Want to be an Entrepreneur? Get a job.

Machine Learning says so.

The resounding success stories of the Mark Zuckerbergs, Steve Jobs and Bill Gates of the world shaped the idea of who successful entrepreneurs are: Genius college drop outs that built their billion dollar company in their dorm room.

Does this mean that you should stop studying for your final exams, and run with that idea your roommate and you had during your weekly pizza night?

Does this mean that you don’t stand a chance because you've been employed and out of college for 10 years?

No.

Our research shows that most successful early-stage entrepreneurs have 10 to 12 years of employment, prior to founding their company. The myth of the genius drop out is exactly that, a myth.

Not only are 94% of US industry and government leaders, college graduates (Wai, Jonathan, and Heiner Rindermann What goes into high educational and occupational achievement? Education, brains, hard work, networks, and other factors. Duke University: 2017), but we also discovered using Machine Learning techniques that the most predictive factor of early stage success for start-up founders is how long they have been employed before founding their company.

“Wait a minute. What do you mean by early stage success?”

This is a good question. Success is a subjective metric and there are as many definitions as there are humans on earth. However for the sake of clarity (and to make the task a little less daunting for us) we defined success as the ratio between the company’s valuation at Series B over it’s valuation at Series A:

Success = (Valuation at Series B) / (Valuation at Series A)

Using this as our measure of early-stage success, we proceeded to rank 16 different characteristics of start-up founders with regards to how useful they are in predicting that ratio.

To do so we used different Machine Learning techniques, assigned points according to the characteristics’ ranking with each technique and compiled all the points. We ended up with the following table:

As you can see, the most useful thing to look at to predict whether a founder will be successful at early-stages or not is how long they have been employed prior to starting their company.

“What does it translate in terms of success ratio?”

I’m glad you asked. After we got these results we fitted a polynomial model into our data set and here are the results:

Founders that have had 10 to 12 years of prior employment tend to have the highest success ratio. It is over 50% bigger than that of people with no prior employment.

Intuitively, this makes sense. With longer periods of employment come increased experience, a sharpened awareness of real industry problems, as well as a strong and wide network.

As with all studies, there are limitations to our results. However, we believe our approach to this entrepreneurship research as well as the robustness of the method can justify further research to be undertaken in this direction to validate, or dismiss, by utilizing a more comprehensive data set.

Today, in the age of Big Data, there is an exponentially growing quantity of information available to all. This allows new research methodologies that are more quantitative in nature to be applied to new fields of research. Machine Learning and Data Science can offer entrepreneurship research a new set of tools to showcase patterns and results unseen before. Only a decade ago studies like these were impossible to conduct.