, we solve it by automating a lot of the ways that we segment our data. That way, we can surface the cuts or geographies that we see differences. That’s something that we build into tools that we have. So, you can make easy slices of the data and see if this difference has the same impact in France as it does in Australia and find out what’s going on there.
For the second point, this is where we’ve really pushed on the diversity of our data science team. When we think about what kinds of questions we want to ask, we need to make sure we have different perspectives asking those questions. That way, we can really get that global view of our data and not exclude certain populations or ask questions in a way that might not benefit all of our users.
HOST: Can you give an example of something you learned as Head of Data Science that wasn’t as successful as you thought it would be?
ELENA: One example that was a pretty surprising result early on was with one of our operations teams. Coming into data science from education kind of gives you a perspective that there’s not one set way that data science should work. When I started, it was like anything can benefit from looking at data in a more rigorous way.
So, we looked at our operations teams that had been expanding globally. We were in kind of a battle with some companies in Europe and had a lot of people on the ground. One of the things that those teams was doing was calling hosts and giving them suggestions on how to improve their listings. This was something that we hypothesized would be really great and would change the quality of the listings.
If you had looked at the listings of the people who had been called compared to the people who had not been called, the homes of the people who were called were better. They generally looked better and got more bookings. You might have concluded that the calls had caused the increase in bookings. But, we were wondering if that was really true.
So, we ran an experiment. And we found that the phone calls were really not making much of a difference at all. Part of the reason that we were seeing that difference was the people who were being called and picking up the phone were usually better hosts already. Those calls weren’t really having the impact that we had hoped for. That was a very surprising result, where an experiment was so helpful because we could actually see this data and see the causal impact of what we were doing.
HOST: What are you excited about in the data science field?
ELENA: This is such an exciting moment of revolution! In the past, we really didn’t have access to this kind of data. That’s incredible and a reason why this field is growing. All of a sudden, we can ask questions that we couldn’t ask before and have tons of data at our fingertips to test hypotheses. That’s really where the science comes in to take what we think is true and test it.
In the same way, some of the more traditional sciences—such as biology or physics—the availability of data or certain technologies led to breakthroughs. It feels like that’s where we are with data in this social world with how people are using digital products. That’s really cool.
The other thing that gets me really excited is that people are looking at what is true and trying to understand that using data. One thing that has been so cool at my company is that we’ve launched this program called Data University. It’s built to empower every Airbnb employee to look at data themselves.
One thing that sometimes scares me is that I don’t want people who have data to be the keepers of knowledge or power. Instead, I want them to share that knowledge and enable people to think more critically and make conclusions themselves… I think this energy around data science will translate to helping us understand statistics and probability in our lives a little bit better.