Bayes Impact: Big data for good

In a scene in The Graduate (1967), Benjamin Braddock, the character played by Dustin Hoffman, was given the most famous bit of career advice since Horace Greeley advised readers to "go west, young man.""Ben. I just want to say one word to you. Just one word. Are you listening?...Plastics."Today, math-literate young people might need two words to point them towards a sizzling-hot sector of the economy: big data.Big data is a big deal, in the nonprofit world as well as in business. How else to explain the overnight success of Bayes Impact, a nonprofit formed less than 18 months ago by a couple of not-long-out-of-Berkeley grads that has already attracted about $2 million in support from foundations and clients including Y Combinator, the Bill and Melinda Gates Foundation, the Robert Wood Johnson Foundation, the Michael J. Fox Foundation for Parkinson's Research, LinkedIn for Good, Microsoft, Youth Villages, Teespring, the Tableau Foundation, the U.S. Department of Veterans Affairs and the government of France.Eric Liu, 24, and Paul Duan, 23, the co-founders of Bayes Impact, clearly are onto something. They describe themselves as "practical idealists who believe that, applied properly, data can be used to solve the world's biggest problems."

“We want to improve the ways that people get health care, education and housing services,” Liu told me, when we met for coffee in Washington. He'd just come from a meeting with the VA, to talk about how Bayes Impact can help the agency place unemployed veterans in jobs, by matching their military skills to the needs of the civilian job market. "It's really exciting."

Big data for nonprofits is a topic of conferences and courses and informal networks like Data Analysts for Social Good. It's the idea behind other nonprofits, too, notably DataKind ("harnessing the power of data science in the service of humanity"), which connects data scientists with nonprofits around the world. What's not clear--yet--is how it's going to make nonprofits more effective.

Andrew Means, the founder of Data Analysts for Social Good and associate director of the Center for Data Science & Public Policy at the University of Chicago, told me: “There’s a lot of interest in data, but not a lot of understanding, or many examples of how it’s being used well.”

San Francisco-based Bayes Impact -- it's named after a theorem of probability theory -- took root at U-C Berkeley, where Liu and Duan met as undergraduates. Liu, who grew up in Silicon Valley and majored in business, joined a venture capital firm after graduation, analyzing private companies that developing software for online advertising and financial services. Duan's family left China after the Tiananmen Square protests for France, where he was born. Trained as a data scientist, he worked on fraud detection at Eventbrite.

Both saw first-hand the power of data science and thought it could be put to better use. As Jeff Hammerbacher, a data scientist and early Facebook employee, once said, semi-famously: "The best minds of my generation are thinking about how to make people click ads. That sucks."

Liu and Duan began talking to people in the nonprofit and foundation world, as well as in Silicon Valley, and they volunteered to do a couple of projects for the San Francisco city government, about ambulance dispatch and housing inspections. One day, they got a call from Y Combinator, which incubates startup companies and, occasionally, nonprofits. They were offered a $100,000 donation, the chance to join a intensive, three-month course in how to start and grow a company and, perhaps just as important, access to YC's well-connected alumni network..

Through Y Combinator, they met Zidisha, an online lending nonprofit, which they helped with fraud detection. The Gates Foundation funded them to rapidly evaluate educational video games.They worked on a organ transplant project for the US Department of Health and Human Services. They worked with the state of California on a data-driven approach to reduce prison overcrowding, by identifying better candidates for parole, and they are part of a Robert Wood Johnson Foundation initiative aimed at reducing hospital readmissions.

It all sounds a little scattered to me, I told Liu. He didn't disagree. "We've worked across a whole lot of different areas," he said. Demand for their work has exceeded their time. Bayes Impact now has eight employees, and is hiring six more. And, Liu told me, the nonprofit now intends to focus on the biggest providers of social programs in the world--governments. "We've been thinking a lot about scale," he said.

So Bayes Impact is evolving -- or should I say pivoting? -- from a nonprofit consultancy into a product-driven NGO, primarily aimed at building digital social services for governments. "With one product, if it's the right product, we can help a whole lot of people,"Liu says. Their first product seeks to improve the performance of the French government's national unemployment agency, by better matching job-seekers with job openings. A different version of the product will be deployed later this year by the VA.

Meantime, Bayes Impact continues to gather momentum. The group was awarded a $100,000 prize last month as a runner-up a competition of social entrepreneurs that drew 2,500 entries at the Forbes Under 30 Summit. My guess is that you'll be hearing more about Bayes Impact in the years to come.