Education
Startup Survival Rates: Why the Five-Year Line Decides Your Equity Outcome
Only 49.2% of new businesses reach year five. The same five-year line sets your QSBS exclusion, your Section 1244 loss, and what your 409A record is worth.
By 409.AI Team - 2026-08-24
# Startup Survival Rates: Why the Five-Year Line Decides Your Equity Outcome
Every founder has heard some version of "90% of startups fail." Nobody ever cites a source for it, and the number is wrong. The real figures are less dramatic and far more useful, because they tell you *when* the risk sits rather than just how much of it there is.
The SBA Office of Advocacy publishes them, drawing on the Bureau of Labor Statistics' Business Employment Dynamics series. Across 1994 to 2022, an average of 67.7% of new employer establishments survived at least two years. The five-year survival rate was 49.2%. Ten years: 33.9%. Fifteen years: 25.5% ([SBA Office of Advocacy, *Frequently Asked Questions About Small Business*, February 2026](https://advocacy.sba.gov/wp-content/uploads/2026/02/FINAL_FAQsAboutSmallBusiness_2026_012826.pdf)).
So roughly half make it to year five, not 10%. But here is the part worth pinning to the wall: the same data shows that 69.5% of establishments that *reach* five years go on to reach ten, and 76.1% of those that reach ten also reach fifteen. Risk is not spread evenly across a company's life. It is front-loaded into exactly the window where founders are busiest, most cash-constrained, and most inclined to postpone anything that looks like administrative overhead.
That includes almost everything that later determines what your equity is worth after tax.
What the numbers actually measure
Before building anything on top of these figures, it's worth knowing what they count. BED tracks *establishments*, meaning individual work locations with paid employees, across the private sector in every industry. A restaurant, a machine shop and a seed-stage software company all sit in the same pool.
A "closure" in this data means the establishment stopped reporting employment. That is not identical to failure. A location can drop out because it was acquired, consolidated into another site, or relocated. In the other direction, the series excludes businesses with no employees, which is most of the 36.2 million small businesses in the country: 82.3% of them are nonemployer firms.
None of that undermines the shape of the curve, and the shape is what matters. Attrition is heaviest early and thins out sharply for the companies that get past the first few years. Your own sector may run above or below the all-industry line. The timing of the risk is the same either way.
Five years is not an arbitrary milestone
Here's why the five-year line should mean something specific to a founder rather than something vaguely motivational. Section 1202 of the Internal Revenue Code, the qualified small business stock exclusion, is built on a five-year holding period. Get there, and gain on qualifying stock can be excluded from federal income tax up to the greater of a per-issuer dollar cap or 10 times your basis. Miss it, and the calculus changes.
The One Big Beautiful Bill Act (P.L. 119-21, signed July 4, 2025) softened the cliff without removing it. For QSBS acquired after that date, IRC §1202(a)(5) sets a tiered schedule: 50% exclusion at three years, 75% at four, and 100% at five years or more. The Act also raised the per-issuer cap from $10 million to $15 million and lifted the company-level gross assets ceiling to $75 million. We covered the mechanics in [what the One Big Beautiful Bill Act changed about Section 1202](https://409.ai/articles/qsbs-one-big-beautiful-bill-act-section-1202-changes).
Put the tiers next to the survival curve and the overlap is uncomfortable. The tax code's full reward arrives at exactly the point where, historically, half the cohort is already gone.
Say you incorporated in early 2026 and took founder stock at a nominal price, so your basis is close to zero. Four years in, an acquirer offers a deal that puts $8 million of gain in your hands. At the four-year mark, §1202 excludes 75%, so $6 million comes out of income and $2 million stays in. That remaining slice is not taxed at the 20% long-term rate either. Under IRC §1(h)(4), section 1202 gain counts as 28-percent rate gain, and the 3.8% net investment income tax can apply on top.
Hold twelve more months and the same $8 million is fully excluded, under the $15 million cap. The gap between those two outcomes is worth more than most seed rounds, and it turns on a date.
Selling early isn't always avoidable, and it isn't always fatal to the benefit. [Section 1045 lets you roll a QSBS gain into replacement stock](https://409.ai/articles/section-1045-qsbs-rollover-defer-gain-early-sale) within 60 days and tack the old holding period onto the new shares. That's a genuine escape hatch, though it requires you to have qualified in the first place.
The clocks that start in month one
Qualifying is the part founders get wrong, and they usually get it wrong early.
QSBS status is tested at issuance. The stock has to be issued by a domestic C corporation that meets the $75 million aggregate gross assets test both before and immediately after the issuance (§1202(d)(1)), and it has to be acquired at original issue for money, property, or services. Proving those conditions held on a particular date is a far easier exercise in the year they were met than seven years later, which is the argument for documenting eligibility with a [QSBS attestation](https://409.ai/products/qsbs) while the records are still close at hand. And the clock starts when the stock exists, not when you started working on the idea and not when an investor's money landed.
That start date is where a lot of value quietly disappears. A SAFE is not stock. Money can sit on your balance sheet for two years under a SAFE while the QSBS clock has not started at all, because [your Section 1202 clock starts at conversion](https://409.ai/articles/safes-qsbs-holding-period-conversion-section-1202). A company that raises on SAFEs, converts at a priced round in year three, and sells in year five has held qualifying stock for two years. The five-year survival milestone it just cleared buys it nothing under §1202.
The [83(b) election](https://409.ai/articles/the-83b-election-explained-for-founders) has a harder deadline still. IRC §83(b)(2) gives you 30 days after the transfer to file, with no extension and no cure. That filing is made in month one of a company with a 67.7% chance of seeing month twenty-four. The founders who skip it are usually the ones being sensible about odds. They are also the ones who, if things go well, discover that their vesting shares were taxed as ordinary income at each vest.
Both of these are decisions made when the company is least likely to survive and most likely to treat paperwork as a distraction. Neither can be fixed retroactively.
The failure case has a tax treatment too
Half of this cohort will not reach year five, so an honest article has to say what happens then.
Section 1244 is the provision founders and angels rarely hear about until it's too late to qualify. It converts what would be a capital loss on small business stock into an *ordinary* loss, up to $50,000 a year, or $100,000 on a joint return (§1244(b)). Ordinary matters: capital losses offset capital gains and then trickle out against ordinary income at $3,000 a year, while an ordinary loss lands against your salary now.
The catch is structural. Under §1244(c)(1)(B), the stock must have been issued for money or other property, not for services. And under §1244(c)(3)(A), the corporation must not have received more than $1 million in total for stock and paid-in capital as of the issuance. That test is measured at issuance, which means a company that raises a $5 million Series A doesn't retroactively disqualify the founder shares issued when it was capitalized with $100,000. We walk through the qualification rules in [how founders and angels turn a failed startup into an ordinary loss](https://409.ai/articles/section-1244-ordinary-loss-failed-startup-stock).
Section 1202 and Section 1244 are the two ends of the same distribution. Both are decided by how the stock was issued in the first eighteen months.
What compounds in the meantime
Between issuance and outcome sits the valuation record, and it behaves the same way: cheap to maintain, expensive to reconstruct.
A [409A valuation](https://409.ai/products/409a) supports the strike price on every option you grant. Under Treas. Reg. §1.409A-1(b)(5)(iv)(B)(2), the presumption of reasonableness attaches to an independent appraisal performed within the previous twelve months, provided nothing material has changed since. Rounds, major customer wins, secondary transactions and restructurings all reset that clock, which is why [409A frequency](https://409.ai/articles/409a-valuation-frequency-how-often-should-you-get-one) is a question about events rather than a calendar. [Letting a valuation lapse](https://409.ai/articles/409a-valuation-deadline-correction-procedures) pushes the consequences onto employees, who bear the §409A penalties on discounted options, not the company that let the date slide.
The survivorship data makes the case for keeping this current better than any compliance argument does. Roughly a third of establishments reach year ten. Those are the companies that get acquired, raise growth rounds, or go public, and every one of those events involves someone reading backwards through your option grants. A diligence team in year eight can tell the difference between eight annual valuations and three valuations plus a scramble.
It matters in the bad years too. A [down round resets your 409A and pushes existing options underwater](https://409.ai/articles/down-round-409a-underwater-options-repricing), and the repricing that follows has to thread §409A, ISO and ASC 718 rules at once. Companies doing that in year four are, statistically, on the good side of the survival curve. They just don't feel like it.
The line worth drawing
The useful reading of the SBA data isn't "half of you will fail." It's that the years when compliance feels most optional are the years that set the terms for everything after.
If your company is going to be one of the 33.9% that reaches ten years, almost every decision that determines what that decade is worth to you personally was made in the first eighteen months, when it looked like a coin flip. Issue the stock into a C corporation that clears the gross assets test. File the 83(b) inside 30 days. Know the date your QSBS clock actually started, which for anyone who raised on SAFEs is later than they think. Then keep the 409A current, because the record is what a buyer reads.
None of that improves your odds of reaching year five. It decides what reaching it is worth.