Why do most startups fail?

The short answer

The “90% of startups fail in year one” statistic is one of the most repeated and least true claims in business media. The actual BLS data shows roughly 20% of new U.S. businesses close within their first year — high, but nowhere near 90%.

What is true is that failure compounds over time. By year five, half are gone. By year ten, two-thirds. The early survival is misleadingly high because companies that have not yet hit their cash crunch or competitive reckoning are still technically operating.

The reasons companies fail are not random. CB Insights’ analysis of more than 100 startup post-mortems found that “no market need” was cited in 42% of cases — far more than running out of cash (29%) or wrong team (23%). Cash always runs out, but the underlying cause is usually that the product never found a buyer in volume.

→ New to investing fundamentals? Financial education across macro regimes

What the data shows

The BLS Business Employment Dynamics survival data (2024 release):

• 1-year survival rate for U.S. businesses: approximately 79.6% (about 20% fail in year one)
• 5-year survival rate: approximately 50.6% (about 50% fail by year five)
• 10-year survival rate: approximately 34.7% (about 65% fail by year ten)
• Industry variation is large: agriculture has approximately 87.5% one-year survival vs information sector at approximately 74.9%
• VC-backed startups face a different set of failure rates: roughly 60% of pre-seed-funded startups never reach Series A; the failure rate drops sharply once a company reaches Series C and beyond

The exception worth noting: the 90% failure number is not entirely fabricated — it roughly approximates the failure rate of crypto and blockchain startups (~95%) and venture-backed tech startups generally (~63% within five years). The mistake is conflating these high-risk subsegments with all small businesses.

→ Dataset: U.S. Bank Lending Standards (Fed SLOOS)

Why it happens — the macro mechanism

Startup failure follows predictable patterns shaped by demand, capital, and competition.

Channel 1 — Demand-side mismatch. The 42% “no market need” finding from CB Insights captures the dominant failure mode: founders build products that solve problems customers do not actually have, or solve them in ways customers will not pay for. This is fundamentally a discovery problem, not a financial one. The companies that fail this test are often well-funded — they simply spend that funding on the wrong product.

Channel 2 — The myth-versus-data gap — the angle worth highlighting. The persistent “90% fail in year one” claim survives despite empirical refutation because it serves narrative functions for both pessimists (warning against entrepreneurship) and optimists (celebrating survival). The actual data tells a different story: the first year is the easiest because runway from initial capital, founder energy, and customer goodwill have not been exhausted. Years 3-7 are the most lethal — when initial capital is exhausted but scale economies have not yet materialized.

Channel 3 — Macro-cyclical exposure. Startup failure rates are deeply procyclical. In capital-abundant regimes (2010-2021), even mediocre startups can keep raising rounds and stay nominally alive; in capital-constrained regimes (2022-2024), the same companies face down rounds, layoffs, and shutdowns within months. The shift is mechanical: when LP capital tightens, VCs reserve dry powder for existing winners and stop funding marginal portfolio companies. Understanding VC mechanics illuminates why this acceleration is built into the system.

Synthesis by regime: in the 2009-2021 capital-abundant period, the average venture-backed startup survived for unusually long periods because successive rounds extended runway regardless of fundamentals; in the 2022-2024 normalization, failure rates among venture-backed firms accelerated sharply, with PitchBook data showing dead-company counts in 2023-2024 reaching multi-year highs even though the underlying failure causes (no market need, weak team) had not changed.

Startups do not die from running out of cash — they die because they ran out of cash without first finding a customer who would pay enough to keep the lights on.

→ Conceptual framework: Equity markets and the economic cycle

What it means for different economic actors

Founders need to understand that the failure modes are well documented and largely avoidable in advance. Customer development before product development is the most-cited research-supported approach for reducing the “no market need” failure mode.

VC investors face an inversion of conventional risk thinking: they need failure rates to be high in their portfolio because the few survivors are what generates returns. A VC fund with 60% failure rate is normal; one with 20% failure rate suggests insufficient risk-taking and is unlikely to generate top-quartile returns.

Bank lenders see failure differently — for banks, even one default can wipe out the spread earned on dozens of performing loans, which is why bank credit to startups is rare and heavily collateralized. The Altman Z-score remains widely used by lenders to flag distress before it becomes default.

A common error is to interpret survival as success. A business that has been operating for ten years but generates negative free cash flow is “alive” in the BLS data, but is not creating value.

Practical observation

Go deeper

Related questions

• How do small businesses finance growth?
• How does venture capital actually work mechanically?
• Why is cash flow more important than profits?
• What is working capital management?

Frequently asked questions

How does the failure rate vary by industry?

BLS data shows substantial industry variation. Agriculture and forestry have the lowest failure rates — approximately 12.5% in year one and 49.5% by year ten. The information sector has among the highest — approximately 25.1% in year one and 70% by year ten, reflecting both higher risk-taking and faster competitive obsolescence. Mining and oil and gas extraction have the highest failure rates of all (approximately 30.8% in year one), driven by commodity-cycle exposure. The variance across industries is more than 2× — meaning a “startup failure rate” without an industry qualifier is almost meaningless.

Why do venture-backed startups fail at higher rates than typical small businesses?

VC-backed startups deliberately pursue high-risk, high-growth business models that have a much wider distribution of outcomes. The CB Insights and PitchBook data both suggest VC-backed tech startups fail at roughly 63% within five years, versus 50% for the BLS small-business universe. This is not a failure of VC selection — it is the design of the model. VCs need a small number of extreme winners, which requires a portfolio of high-risk bets where most will fail.

Are serial entrepreneurs really more likely to succeed than first-timers?

Yes, but the magnitude is often overstated. Empirical research from Kauffman Foundation and academic studies (Gompers, Kovner, Lerner, Scharfstein) suggests serial entrepreneurs with a prior success have approximately 30% probability of success in their next venture, versus approximately 18% for first-time founders and approximately 20% for serial founders with prior failures. The pattern is robust but the absolute success probabilities remain modest — even prior winners fail 70% of the time on their next attempt.

Last updated — 12 July 2026