Editor’s note: Aileen Lee is founder of Cowboy Ventures. a seed-stage fund that backs entrepreneurs reinventing work and personal life through software. Previously, she joined Kleiner Perkins Caufield & Byers in 1999 and was also founding CEO of digital media company RMG Networks, backed by KPCB. Follow her on Twitter @aileenlee . 

Many entrepreneurs, and the venture investors who back them, seek to build billion-dollar companies.

Why do investors seem to care about “billion dollar exits”? Historically, top venture funds have driven returns from their ownership in just a few companies in a given fund of many companies. Plus, traditional venture funds have grown in size, requiring larger “exits” to deliver acceptable returns. For example – to return just the initial capital of a $400 million venture fund, that might mean needing to own 20 percent of two different $1 billion companies, or 20 percent of a $2 billion company when the company is acquired or goes public.

So, we wondered, as we’re a year into our new fund (which doesn’t need to back billion-dollar companies to succeed, but hey, we like to learn): how likely is it for a startup to achieve a billion-dollar valuation? Is there anything we can learn from the mega hits of the past decade, like Facebook. LinkedIn and Workday ?

To answer these questions, the Cowboy Ventures team built a dataset of U.S.-based tech companies started since January 2003 and most recently valued at $1 billion by private or public markets. We call it our “Learning Project,” and it’s ongoing.

With big caveats that 1) our data is based on publicly available sources, such as CrunchBase, LinkedIn, and Wikipedia, and 2) it is based on a snapshot in time, which has definite limitations, here is a summary of what we’ve learned, with more explanation following this list*:

Learnings to date about the “Unicorn Club”:

We found 39 companies belong to what we call the “Unicorn Club” (by our definition, U.S.-based software companies started since 2003 and valued at over $1 billion by public or private market investors). That’s about .07 percent of venture-backed consumer and enterprise software startups.

On average, four unicorns were born per year in the past decade, with Facebook being the breakout “super-unicorn” (worth >$100 billion). In each recent decade, 1-3 super unicorns have been born.

Consumer-oriented unicorns have been more plentiful and created more value in aggregate, even excluding Facebook.

But enterprise-oriented unicorns have become worth more on average. and raised much less private capital, delivering a higher return on private investment .

Companies fall somewhat evenly into four major business models. consumer e-commerce, consumer audience, software-as-a-service, and enterprise software.

It has taken seven-plus years on average before a “liquidity event” for companies, not including the third of our list that is still private. It’s a long journey beyond vesting periods.

Inexperienced, twentysomething founders were an outlier. Companies with well-educated, thirtysomething co-founders who have history together have built the most successes

The “big pivot” after starting with a different initial product is an outlier .

San Francisco (not the Valley) now reigns as the home of unicorns .

There is very little diversity among founders in the Unicorn Club.

Some deeper explanation and additional findings:

1) Welcome to the exclusive, 39-member Unicorn Club: the Top .07%

Figuring out the denominator to unicorn probability is hard. The NVCA says over 16,000 internet-related companies were funded since 2003; Mattermark says 12,291 in the past 2 years; and the CVR says 10-15,000 software companies are seeded each year. So let’s say 60,000 software and internet companies were funded in the past decade. That would mean .07 percent have become unicorns. Or, 1 in every 1,538 .

Takeaway: it’s really hard, and highly unlikely, to build or invest in a billion dollar company. The tech news may make it seem like there’s a winner being born every minute — but the reality is, the odds are somewhere between catching a foul ball at an MLB game and being struck by lightning in one’s lifetime. Or, more than 100x harder than getting into Stanford .

2) Facebook is the super-unicorn of the decade (by our definition, worth >$100B). Every major technology wave has given birth to one or more super-unicorns

Facebook is what we call a super-unicorn: it accounts for almost half of the $260 billion aggregate value of the companies on our list. (As such, we excluded them from analysis related to valuations or capital raised)

Prior decades have also given birth to tech super-unicorns. The 1990s gave birth to Google, currently worth nearly 3x Facebook; and Amazon, worth

$160 billion. The 1980’s: Cisco. The 1970s: Apple (currently the most valuable company in the world), Oracle, and Microsoft; and Intel was founded in the 1960s.

What do super-unicorns have in common? The 1960s marked the era of the semiconductor; the 1970s, the birth of the personal computer; the 1980s, a new networked world; the 1990s, the dawn of the modern Internet; and in the 2000s, new social networks were built.

Each major wave of technology innovation has given rise to one or more super-unicorns  — companies that could change your life to work at or invest in, if you’re not lucky/genius enough to be a co-founder. This leads to more questions. What is the fundamental technology change of the next decade (mobile?); and will a new super-unicorn or two be born as a result?

Only four unicorns are born per year on average. But not all years have been as fertile:

The 38 companies on our list outside of Facebook are worth about $3.6 billion on average. This might feel like a letdown after reading about super-unicorns, but remember, startups generally start as ideas that most people think are crazy, dumb, or not that important (remember when people ridiculed Twitter as the place to share that you were eating a ham sandwich?). Only after many years and extraordinary good fortune, a few grow into unicorns, which is extremely rare and pretty awesome.

Unicorn founding was not front-end-loaded in the past decade. The best year was 2007 (8 of 36); the fewest were born in 2003, 2005 and 2008 (as far as we know today; there are none yet founded in 2011 to today). From this snapshot in time, it’s not clear whether the number of unicorns per year is changing over time.

It would be interesting to plot the trajectory of unicorns over time  — which become more valuable and which fall off the list — and to understand the list of potential unicorns-in-waiting, currently valued at <$1 billion. Hopefully for a future post.