01 Sep 2015

Data-Driven Diligence

A number-crunching VC fund skips the pitch
by Francis Storrs


Illustration by QuickHoney

On the surface, there is seemingly nothing sexy about a startup that sells foam mattresses online. But in January 2014, David Coats (MBA 1992) and Trevor Kienzle (MBA 1991) made a significant seed-round investment anyway after just 12 days. It wasn’t an elevator pitch that sold them on Casper—it was the numbers. The company had tested well in a revolutionary predictive model built by the venture firm.

The idea to vet possible investments in such a way hit Coats about a decade ago: Maybe, he figured, he could use data analysis to quickly find undervalued startups—the same way, he’d later learn through Moneyball, that baseball teams were using analytics to find undervalued players. The problem was, Coats needed data—lots of it. “I kept asking VCs and others in the industry, ‘Has anyone ever aggregated venture capital financing data and looked for patterns?’” he recalls. After a fruitless search, Coats gave up on asking and partnered with Kienzle and others to dig up the data themselves.

In 2006, Coats enlisted a board of advisors, including HBS associate professor Matthew Rhodes-Kropf, and quit his VC job to launch Correlation Ventures. Kienzle joined soon after. While Coats and Kienzle had both been VCs for a decade, the company didn’t make a single investment in its first four years. Instead, the pair met with many other VCs and entrepreneurs and negotiated nondisclosure agreements with data providers. The two eventually aggregated data on more than 60,000 financings dating back to 1987, representing some 98 percent of the pool of deals in that period. Coats calls it “the most complete and accurate database of US-based venture capital financings in the world.”

In 2010, Correlation completed a first closing of their fund and began investing. By 2011, it had raised $166 million, allowing their unique model to finally take shape. While a traditional VC might back 15 to 20 companies over a decade, Correlation’s algorithms helped them make that many investments in as little as six months. The company keeps its role simple, coinvesting behind experienced lead VCs and never demanding board seats.

Most important of all, Correlation swears total allegiance to the FICO-like score produced by its analytics. “There are opportunities that we, with our classic VC hats on, are like, ‘Oh, I love this deal,’ but then it doesn’t get past the discipline of the analytics,” Coats says. He and Kienzle can say no in as little as two days—traditional VCs take months—and always make investment decisions within two weeks or less.

To date, Correlation has invested in an enormous portfolio of 112 companies, and expects to get to 120 in the first fund before investing from a second. It’s also made two additional investments in Casper, which is proving wildly popular. The model worked as intended, says Kienzle, by recognizing “a promising investment opportunity in a sector—mattresses and bedding—that had not received much VC attention, and most people would not have considered ‘hot.’” Same with the storage software company Virsto, which Correlation put $1 million into in 2011. When it was later acquired for $200 million, Correlation’s bet paid off handsomely.

Big windfalls are nice, but the real secret to the Correlation model is that it’s built to accumulate smaller successes. “It’s really designed to tilt the odds in our favor a little bit, kind of like counting cards in blackjack,” Coats says. “If you play enough hands, you should win.”

Featured Alumni

Featured Alumni

Class of MBA 1992, Section E
follow @CoatsDavid
Class of MBA 1991, Section C
follow @TrevorKienzle

Post a Comment