Short answer
Three numbers decide whether a new business survives its first two years, and none of them is a statistic about other people. Your personal runway in months; your self-employment tax, which is 15.3% on top of income tax and catches almost everyone; and the gap between invoicing and being paid. The widely quoted survival rates are real but measure establishments going to zero employment, not businesses failing — and by sector they run the opposite way to the folklore: professional services survive worse than restaurants and retail (BLS).
1. What the survival numbers actually measure
The Bureau of Labor Statistics does publish survival data, and the headline arithmetic that circulates from it is roughly right: of private-sector establishments born in March 2013, 79.6% survived one year, 50.6% survived five years and 34.7% survived ten (BLS Business Employment Dynamics, BLS). The most recent cohort, born 2024, was at 77.9% after one year.
Two things about that data change what it means, and both are in BLS’s own definitions.
It counts establishments, not businesses. An establishment is a single physical location. A firm that opens a second shop and later closes it registers a death while the business itself is growing.
A “death” is not a failure. BLS defines one purely by employment reporting: units with no employment, or zero employment reported, in the third month of four consecutive quarters (BLS technical note). There is no test of solvency, profitability or intent anywhere in it. A founder who sells the business, retires, incorporates differently or simply stops hiring produces the same record as one who goes broke.
That matters for a claim you will see attached to these figures — that some share of the closures, often given as about 30%, were voluntary. BLS publishes no reason-for-closure breakdown at all. Business Employment Dynamics is built from unemployment-insurance administrative records, which carry no closure-reason field, so the split cannot exist in that dataset even in principle (BLS BED FAQ). The Census Bureau’s Annual Business Survey does ask why a business ceased, but that is a different agency and a different programme, and it does not rescue a figure credited to BLS.
By sector, the data runs opposite to the folklore. The belief that restaurants fail fastest and professional services are safe is not what BLS reports. Five-year survival for the March 2019 cohort:
| Sector, establishments born March 2019 | Survived 1 year | Survived 5 years |
|---|---|---|
| Retail trade | 84.9% | 60.7% |
| Accommodation and food services | 82.6% | 57.4% |
| All private sector | 79.2% | 51.5% |
| Professional, scientific and technical services | 78.0% | 50.7% |
Professional services sit below the all-industry average, and both retail and food services sit above it. The 2013 cohort tells the same story over ten years: professional services 28.6% against 34.7% for all private establishments (BLS, BLS, BLS). Choosing a sector because consultancies are supposed to be durable is choosing on a premise the data does not support.
2. Personal runway, which is the number that actually decides it
Runway is how many months of household expenses you can cover without taking anything out of the business. It is the only figure on this page that is entirely within your control before you start, and it determines how many bad decisions you can afford to avoid.
Six months forces revenue quickly, which in practice means underpricing, accepting poor-fit clients, and committing to whoever pays first. Twelve months lets you wait for the right work. Eighteen absorbs a delayed launch and a slow ramp without a crisis. These are planning conventions rather than findings — no agency publishes an optimal runway — but the mechanism behind them is real: the shorter the runway, the more your pricing is set by your own cash position rather than by the market.
The cheapest way to extend it is to overlap. Building the business while still employed converts runway from savings into salary, and it is the one lever that costs nothing but time.
3. What it costs to start, and why no agency publishes that
There is no federal figure for average startup costs by business type. SBA’s own guidance is a worksheet and a method, not a table of numbers, and it tells owners to price their own case: talk to vendors, mentors and service providers about what comparable businesses actually pay (SBA). The Census Bureau collects startup capital in banded ranges, but by demographic and sector rather than by the service-versus-storefront split people want.
So the ranges quoted confidently across the web — including in an earlier version of this page — correspond to nothing anyone measured. The honest version of this section is a method: list every cost you can name, get three real quotes for the three largest, add the ones founders reliably forget (professional insurance, the accounting software you will need from month one, the first year’s tax preparation), and then add a contingency you do not touch. What matters is that the total is yours and checkable, not that it matches a published band.
4. Where the money actually comes from
Personal and family savings dominate, and the reliable figures come from the Census Annual Business Survey rather than from any round number in circulation. About 65% of new businesses used personal or family savings to cover startup costs (2018 survey, covering 2017), and 69% of business owners said the same in a 2022 survey. The next largest source is a bank loan, at about 17% (Kauffman, analysing Census ABS, Census ABS).
Note the verb. Those figures say used, not primarily funded by, and no primary source publishes a “primarily” figure. A separate 77% that circulates attached to this question comes from the Federal Reserve’s Small Business Credit Survey and measures something different: the share of startup non-employer firms that used personal funds in response to financial challenges in the prior twelve months (Federal Reserve, 2024 Report on Startup Firms). That is a distress response, not a funding structure.
Venture capital is rarer than even the sceptical version suggests. No federal agency publishes the share of startups that raise venture capital. The nearest defensible figure comes from Census data analysed by Kauffman: about 0.5% of businesses received equity investment from venture capitalists. A separate and widely repeated “under 1%” is a share of startup funding dollars from an academic paper, not a share of firms — two different quantities that get quoted interchangeably.
SBA lending, with the programmes kept apart. The 7(a) programme lends up to $5 million, with terms up to ten years for working capital and up to 25 for real estate, at a rate set as a base rate plus a spread SBA caps by loan size — base plus 6.5% at $50,000 or less, down to base plus 3.0% above $350,000 (SBA, SBA terms). The Microloan programme is separate: up to $50,000, up to seven years, and SBA states the average microloan is about $13,000 (SBA).
That $13,000 is worth naming, because it circulates as “the average small business loan”. It is the average of a small specialty programme capped at $50,000. SBA’s main lending channel is an order of magnitude larger, and SBA publishes no “average loan for new businesses” figure at all.
5. Structure: LLC, S-corp, C-corp
An LLC gives liability separation without changing how you are taxed by default — a single-member LLC still files Schedule C. The S-corp election is a tax choice layered on top, not a different entity: you pay yourself a reasonable salary subject to payroll taxes and take the remainder as a distribution that is not subject to self-employment tax.
The saving is real and so is the cost. On $100,000 of net income with a $60,000 reasonable salary, the $40,000 distribution avoids 15.3%, about $6,120. Against that sit payroll processing, a separate return, and the fact that “reasonable” is a standard the IRS enforces rather than a number you choose. The election tends to pay for itself once net income is consistently well above the salary you would have to justify — which is a judgement about your own figures, not a threshold anyone publishes.
A C-corporation pays 21% at the entity level and is rarely right for a small business unless you are raising venture capital, which effectively requires it.
6. Self-employment tax, which is the shock
Self-employment tax is 15.3% — 12.4% Social Security plus 2.9% Medicare — and it is charged on 92.35% of net earnings, on top of income tax (IRS Topic 554). On $100,000 of net profit that is $14,129.55. Half of it is deductible.
Two ceilings matter. The 12.4% Social Security portion stops at the wage base, $184,500 for 2026 (IRS Publication 15); the 2.9% Medicare portion has no ceiling, and an Additional Medicare Tax of 0.9% applies above $200,000 for single filers and $250,000 married filing jointly — thresholds that are not indexed and so catch more people each year.
The practical consequence is that $100,000 of business income is not $100,000 of salary. Set aside a fixed share of every payment received into an account you do not spend from, and pay quarterly: 15 April, 15 June, 15 September 2026 and 15 January 2027. The safe harbour that avoids an underpayment penalty is 100% of the prior year’s tax, rising to 110% if prior-year AGI was above $150,000 (IRS Form 1040-ES).
7. The Solo 401(k), and the contribution almost everyone computes wrongly
For 2026 the one-participant 401(k) allows an employee deferral of $24,500, a catch-up of $8,000 at 50 or over, an enhanced catch-up of $11,250 at ages 60 to 63, and a total annual addition limit of $72,000 (IRS Notice 2025-67).
The employer half is where the error lives. It is not 25% of net profit. IRS defines the base as earned income: net earnings from self-employment after deducting both half the self-employment tax and the contribution itself (IRS). Because the contribution is subtracted from its own base, the 25% rate resolves to 20% of earned income — and to about 18.6% of Schedule C net profit (IRS, calculation of plan compensation for sole proprietorships).
| Employer contribution on $150,000 of Schedule C net profit | Amount |
|---|---|
| Net profit | $150,000 |
| × 92.35%, then 15.3% self-employment tax | $21,194 |
| Less one half of that tax | −$10,597 |
| Earned income | $139,403 |
| Maximum employer contribution, 20% of earned income | $27,881 |
| Plus the employee deferral | $24,500 |
| Total | $52,381 |
Applying 25% to net profit instead gives $37,500 and a total of $62,000. That is an excess annual addition of nearly $10,000 requiring correction — and note that it does not breach the $72,000 limit, so checking against the headline cap will not catch it. An earlier version of this page carried exactly that error.
8. The 2026 tax figures that changed
The qualified business income deduction is now permanent, and its thresholds moved. For 2026 the limitations begin at $201,750 of taxable income for single filers and $403,500 married filing jointly, with the phase-in range widened to $75,000 and $150,000 respectively (Revenue Procedure 2025-32). The One Big Beautiful Bill Act made the deduction permanent, widened that range, and added a minimum deduction of $400 for taxpayers with at least $1,000 of qualified business income, effective for tax years beginning after 31 December 2025 (IRS). Figures of $191,950 and $383,900 are the 2024 thresholds and are two years stale.
Mileage has two rates in 2026. The business standard mileage rate is 72.5 cents per mile from 1 January to 30 June and 76 cents from 1 July to 31 December (IRS Notice 2026-10, Announcement 2026-11). 70 cents was the 2025 rate. A log kept at a single rate for the year will be wrong for half of it.
Startup costs are deductible up to $5,000 in the first year, reduced dollar for dollar as total startup costs exceed $50,000 and gone entirely at $55,000, with the remainder amortised over 180 months (IRS Publication 583).
The home office simplified method is $5 per square foot up to 300 square feet, so $1,500 at most, and capped again by gross income from the business use of the home (IRS).
Self-employed health insurance is an above-the-line deduction, but it cannot exceed earned income from the business the plan is established under, and it is unavailable for any month you were eligible for a subsidised employer plan — eligibility, not enrolment, which catches founders whose spouse has cover they declined (IRS Form 7206 instructions).
9. The gap between invoicing and being paid
Profitable businesses run out of money in the gap between doing the work and being paid for it. If you invoice on net-30 terms and clients actually pay in 45 days, every month of expenses is a month you are financing. At $15,000 a month in outgoings and a 45-day cycle, roughly $22,500 has to exist somewhere before any of it comes back.
The levers are unglamorous and they work: invoice on delivery rather than at month end, take a deposit or bill against milestones, offer a small discount for payment within ten days, and chase on the day a payment becomes late rather than the week after. A line of credit bridges the gap but does not close it — the timing is the problem, and only collection speed fixes timing.
10. Health and business insurance
Leaving employer cover is usually the largest single line to replace. The two routes are COBRA, which continues the former employer’s plan at its full cost for up to 18 months, and the ACA marketplace, where premium tax credits are set against the income you expect this year — which for a founder in year one is often far lower than the salary just left behind. Estimate deliberately and update the estimate when income changes, because the credit reconciles at filing.
Beyond health cover, the policies a small business actually needs are general liability, professional liability or errors and omissions where you advise clients, and cyber liability where you hold client data. One gap worth checking rather than assuming: a homeowner’s policy generally does not cover business-related claims, and a home-based business rider or commercial endorsement is what extends it.
11. The mistakes that cost the most
Treating business income as salary. Self-employment tax is 15.3% on top of income tax, and nobody withholds it for you.
Computing the Solo 401(k) employer contribution as 25% of net profit. It is 20% of earned income, about 18.6% of net profit, and the difference is an excess contribution the headline $72,000 cap will not flag.
Choosing a sector on the survival folklore. Professional services survive worse than retail and food services in every BLS cohort checked.
Launching on six months of runway when twelve was possible. A short runway prices your work for you.
Using a single mileage rate for 2026. It changed on 1 July.
Quoting 2024 QBI thresholds. They moved, the deduction became permanent, and a $400 minimum was added for 2026.
Assuming the self-employed health deduction is available because you are not enrolled elsewhere. Eligibility for a subsidised employer plan disqualifies the month, enrolled or not.
Funding the business from a retirement account. A $50,000 early withdrawal triggers a 10% additional tax of $5,000 plus ordinary income tax at your own marginal rate, so the total depends on a bracket no general figure can know. The compounding you give up needs a rate and a horizon before it means anything: at 7% over 25 years, $50,000 becomes about $271,000; at 5% over the same 25 years, about $169,000.
12. Go deeper: the decision behind each section
Each decision above has a page of its own. Start with the business-startup decision engine to sequence runway, structure and tax against your own launch date. The self-employed retirement and tax guide goes further into the Solo 401(k) and SEP comparison, the emergency-fund guide covers the runway itself, and the career-change guide covers the overlap period while you are still employed. The decision tools run the numbers for your own case.
13. Common questions
Do half of new businesses really fail in five years? About half of establishments born in a given year are no longer reporting employment five years later. BLS measures establishments rather than businesses and defines a death by zero employment reporting, with no test of solvency or intent, so “fail” is doing work the data does not support.
Are restaurants the riskiest business to start? Not in the BLS data. Accommodation and food services had 57.4% five-year survival for the 2019 cohort against 50.7% for professional, scientific and technical services.
What is the average small business loan? $13,000 is the average SBA microloan, a programme capped at $50,000. SBA publishes no average loan figure for new businesses, and its main 7(a) channel lends up to $5 million.
How much can I put into a Solo 401(k) at $150,000 of net profit? $24,500 as the employee deferral plus $27,881 as the employer contribution, which is 20% of earned income rather than 25% of net profit. $52,381 in total, before any catch-up.
What is the 2026 mileage rate? 72.5 cents per mile to 30 June and 76 cents from 1 July.
When does the S-corp election make sense? When the distribution left after a defensible salary is large enough that 15.3% of it clearly exceeds payroll processing, a second tax return and the burden of justifying that salary. That is arithmetic on your own numbers, not a published income threshold.
Sources
Every figure links to the body that publishes it, and every figure that changes on a schedule carries the year or the effective date. Where no agency publishes a number — startup costs by business type, an optimal runway, the share of startups that raise venture capital — this page says so rather than supplying one. Three IRS pages were stale at the time of writing and are deliberately not cited: the standing QBI newsroom page, the self-employment tax page carrying a 2024 wage base, and the mileage page published before the mid-year change.
- Survival: BLS BED establishment survival · BLS technical note · BED FAQ · BLS, 2013 cohort at ten years · professional services · accommodation and food · retail
- Funding: Kauffman on Census ABS · Census Annual Business Survey · Federal Reserve Small Business Credit Survey, startup firms · SBA 7(a) · 7(a) terms and rate caps · SBA Microloans · SBA, calculate your startup costs
- Tax: Revenue Procedure 2025-32 · IRS on the OBBBA business provisions · Notice 2026-10 · Announcement 2026-11 · IRS standard mileage rates · Topic 554, self-employment tax · Publication 15 · Form 1040-ES · Publication 583 · Form 7206 instructions · home office simplified option
- Retirement: IRS Notice 2025-67 · one-participant 401(k) plans · plan compensation for sole proprietorships