Key takeaways
- 20.4% of all U.S. businesses fail in year one, 49.4% by year five: These are BLS 2024 figures for all private-sector establishments, not the "90% of startups fail" headline that applies only to venture-backed companies.
- 75% of venture-backed startups never return cash to investors: Harvard Business School research (Shikhar Ghosh, 2,000 companies) found the real VC failure rate far exceeds the 20-30% that venture firms self-report.
- The #1 root cause is poor product-market fit at 43%: CB Insights analyzed 431 failed VC-backed companies in 2024 and found "ran out of capital" (70%) is the symptom, not the cause.
- Startup shutdowns rose 25.6% in 2024 to 966 recorded closures: Carta data shows the 2021 funding boom produced a wave of failures peaking in 2024-2025, with 74% of shutdowns at pre-seed or seed stage.
Nine out of ten. That is the startup failure number you will find in every pitch deck disclaimer, every accelerator slide, every founder's Twitter bio. It is also wrong, or at least incomplete. The real startup failure statistics for 2026 depend on what you count as a startup, what you count as failure, and where you pull your data.
I built Preuve AI to help founders validate before they build, so I spend a lot of time tracing these numbers to their primary sources. Most stat roundups recycle the same handful of figures without checking where they came from or whether they still hold. This page is my attempt to fix that. Every number below links to the original study, government dataset, or research report it came from.
If you want the narrative on why startups fail (the CB Insights deep-dive, the PMF argument), I wrote that separately in why startups really fail in 2026. This page is the reference table. Bookmark it or argue with it.
Will your idea survive the market?
Preuve AI runs 10 agents against live market data and links every claim to a source. Free analysis in 60 seconds.
What percentage of startups fail?
The answer depends entirely on which population you are counting. The two numbers that get mixed up constantly are the BLS "all businesses" rate and the VC-backed startup rate. They measure different things.
| Metric | Rate | Source |
|---|---|---|
| All U.S. businesses, year 1 | 20.4% | BLS Business Employment Dynamics, 2024 |
| All U.S. businesses, year 5 | 49.4% | BLS Business Employment Dynamics, 2024 |
| All U.S. businesses, year 10 | 65.3% | BLS Business Employment Dynamics, 2024 |
| VC-backed, never return investor capital | 75% | Harvard Business School (Ghosh), 2,000 companies |
| VC-backed, liquidate all assets (total loss) | 30-40% | Harvard Business School (Ghosh), 2,000 companies |
| VC-backed, fail to achieve venture returns | ~90% | Startup Genome Project |
| Startups that achieve unicorn status | ~1% | CB Insights |
The gap between 20.4% (year-one BLS) and 90% (Startup Genome) is not a contradiction. The BLS counts every private-sector establishment: restaurants, dentist offices, landscaping companies. The 90% figure applies to innovative, scalable startups attempting venture-scale growth. Mixing these populations is how most stat pages get the number wrong. You need to know which one applies to you.
What are the top reasons startups fail?
The most widely cited source here is CB Insights' 2024 study of 431 failed VC-backed companies. It replaced their earlier analysis of 110+ post-mortems (the source of the famous "42% no market need" stat from 2014). The updated study has 4x the sample size and distinguishes root causes from symptoms.
| Failure Cause | % of Failures | Classification |
|---|---|---|
| Ran out of capital | 70% | Symptom |
| Poor product-market fit | 43% | Root cause |
| Bad timing / macro conditions | 29% | Root cause |
| Unsustainable unit economics | 19% | Root cause |
Percentages exceed 100% because companies cite multiple causes. The part worth paying attention to: "ran out of capital" is almost always the final symptom, not the root problem. The 431 companies in the study raised a combined $17.5 billion in equity before dying. The median company raised $11 million. These teams had money. What they lacked was evidence that anyone wanted what they were building.
The median time from last fundraise to shutdown was 22 months, and nearly 25% were "walking dead" for 3+ years before officially closing (CB Insights). That pattern matters: most founders knew something was wrong long before they admitted it. I wrote a deeper look at how this connects to validation in why startups really fail.
The Failory cross-check: Failory's independent analysis of 80+ failed startups found marketing problems (56%) topped their list, followed by team (18%) and finance (16%). Different taxonomy, consistent signal: the problem is demand-side, not supply-side.
What is the startup failure rate by industry?
The Bureau of Labor Statistics tracks survival rates for every private-sector establishment by NAICS industry code. This is the most reliable industry-level dataset because it covers all U.S. businesses, not just startups that self-report or take VC money. The table below uses 2024 BLS data, analyzed by Commerce Institute.
| Industry | 1-Year | 5-Year | 10-Year |
|---|---|---|---|
| Agriculture, forestry, fishing, hunting | 12.5% | 33.8% | 49.5% |
| Real estate and rental/leasing | 16.1% | 41.3% | 57.8% |
| Retail trade | 15.8% | 41.7% | 58.3% |
| Manufacturing | 17.4% | 42.3% | 56.4% |
| Construction | 24.0% | 46.1% | 59.9% |
| Finance and insurance | 19.1% | 46.8% | 62.5% |
| Health care and social assistance | 17.3% | 44.9% | 64.3% |
| Transportation and warehousing | 20.6% | 49.9% | 66.0% |
| Professional, scientific, technical | 23.0% | 53.7% | 69.1% |
| Information (incl. tech/software) | 25.1% | 55.7% | 70.9% |
| Mining, quarrying, oil and gas | 20.6% | 59.8% | 75.5% |
| All industries average | 20.4% | 49.4% | 65.3% |
Two things jump out. First, the Information sector (which includes software and tech startups) has the highest 10-year failure rate at 70.9%. If you are building a SaaS product, your baseline odds are worse than average. Second, the spread is enormous: agriculture's 10-year rate (49.5%) is 26 points lower than information (70.9%). Industry selection is a risk factor most founders ignore.
Note: the BLS does not have a "startup" category. These numbers cover every new establishment in each sector, from a solo consultant registering an LLC to a venture-backed SaaS company. For tech-specific startup data, see the startup validation benchmarks I published with data from Preuve AI scans.
How does the startup failure rate change by funding stage?
Failure rates drop sharply as startups progress through funding rounds. The data here comes from multiple sources because no single dataset covers the full pipeline with consistent methodology.
| Funding Stage | Failure Rate | Source |
|---|---|---|
| Pre-seed / Seed (fail to reach Series A) | 60-70% | SPDLoad / industry aggregate |
| Series A (fail to reach Series B) | ~35% | SPDLoad / industry aggregate |
| Series B and beyond | ~1% | SPDLoad / industry aggregate |
The trend is steep: the earlier you are, the more likely you are to fail. Seed-stage companies face 60-70% attrition before they even reach Series A. Once a startup closes a Series B, it has usually demonstrated enough traction that outright failure drops to roughly 1%.
This maps directly to Carta's 2024 shutdown data: 74% of recorded startup shutdowns occurred at pre-seed or seed stage, with 41% specifically at seed. The highest-risk period is exactly when most founders are skipping validation because they "just want to build."
Skip weeks of manual research
Get complete market research, sourced proof, competitor map, and pricing data for your idea instantly.
How many startups shut down in 2024 and 2025?
The 2021-2022 funding boom is now producing its downstream wave of failures. Two independent tracking platforms confirm the spike.
The different totals reflect different customer bases. Carta tracks its own platform customers. AngelList tracks winddowns on its platform. Both show the same direction: shutdowns are accelerating.
The sector breakdown from Carta is worth noting: enterprise SaaS led at 32% of shutdowns, followed by consumer (11%), health tech (9%), fintech (8%), and biotech (7%) (Carta via TechCrunch). CB Insights' parallel data shows healthcare and biotech destroyed the most capital: 62 failed healthcare/biotech companies burned through $5.1 billion.
The root cause is not a mystery. TechCrunch's analysis frames it plainly: VCs did not get better at picking winners in 2021. They just funded more companies. The hit rate stayed flat (or got worse), so the absolute number of failures had to rise. The 2021 vintage is now 3-4 years old, which is exactly when the median failed company runs out of runway.
What does "startup failure" actually mean?
This is the question none of the stat pages answer, and it is the one that matters most for how you read every number above.
There is no single definition. The major sources each define it differently:
BLS: cessation of operations.
An establishment that no longer has employees. This includes everything from a planned retirement to a bankruptcy. It does not distinguish between "I chose to close" and "I was forced to close."
Harvard (Ghosh): failure to return investor capital.
A company that raised VC money and never paid it back. This counts profitable small businesses that returned 0.8x as "failures" alongside total wipeouts.
CB Insights: company shutdown.
The company ceased operations entirely. Acqui-hires and pivots are not counted as failures, even if investors lost money.
Startup Genome: failure to achieve venture returns.
A startup that does not generate the 10x+ returns VCs target. By this definition, a profitable $5M/year company is a "failure" if it raised a $20M Series A.
I think about this constantly because I built a validation tool that helps founders gauge risk before building. The definition changes the math. If "failure" means your company literally shuts down, the rate is closer to the BLS numbers (20% year one, 50% year five). If it means "you did not hit venture-scale returns," the rate is 90%+.
For a solo founder or small team deciding whether to build, the BLS definition matters more than the Startup Genome definition. You probably are not optimizing for venture returns. You are trying to not go broke. The relevant question is not "will I become a unicorn" but "will anyone pay for this." That second question is answerable before you write code.
How do first-time founders compare to repeat founders?
Experience matters, but less than you might hope.
| Founder Type | Success Rate | Source |
|---|---|---|
| First-time founders | 18% | Failory / BLS aggregate |
| Repeat founders (prior success) | 30% | Failory / BLS aggregate |
| Repeat founders (prior failure) | 20% | Failory / BLS aggregate |
The 12-point gap between first-time (18%) and previously-successful repeat founders (30%) is real but modest. A prior failure barely moves the needle (20% vs 18%). What repeat founders learn is not magic. It is usually validation discipline: they stop assuming demand and start testing it. If you are a first-time founder, you can borrow that discipline without the expensive lesson. That is the whole premise behind startup validation benchmarks.
How to read these statistics without getting paralyzed
I have spent the last section of this page dropping sobering numbers on you, so here is the counterweight: every one of these statistics describes what happens on average, across all founders, including those who did zero validation.
The CB Insights data makes this point clearly. The #1 root cause (poor product-market fit, 43%) is the one cause you can address before you spend money. Running out of cash (70%) is the symptom of ignoring that cause. Bad timing (29%) is harder to control, but you can at least check whether the market is ready right now. I wrote about how to do that in dead startup ideas worth reviving, where every case study was a company that failed because the timing was wrong, not because the idea was wrong.
The Startup Genome Project found that 74% of failed startups scaled prematurely, building teams and spending on marketing before confirming product-market fit. Startups that pivoted 1-2 times showed 3.6x better user growth than those that pivoted zero or more than 2 times. The pattern is consistent across every dataset I have reviewed: validation before building is the single highest-leverage activity a founder can do.
That is not a sales pitch. You can validate manually for $0. Talk to potential customers, check search volume, scan competitor revenue. Or you can run your idea through a free scan and get the same signal in minutes. Either way, the point is the same: know whether anyone wants what you are building before you build it.
FAQ
What percentage of startups fail?
It depends on how you define "startup." For all new U.S. private-sector businesses, the Bureau of Labor Statistics reports a 20.4% failure rate in year one, 49.4% by year five, and 65.3% by year ten (2024 data). For venture-backed startups specifically, Harvard Business School researcher Shikhar Ghosh found that 75% never return cash to investors, based on a study of 2,000 companies.
What is the number one reason startups fail?
Poor product-market fit. CB Insights analyzed 431 failed VC-backed companies in 2024 and found 43% cited poor product-market fit as a root cause of failure. Running out of capital affected 70% of failures, but CB Insights identifies that as the final symptom rather than the underlying cause.
What industry has the highest startup failure rate?
According to BLS 2024 data, the Information sector (which includes tech and software) has the highest 10-year failure rate at 70.9%, followed by mining/oil and gas at 75.5%. Agriculture, forestry, fishing, and hunting has the lowest 10-year failure rate at 49.5%.
Do most startups fail in the first year?
No. Most startups that fail do so between years two and five. The BLS first-year failure rate for all U.S. businesses is 20.4%. The danger zone is the period after initial launch when founders discover whether real demand exists, which is why 70% of startup failures occur between years two and five.
How many startups shut down in 2024?
Carta recorded 966 startup shutdowns in 2024, a 25.6% increase from 769 in 2023. AngelList tracked 364 winddowns in the same period, a 56.2% increase. The discrepancy reflects different tracking methodologies and customer bases.
Vincent
5 years in B2B growth, building Preuve AI in public. 82% of ideas it scores aren't ready, the point is finding out in 5 minutes, not 3 months.
Follow on X →
Building is expensive. Validation is free.
Run your idea through 10 AI agents before you write a line of code. Every claim source-linked.
