Why startups fail is rarely explained by the famous 90% claim, and the real causes are easier to spot than most founders think. This guide from The Business Perspective shows why startups fail using official survival data and post-mortem studies.
Key Takeaways
Everyone has heard the number. Nine in ten startups fail. It shows up in pitch decks, podcasts and family dinners, and almost nobody can say where it came from. Understanding why startups fail starts with checking that number before you believe it.
The Business Perspective went to the primary data, including U.S. Bureau of Labor Statistics survival tables and post-mortem studies, to see what holds up. The short version is that the 90% claim is narrower than it sounds, and the real causes are more useful than the scary statistic. Every figure below names its source.
We also built a free 20-page companion with worksheets for your own company. You will find the download in the box below.
Free download: The Startup Failure Field Guide
Twenty pages from The Business Perspective with a seven-cause risk scorecard, runway and unit-economics worksheets, a pre-mortem and a 30-day rescue plan.
Only under a narrow definition. The 90% figure is tied to venture-backed startups over roughly ten years, and one review says its origin is unclear. For all new businesses, official U.S. data shows about half still open after five years. What counts as failure changes the answer.
Here is the awkward part. The closest source is Startup Genome, which one analysis summarises as finding that only about one in twelve entrepreneurs succeed, per Nanoglobals. The same review says the exact origin of the 90% figure is not clear, and that the year of the report is itself inconsistent across citations.
A second number competes with it. Harvard Business School’s Shikhar Ghosh found that about 75% of venture-backed startups never returned cash to investors, per a Wall Street Journal report cited by Failory. His dataset covers roughly 2,000 companies.
| Claim | Source | What it measures |
|---|---|---|
| 9 in 10 startups fail | Startup Genome (origin unclear, per Nanoglobals) | Startups over about ten years, against an investor-style bar for success |
| 75% of VC-backed startups fail | Shikhar Ghosh, Harvard Business School, via WSJ and Failory | Venture-backed companies that never returned cash to investors |
| About half of businesses close within five years | U.S. Bureau of Labor Statistics | All new private-sector business locations, not just startups |
Notice that the three do not measure the same thing. One judges startups against a venture bar, one counts cash returned and one simply counts who is still open. That is why a single headline number misleads. Our report on why startups fail and how businesses fail worldwide covers the wider picture.
About half of new business locations are still open after five years, and roughly a third after ten. U.S. Bureau of Labor Statistics data for the 2015 opening cohort shows 79.6% survived one year, 69.1% two years, 50.2% five years and 34.7% ten years. The curve has hardly changed across cohorts.
| Opening cohort | 1 year | 2 years | 5 years | 10 years |
|---|---|---|---|---|
| 2005 | 80.1% | 68.7% | 46.8% | 33.8% |
| 2010 | 78.6% | 68.6% | 51.1% | 35.3% |
| 2015 | 79.6% | 69.1% | 50.2% | 34.7% |
| 2020 | 80.9% | 72.3% | 51.4% | not yet available |
These figures come from the BLS Business Employment Dynamics survival table, where cohorts are years ended March. Do the arithmetic on the 2015 row and about two-thirds of locations were closed within ten years, our calculation (100% minus 34.7%). The pattern is steady across a decade of cohorts.
The 2022 cohort is worth a glance. Its one-year survival rate, 76.3%, was the lowest since the 2008 cohort at 75.2%, per BLS. That is a small dip, not a collapse, but it shows the data moves with the economy.
One warning before you quote any of this. BLS counts establishments, meaning individual business locations. A closed shop is not always a failed company, and a barber shop is not a venture-backed startup. Treat the table as a base rate for new businesses, not for tech startups.
It depends on who is counting. Governments usually count closures. Venture investors count whether they got their money back, or whether the company hit a large multiple such as 10X. Many measures treat anything short of closure as survival, including zombie companies and small acquisitions, per Nanoglobals.
| Definition | What it means | Who uses it |
|---|---|---|
| Closed | The business stopped operating or became insolvent | Government data such as BLS |
| Never returned cash | Investors did not get their money back | Venture studies such as Ghosh’s |
| Missed the venture bar | No large multiple, such as 10X | Venture investors |
| Zombie | Still alive, not growing, unable to raise | Founders and advisers |
| Soft landing | Acquired in a small deal, often for the team | Mixed; often counted as survival |
Nanoglobals argues that the 90% number is roughly accurate only when you use the harshest definition, a venture-backed startup measured over ten years against a high-multiple bar. Loosen the definition and the figure falls fast. A founder building a profitable business that never raises venture money can be a success on every measure that matters to them.
So before you worry about the number, define your own. Our field guide has a short exercise for exactly this, and it takes five minutes. For founders weighing investors, our startup term sheet guide explains the terms that shape what an investor counts as a win.
Two public datasets agree that the market comes first. CB Insights found “no market need” was the top reason in 42% of 101 post-mortems in 2018. Failory’s interviews with founders of more than 80 failed startups put marketing problems, mostly missing product-market fit, at 56%. Money and people follow.
| Source | Finding | Sample and caveat |
|---|---|---|
| CB Insights (2018) | No market need was the top reason in 42% of cases | 101 post-mortems; percentages for other reasons not available to us |
| Failory founder interviews | Marketing 56%, team 18%, finance 16%, tech 6%, operations 2%, legal 2% | 80+ failed startups; self-selected, and Failory interprets the answers |
Both samples are small, and both rely on founders explaining their own failures. That is useful, but it is not a census. Failed companies that never wrote a post-mortem are missing, which is our caveat rather than the authors’.
Even so, on why startups fail, the direction is hard to miss. Building something people will not pay for outranks bad code, bad law and bad operations. Our startup funding coverage, such as how to raise seed funding for startups, keeps returning to the same point: investors fund evidence of demand, not enthusiasm.
Seven causes keep appearing: no market need, running out of cash, the wrong team, being outcompeted, broken unit economics, weak retention and lost focus. This grouping is The Business Perspective’s own, built from CB Insights and Failory data. Most startups face several at once, and one cause usually triggers the next.
This is the big one. CB Insights found it was the top reason in 42% of 101 post-mortems, and Failory’s interviews put marketing, mostly missing product-market fit, at 56%. The warning sign is simple: people say “nice idea” but never pay.
The fastest test is to ask for money, or a signed letter of intent, before you build the next feature. Praise is cheap. A purchase order is a signal.
Finance was the cause in 16% of Failory’s founder interviews. Running out of money is often the last blow even when something else started the decline. The formula is plain: runway equals cash in the bank divided by net monthly burn.
Our rule of thumb, not a benchmark, is that fundraising takes longer than founders expect. Know your runway in months and start early. If you are weighing a first round, see our guide to pre-seed versus seed rounds.
Team problems were behind 18% of cases in Failory’s interviews. They rarely look like bad people. They look like missing skills, unclear roles and decisions nobody owns.
A quick check helps. Ask each founder, separately, to write down who owns product, revenue, hiring and money. If the answers differ, you have found the problem, and a written agreement on equity and vesting is the next step. For what employees will ask you, see how candidates negotiate startup equity.
Being outcompeted appears among CB Insights’ leading failure reasons, though its percentage was not available to us. The danger is not always a bigger rival. It is a rival that is faster, cheaper or simply easier to buy.
Run a win/loss log for 30 days and record the real reason from each lost customer, not the one in your CRM notes. That single habit turns vague worry about competitors into specific fixes.
Pricing and costing issues also feature in CB Insights’ list. The pattern is simple. Each new customer costs more to win and serve than they ever pay, so growth makes the hole deeper.
Three numbers tell you where you stand: customer acquisition cost, payback in months and customer lifetime. Our field guide includes a worksheet with a worked example, and the arithmetic is shown so you can check it yourself.
Technology problems were the cause in 6% of Failory’s interviews, but the deeper issue is whether users stay. If each new signup cohort drops to near zero, growth is hiding a leak and better acquisition will not fix it.
Plot retention by signup month. It is the quickest way to find out whether the product, or the timing, is the problem.
This cause is our own addition. It is not a measured category in the datasets above, so read it as editorial judgment. A company that chases every request ends up serving no customer well, and the roadmap changes monthly while nothing gets finished.
Ask five team members for the top priority this quarter. Five answers means the problem is focus, not effort. Strong boards force this clarity, which is why our startup board meeting agenda puts one metric first.
Score each of the seven causes from 0 to 3 honestly, then add the total. Know your runway in months, your payback period per customer and whether paying customers stay. Our scorecard is a planning tool, not validated research, but it forces you to see all seven risks together on one page.
| Cause | Early warning sign | First fix |
|---|---|---|
| No market need | Praise but no payment | Ten customer conversations and one paid pilot |
| Out of cash | Runway unknown or under about nine months | Calculate burn, cut three costs, start the raise early |
| Wrong team | Two people both think they lead | Write who owns what, with decision rights |
| Outcompeted | Lost deals with no clear reason | 30-day win/loss log |
| Broken unit economics | Payback longer than customers stay | Fix price, cost or churn before buying growth |
| Weak retention | New cohorts drop off quickly | Plot retention by signup month |
| Lost focus | Five people, five priorities | One metric, one customer, one goal |
The Business Perspective’s free field guide turns this table into a fillable scorecard for diagnosing why startups fail, with a runway calculator, a unit-economics worksheet, a pre-mortem template and a 30-day rescue plan. It is twenty pages and works for solo founders and teams alike.
Free download: The Startup Failure Field Guide
Twenty pages from The Business Perspective with a seven-cause risk scorecard, runway and unit-economics worksheets, a pre-mortem and a 30-day rescue plan.
Critics make three points. The 90% figure may be fair for venture-backed startups judged on a high bar, and government data counts locations, not startups. Post-mortem studies are also small and self-selected. We agree with all three, which is why every number here carries its source and its limits.
Here is what sceptics say, and they have a point.
Our view at The Business Perspective is that none of this makes the headline number useful. It makes the definition useful. Decide what failure means for your company, then measure against that. Our AI bubble analysis shows how fast investor definitions of success can shift when capital tightens.
Readers most often ask whether the 90% figure is true, what the top cause is, how many businesses survive five years, what share of venture-backed startups fail, how long failure takes and how to spot the risk. The Business Perspective’s six answers below are short and sourced.
Only under a narrow definition. The 90% figure is tied to venture-backed startups over about ten years, and one review says its exact origin is unclear. Official U.S. data shows roughly half of all new business locations are still open after five years, so the answer depends on what you count as failure.
Lack of market need tops the lists. CB Insights found it was the leading reason in 42% of 101 startup post-mortems it analysed in 2018. Failory’s founder interviews point the same way, with marketing problems, mostly missing product-market fit, cited in 56% of cases. Both samples are small, so treat them as patterns.
About half. U.S. Bureau of Labor Statistics data shows 50.2% of private-sector establishments opened in the year to March 2015 were still open after five years, and 34.7% after ten. These figures cover all new business locations, not only startups, so tech companies may behave differently.
A widely cited figure is 75%. Harvard Business School’s Shikhar Ghosh, as reported by the Wall Street Journal and cited by Failory, found about three in four venture-backed companies never returned cash to investors. That is a narrower definition of failure than closing down, based on roughly 2,000 companies.
Closures cluster early but continue for years. BLS data for the 2015 cohort shows about 20% of new establishments were gone within one year, about 31% within two and about 50% within five. Failure can also be slow, with zombie companies surviving for years without growth, so timing varies widely.
Score the seven causes honestly: market need, cash runway, team, competition, unit economics, retention and focus. Know your runway in months, your payback period per customer and whether paying customers stay. Our free field guide includes a one-page risk scorecard, a runway calculator and a 30-day plan to act on the results.
Stop quoting the 90% number and start checking your own risks. Define what failure means for your company, know your runway and unit economics, talk to paying customers and score the seven causes honestly. That is how you act on why startups fail instead of just worrying about it.
The honest takeaway is that the scary number tells you little, but the causes tell you a lot. Most failures start with a market that does not care, and most are made worse by cash, team and focus problems that were visible months earlier. That is good news, because visible problems can be fixed.
Take the field guide into your next team meeting, fill in the scorecard together and pick one cause to fix this month. The Business Perspective will keep publishing practical, sourced guides like this one for founders and investors. If you are on the other side of the table, our ten biggest AI agent rounds of October 2026 show where capital is flowing now.
Free download: The Startup Failure Field Guide
Twenty pages from The Business Perspective with a seven-cause risk scorecard, runway and unit-economics worksheets, a pre-mortem and a 30-day rescue plan.
Sources: U.S. Bureau of Labor Statistics (Business Employment Dynamics, Table 7), CB Insights (analysis of 101 startup post-mortems, 2018, via StartupNews), Failory (founder interviews and summary of Shikhar Ghosh’s Harvard Business School research reported in the WSJ), Nanoglobals (analysis of the 90% claim). The seven-cause grouping is The Business Perspective’s own framework. This article is education, not legal, tax or financial advice. Published by The Business Perspective Editorial Team to explain why startups fail.
