The Startup Failure Myth

An often cited statistic (sometimes quoted from the Small Business Administration) is 90% of businesses fail within the first 12-18 months. This stat is used to illustrate how difficult it is to start a company. I heard this many times while at graduate school. It was very discouraging.

We love to praise successful entrepreneurs any chance we get, but we also love to make it sound like starting a company is so unbelievably difficult you probably shouldn’t even attempt it. This FUD holds back many people from starting something on their own or joining an early stage startup. I’m not saying that starting a company is an easy way to riches and fame, but I also don’t think we should use negative statistics about how much startups fail unless they are completely valid.

Fortunately, the “90% fail” stat is a load of crap. Like many statistics that are thrown around, a single percentage can rarely be applied to a large population. Another great example of this is the divorce rate. We often hear that more than 1 out of every 2 marriages end in divorce. Those are some odds! It really makes you feel good about entering a new marriage. Fortunately, they are the wrong odds for most people. The thing about the divorce rate is that it varies dramatically depending on your demographic:

if you are a reasonably well-educated person with a decent income, come from an intact family and are religious, and marry after age twenty-five without having a baby first, your chances of divorce are very low indeed.

The same applies to your odds of starting a successful company. Even if 90% of companies fail in the first 18 months (which hasn’t been substantiated), that percentage would be applied across all companies. The odds of someone with no training or financial backing being successful at starting a company is not the same as someone with a college education and a couple jobs under their belt.

When I was at MIT, I attended a lecture by Ed Roberts where he reviewed some of his findings from studying technology-based companies for 25 years. Here are some relevant points from my notes back in 2005:

Who becomes technology-based entrepreneurs? (”becomes” not necessarily “who are successful”)

  • Disproportionate share came from families where the father was self-employed
  • Women were extraordinarily exceptional - very few
  • Entrepreneurs have a moderate need for achievement, moderate need for power, but a high need for independence
  • Development-oriented work background (not research)

50% of companies in his study were founded on a part-time basis despite this probably violating employee and IP agreements

SBA says 80-90% of companies fail within 18 months, but their data showed only 16% failed after 5-7 years

Average entrepreneur has less than a high school diploma, but their data showed avg tech entrepreneur has a college diploma

Didn’t find a distinction based on where entrepreneur went to college

83% of entrepreneurs were highly satisfied with their performance 5-7 years post-founding and 17% were not satisfied. The entrepreneurs that were not satisfied did not correlate to those that failed

Successful founders’ characteristics:

  • Made, not born (found no birth characteristics that predict success)
  • High need for achievement and optimally with moderate need for power
  • If you have a high need for power you limit other people from stepping up

Have a hypothesis that the more startups you found, the more successful they become

Age distribution at first founding: gradual shifting to younger and older entrepreneurs (the middle is being vacated)

Median age of founding first company is constantly decreasing: 1950s: 40.5, 1960s: 39, 1970s: 35, 1980s: 32, 1990s: 28

Ed Roberts’ data showed that only 16% of tech-based companies failed after 5-7 years. Also, certain characteristics tend to predict success including larger founding teams and the number of companies started prior.

The moral of the story is that you shouldn’t believe every stat you hear. On aggregate it may be correct (although not in this case), but when applied to your specific situation your odds could be dramatically different — for better or worse.


UDPATE: A gentleman from the SBA wrote me to confirm the SBA has never claimed that 90% of businesses fail in the first year. I asked for some referenceable data and he provided it. Thanks Mike! Here is a report from the SBA written back in 2002 that provides ample data that the success rate is much higher. I’ll jump straight to the punchline:

BITS shows that 66 percent of new employers survive two years or more, 50 percent survive four years or more, and 40 percent survive six years or more

The report also highlights how the SBA is frustrated by the 90% failure myth:

Although the business survival rates presented here simply confirm previous findings, perhaps this kind of independent confirmation is what is needed to dispel the myth that 9 out of 10 businesses close in their first year.

Even if the venture-backed startup failure rate is higher, it isn’t as high as 90%. Next time you hear someone quote the 90% failure myth, correct them!