At our HowTo Conference, panelists taught entrepreneurs how to leverage data science in their startups.
By Vanessa Mason (Contributing Writer, Women 2.0)
Silicon Valley took over national headlines with sensational stories about the power of data science to understand how social media like Facebook can affect our emotions and data-driven experimentation via OkCupid can affect our love lives (or lack thereof). Data science, while obviously powerful and valuable, remains hard to understand and hard to leverage for many entrepreneurs. The second day of the Women 2.0 HowTo Conference set the tone for a lively conversation about data on Day 2 with a keynote by Monica Rogati (Jawbone’s VP of Data) all about the promise of data products. Later in the morning, a panel featuring four female technical powerhouses taught the audience how to get started using data science.
- Clare Corthell, a data scientist & designer at Mattermark, a data-driven deal intelligence platform, who began her career as designer before finding data science
- Elena Grewal, a data scientist at AirBnB, who stumbled into data science while completing her doctoral research on friendship networks in schools
- Anita Lillie who had the opposite career of Corthell, beginning first with math and computer science, and then shifting to design as senior data scientist at Practice Fusion, a digital health company.
How to Use Data Science
Use cases for data science in startups are as numerous and diverse as startups themselves. Corthell’s company relies on data science for mining the Internet for data to understand which companies are growing the fastest and the predictive analytics that can support inferences about fundraising, overall health and other key metrics.
Data science drives building better products for Grewal and her team at AirBnB by predicting factors that shape guest and host behaviors to lead to more meaningful trips for both. At Practice Fusion, Lillie leverages data science to understand how data can support health behavior changes. Her team uses data insights to inform design and messaging that communicate the rationale for recommendations.
How to Predict the Unpredictable
Although data science is all about finding patterns, each of the panelists has stories about finding unpredictable anomalies in their work. Lillie tried to examine the geographic incidence of flu one season only to find the data had 11,000 states listed rather than 50. After she dug deeper, she learned that an early bug in the product caused the anomaly, allowing users to input free text rather than selecting from a list. Figuring out data anomalies is a source of delight for Corthell because it demonstrates the power of machine learning to approximate human inferences.
While data science may eventually bring us a world where tech can know people better than they know themselves, it’s not perfect yet. Grewal anticipated a drop in AirBnB’s review rate once double blind reviews were introduced. The data science team was surprised to find that the opposite occurred as guests and hosts raced to complete reviews to see what was said about them.
How to Build and Grow Your Data Science Team
Most early-stage startups won’t have a data scientist on their team, but doesn’t mean founders shouldn’t think ahead about how to structure a data science team at a later date. Grewal started at AirBnB more insulated from other teams, but over time her team has shifted to a more embedded model with the product team to support more open communication. Lillie noted that embedding data scientists into the rest of the company can help break down barriers and misperceptions about what data scientists do and build internal support for their work. Integration of data science is a bit of a non-issue for Corthell since nearly every employee has been a data scientist at one point.
More From These Speakers
- Making Sense of Big Data: Without These People, Your Company’s Data is Useless
- Get to Know Anita Lillie, Senior Data Scientist at Practice Fusion
- Get to Know Data Science Panelist Elena Grewal, Airbnb Data Science Manager