What Percent of Startups Fail? What the Data Actually Says
About 20% of new US businesses close in year one and two-thirds within 10 years (BLS). Here's where the 90% figure really comes from.

About 20% of new US businesses close within their first year, roughly half are gone by year five, and about two-thirds by year ten, according to the US Bureau of Labor Statistics. That’s the most solid answer to what percent of startups fail.
The popular “90% of startups fail” line is a different claim, and it doesn’t trace back to any official dataset. One set of numbers counts every new business that opens its doors. The other usually refers to venture-backed companies, where “failure” can mean not returning investors’ money. Here’s how to read each figure, and what it means for your own odds.
What percent of startups fail, year by year
The best long-run source in the US is the BLS Business Employment Dynamics program, which tracks every private-sector business establishment from the year it first reports employees. In a January 2024 release, BLS followed the group born in March 2013 through March 2023.
Survival of US private-sector establishments born in March 2013. The biggest single drop happens in year one.
Read as closure rates, that’s roughly 20% gone after one year, 31% after two, 49% after five and 65% after ten. After the first year, the losses keep coming, but at a slower pace each year.
One caveat: BLS counts establishments, meaning business locations, and an establishment that stops reporting employees isn’t always a failure. Some are sold, merged or closed by owners who simply retire. So these figures measure closures, which is a slightly wider net than “failed.”
Where “90% of startups fail” comes from
Search for startup failure rates and you’ll see 90% everywhere, usually credited to a statistics roundup that credits another roundup. I couldn’t find a government dataset or peer-reviewed study behind it.
The closest serious number covers venture-backed companies. In September 2012, the Wall Street Journal reported research by Shikhar Ghosh, a senior lecturer at Harvard Business School, who studied more than 2,000 companies that raised at least $1 million in venture funding between 2004 and 2010. About three-quarters didn’t return investors’ capital.
The National Venture Capital Association’s estimate at the time was far lower, at 25% to 30%, because it counted companies that failed outright. Same companies, different definition of failure.
Why the startup failure numbers disagree
Almost every conflicting startup statistic comes down to two questions: who’s being counted, and what counts as failing. Here are the figures you’ll see most often, matched to their sources.
| Figure you’ll see | Source and date | Who’s counted | What “fail” means |
|---|---|---|---|
| About 20% close in year one | BLS, Jan 2024 (2013 cohort) | All new US private-sector establishments | Stopped operating |
| About half close by year five | BLS, Jan 2024 | Same | Stopped operating |
| About 65% close by year ten | BLS, Jan 2024 | Same | Stopped operating |
| 75% of startups fail | Shikhar Ghosh, via WSJ, Sept 2012 | 2,000+ US companies that raised $1M+ in venture funding, 2004 to 2010 | Didn’t return investors’ capital |
| 25% to 30% fail | NVCA estimate, via WSJ, Sept 2012 | Venture-backed companies | Failed outright |
| 90% fail | No primary source found | Unclear | Unclear |
If someone quotes a startup failure rate without saying who was counted, ask. The answer usually explains the number.
Do tech startups fail more often?
By the BLS numbers, yes. Among businesses born in 2013, the information sector, which includes software publishers, telecom and data hosting, had one of the lowest 10-year survival rates at 29.1%. Professional, scientific and technical services, home to many tech consultancies and agencies, came in at 30.9%. Only mining did worse.
| Industry (2013 cohort) | Open after 1 year | After 5 years | After 10 years |
|---|---|---|---|
| Information | 74.9% | 44.3% | 29.1% |
| Professional, scientific and technical services | 77.0% | 46.3% | 30.9% |
| Accommodation and food services | 82.5% | 55.3% | 38.2% |
| Construction | 76.0% | 53.9% | 40.1% |
| Retail trade | 84.2% | 58.3% | 41.7% |
| All private sector | 79.6% | 50.6% | 34.7% |
Source: U.S. Bureau of Labor Statistics, establishments born March 2013, published January 12, 2024.
Restaurants have a reputation as the riskiest small business, yet in this cohort they outlasted the information sector at every checkpoint. My read: tech is cheap to start, quick to test and quick to wind down when the test fails, so more companies exit early.
Venture-backed startups fail later, and differently
Ghosh’s research also found a timing difference. Startups without venture money tended to fail more often in their first four years, usually because they lacked the cash to keep going while they fixed a broken model. Venture-backed companies more often failed after year four, once investors stopped putting money in.
Recent shutdown data shows the squeeze is real. Carta recorded 966 shutdowns among startups on its platform in 2024, up from 769 in 2023, TechCrunch reported in January 2025, and Carta’s own team said that likely undercounts the real total. SimpleClosure, which handles startup wind-downs, said in December 2025 that Series A companies rose from about 6% to 14% of the closures it handled that year.
What these numbers mean for your odds
None of these figures predicts your company. They’re averages across hundreds of thousands of businesses with very different founders, budgets and markets. What they do show is where the risk concentrates.
Plan around the first year. If roughly one in five new businesses doesn’t reach month 12, your first budget should cover at least that far without assuming new funding or revenue shows up on schedule. Track runway every month, not every quarter. Test whether customers will pay before you build the full product. And set a date, in advance, for deciding whether to keep going, change direction or stop, so the decision isn’t made for you by an empty bank account.
FAQ
What percentage of startups succeed?
It depends on what success means. If it means survival, BLS data shows 34.7% of US businesses born in 2013 were still operating ten years later. If it means returning investors’ money, Shikhar Ghosh’s research on venture-backed firms, reported by the WSJ in 2012, suggests about one in four.
Do most startups fail in the first year?
No, most close later, but year one has the single biggest drop. In the BLS 2013 cohort, 20.4% of new establishments closed within 12 months, while about 65% had closed by year ten. So roughly a third of all closures over the decade happened in the first year.



Discussion