好文 | 预测：谁是潜在创业者？
关键词：风投 | 数据
如果科技公司可以用数据寻找消费者，那么风投也可以用之寻找创业者，甚至是潜在创业者。去年，早期投资公司 Bloomberg Beta 给湾区、纽约两地的 350 位潜在创业者发送邮件，先于其他风投基金直接触达他们。
Bloomberg Beta 认为以下经历将增大某人成为 Future Founders 的概率：（1）曾经在风投投资过公司工作；（2）接受过足够教育，但不用太多；（3）有多年工作经验；（4）早在风投投资之前，就开办过自己的公司。
提示：本文由 Bloomberg Beta 负责人 Roy Bahat 撰写，如果你能顺利打开链接，请在他的博客阅读 ：）
Predicting who will start a company
A new class of Future Founders
Last year, we tried something new at Bloomberg Beta. We' re an early stage fund that invests in making business work better: one way we do this is by taking new techniques from technology companies, and applying them in our own work as investors.
Founders are our customers. Successful technology companies use data to find customer leads before others — could we do the same to find founders? Could we predict who would start venture-backed companies? Sitting at Bloomberg, the appeal of using a new data set to understand a market that once lacked data was clear — that’s what Bloomberg’s been doing for the securities industry for 30 years. (And of course Bloomberg itself was once the kind of startup that Bloomberg Beta would, today, be thrilled to back.)
Last year, our project identified 350 potential Future Founders in the Bay Area and New York (where we focus our investing) — and we reached out to them. When these Future Founders got our email, some thought it was a scam (“if this is a scam, it is the most sophisticated I have ever seen.”)
The group was surprising and remarkable. They were different from the stereotype of a typical VC-backed founder. A more diverse group in every way, and — because the data don’t understand discrimination — a group that the “startup scene” had often missed.
But did last year's class actually become founders?
Short answer: yes.
Longer answer: The first time we met our Future Founders we confirmed we’d repeat this experiment. We built relationships with talented people we’d never otherwise have known — who would be largely invisible to others because they had yet to put their hand up and say “I’m ready to start a company.” They’ll do great things regardless, and maybe work with our startups or bring their friends to us when they want to start something — finding a potential future founder is a needle in a haystack search, it’s just that in this case we believe all the hay is made of gold.
Many found the project an interesting curiosity last year, which we were honored by — we’re now happy to update everyone that there is meat on the bones. The inaugural class is already going places. Eight have started companies, and three are already venture-backed (including Product Hunt, and an unannounced company in which we just invested). Half have changed roles in the last year, and 40% changed companies — kinetic talent.
Future Founders, class of 2015
We learned from the start we got last year, and refined our approach. This year, we worked with the Lester Center for Entrepreneurship at UC Berkeley’s Haas School of Business (where I teach a few weeks a year), led by Prof. Toby Stuart. Weiyi Ng, a PhD candidate there, conducted the analysis with assistance from People.Co, a talent data company.
They reviewed hundreds of thousands of public profiles of people in the technology industry, cross-referenced with data generously provided by AngelList. We analyzed more data this year on each founder, e.g., the number of job moves they’d made.
Many of the same predictors applied again this year, on steroids: for example, whether someone ever worked at a venture-backed company is, unsurprisingly, a major factor. And the more years s/he worked at a venture-backed company, the more likely to start his/her own.
Speaking of gender, we collected data on this for the first time this year: one in five of our predicted future founders are women. This is still shy of where we all want it to be, though about double the current rate of funding for female founders — so the future looks like it will be better than the past.
Other observations: Future Founders have enough education, not too much. Same with years of work experience. They are likely to have started companies in the past without being venture-funded—and unlikely to merely have been self-employed as freelancers. This year’s group was more technical — roughly half have technical education.
We believe our method is now 50x better than random at predicting who will start a venture-backed companies. Though, now that we’ve planted the bug, maybe a few more of them will be nudged in that direction… we’re honored to know and support them either way.