Half of New Businesses Survive Five Years. Surviving Is the Wrong Test.

Nearly three in five South Africans who see a good business opportunity say they would not act on it because the business might fail, according to the Global Entrepreneurship Monitor's South Africa profile. The fear usually arrives with one statistic attached, phrased as "9 out of 10 new businesses fail." It is one of the most quoted numbers in small business, and it puts off people who should at least look at the evidence.

The number does not survive a close look. Government data puts the share of new employer businesses still trading after five years at about half. But the close look does not end in reassurance, and this article will not pretend it does. Take the closure statistic apart and a different number turns up, one that a corporate professional will find harder to argue with: what the survivors earn.

The numbers

• 49.2% of new US employer establishments survived at least five years, and 33.9% survived ten, averaged across 1994 to 2022 (US Small Business Administration, Office of Advocacy, Frequently Asked Questions About Small Business, February 2026, using Bureau of Labor Statistics data).
• 69.5% of US establishments that reach year five go on to reach year ten (US Small Business Administration, Office of Advocacy, February 2026).
• About 75% of venture-backed US companies never return cash to their investors, in a study of roughly 2 000 companies (Shikhar Ghosh, Harvard Business School, reported 2012).
• Nearly three in five South Africans who see a good opportunity would not start a business for fear it might fail (Global Entrepreneurship Monitor, South Africa economy profile).
• 27.1% of South African informal business owners netted R0 to R100 in the month before the survey, and 4.6% netted more than R15 000 (Stats SA, Survey of Employers and the Self-Employed 2023, released 18 March 2025).
• The median US entrepreneur still in business after ten years earned about 35% less than a comparable paid employee (Barton Hamilton, Journal of Political Economy, 2000).

The 9 in 10 figure has no clean source

Harvard Business School's Shikhar Ghosh reported in 2012 that about 75% of venture-backed companies never return cash to their investors, and that in 30% to 40% of cases investors lose most or all of their money. That is the strongest research behind the failure narrative. It describes companies that raised outside capital, which is a small slice of the businesses people actually start.

Ghosh's own definitions show how much the answer depends on the question. If failure means investors lose most of their money, the rate is 30% to 40%. If it means missing projected returns, it is 70% to 80%. The 90% figure only appears at the third definition, where failure means missing a projection the company made itself.

The version repeated in blogs and pitch decks is harder to pin down. It usually traces to a 2019 Startup Genome report. When the founder Stephen Hallgren followed that citation in 2024, it led to a Small Business Trends article he could not find. A review on the founder site We Are Founders notes that Startup Genome gives no clear definition of failure and does not disclose its sample size.

What to do with this: when you meet the figure, ask what it measured, closure or a missed return target. A number without a definition tells you nothing.

Government data puts five-year closure at roughly half

An average of 49.2% of new US employer establishments survived at least five years across 1994 to 2022, according to the US Small Business Administration's Office of Advocacy (February 2026), which tabulates Bureau of Labor Statistics data. Two-year survival averaged 67.7% and ten-year survival 33.9%. The UK's Office for National Statistics reports five-year survival of 43.5% in its best-performing region, the South West, for businesses born in 2019. Across the US and UK figures, five-year closure runs from about half to somewhat more than half.

Half is a poor guide to your own odds, though, for several reasons.

Risk falls with age. Of the establishments that reach year five, 69.5% reach year ten, and of those that reach year ten, 76.1% reach year fifteen (US Small Business Administration, Office of Advocacy, February 2026). The five-year figure describes a founder on day one. It says little about one already in year three.

Sector matters. Five-year survival differs by about 26 percentage points between the best and worst US industries, with capital-heavy, slow-entry industries lasting longest and low-barrier ones churning hardest (BLS Business Employment Dynamics, as tabulated by The Broker Shop, July 2026, and StartBusinessByState, 2026).

The series also counts only establishments with employees. Most US small businesses have none: 81.9% of US small firms are nonemployers (US Small Business Administration, Office of Advocacy, 2024). A consultant working alone from a home office is outside the denominator, which matters if you plan to start alone.

Closure is not always failure, either. Brian Headd's analysis of US Census data found that about a third of closed businesses were judged successful by their owners at closure. Alex Coad argued in 2014 that the research literature leans too far toward recasting closures as successes, and that most exits are unsuccessful. I treat Headd's third as a ceiling, because it rests on owners describing their own outcomes.

What to do with this: if you are already trading, the day-one figure no longer describes you. Each year you survive improves the odds.

South Africa has no reliable failure rate

The claim that 70% to 80% of South African small businesses fail within five years has no tracked cohort behind it that I could find. Academic papers repeat it, and the trail leads to a 2003 conference paper by Brink, Cant and Ligthelm, which presents it as an estimate. Other papers repeat different versions. A 2023 theoretical review of small business in South Africa cites 50% failing in the first year, 70% in the second and 80% within ten years. I could not find the cohort data that would let anyone check either version.

What South Africa does publish is thinner and easy to misread. A registered business is not a trading business. SARS reports that of 900 285 registered VAT vendors in 2024/25, 496 858 (55.2%) were active. Of the 1 228 437 companies assessed for the 2023 tax year, 54.0% declared taxable income of zero and 24.3% an assessed loss (SARS and National Treasury, Tax Statistics 2025). The company register cannot tell you how many businesses survive. I covered the gap between registering a company and running a business in Everyone Wants Out of Corporate. Almost Nobody Is Preparing.

The Global Entrepreneurship Monitor's established-business ownership rate is no better as a proxy. It was 1.8% in the 2022/23 report and 5.9% in 2023 (GEM South Africa). A swing that size looks more like sampling noise than business dynamics, and the measure counts the stock of owners at a point in time rather than following a cohort.

Whatever the true rate is, it pools very different founders. Stats SA's Survey of Employers and the Self-Employed 2023 found that 57.1% of people running informal businesses started because they were unemployed or had no other source of income. A corporate professional who resigns with savings and industry knowledge is a different founder from a retrenched trader with no cushion. An average across both describes neither. The conditions are also harder than in most countries: GEM scored South Africa's entrepreneurial environment 3.6 in 2023, third lowest of the 49 economies it measured.

What to do with this: do not quote a South African failure rate, because there is not one. Use the international range as your reference class, then adjust for your own position rather than for an average nobody has measured.

The number to worry about is what the survivors earn

Barton Hamilton's 2000 study in the Journal of Political Economy found that the median US entrepreneur still in business after ten years earned about 35% less than a comparable paid employee. Survival and success are separate measures, and failure statistics only capture the first.

The picture is more mixed at the top. Kneiding and Kritikos (2013), using German household data, found that the self-employed earn more than employees on average but less at the median. A few large winners lift the average and the typical founder sits below the typical employee.

South African data show the same split between staying alive and earning. In Stats SA's Survey of Employers and the Self-Employed 2023, 22.4% of informal business owners had run their business for more than ten years, up from 11.0% in 2001. In the same survey, 27.1% netted R0 to R100 in the month before the interview, and only 4.6% netted more than R15 000. In every industry, at least half netted R1 500 or less. Those are informal, non-VAT-registered businesses, so they are not your business. They do show that a business can last a decade and pay its owner almost nothing.

Here is a composite. A logistics manager in Durban earning R85 000 a month leaves to run a freight consulting practice. By year three she has four clients and the practice pays her R55 000 a month. She has beaten the failure statistic. She is also R30 000 a month worse off than the day she resigned, before counting the pension contributions and medical aid her employer used to fund. No failure statistic would flag her business as anything other than a success.

Economists Erik Hurst and Benjamin Pugsley found in 2011 that about 75% of new US small business owners reported no desire to grow, and that over half cited non-financial reasons such as flexibility or being their own boss. For those owners, a stable business that pays modestly is the goal. For someone leaving a corporate salary the goal is usually different, and that mismatch is worth settling before you resign, not after.

What to do with this: write down the monthly income the business must pay you, and by when, to beat the salary you are giving up. The break-even calculator shows what the business must sell per day to get there, and the franchise calculator does the same for a franchise.

Why businesses close, and what moves your odds

CB Insights' 2026 analysis of 431 venture-backed companies that shut down since 2023 found that 70% ran out of capital, but it treats that as the last event rather than the cause. Poor product-market fit was the most common underlying reason, at 43%, ahead of bad timing at 29%. GEM's 2024 global survey of people who closed a business points the same way: about 30% said the business was not profitable, and about 16% cited problems getting finance.

I read those as one problem. Revenue arrives more slowly than the founder can afford to fund. Demand is the underlying issue and cash is the clock. It is why reaching the first customer breaks budgets, which I set out in Why Most R5 000 Business Ideas Fail Before They Reach a Single Customer, and why a business can look profitable while the bank balance falls, which the guide on profit versus cash flow explains.

The evidence on what improves the odds is mostly about experience and resources. Pierre Azoulay and colleagues, using US Census data (American Economic Review: Insights, 2020), found that prior experience in the specific industry predicts much higher rates of entrepreneurial success. The average founder of the fastest-growing 1 in 1 000 new ventures was 45. Their outcome is high growth rather than survival, so read it as direction rather than proof. Aaron Chatterji's 2009 research, cited in that paper, found that ventures started by former employees of established firms outperform other new entrants, an advantage linked to industry knowledge. Brian Headd found that firms with more capital and employees survive longer, and that owners with more education are more likely to stay open.

These are correlations. People with industry experience and savings also pick better sectors and can absorb a slow year, so no study can say how much each factor contributes. The direction is consistent, though, and it favours the corporate professional who builds close to what they already know. I made that case in How to Find a Business Idea in South Africa: the best ideas come from deep industry knowledge, not from brainstorming.

Buying changes where you join the curve. Survival attaches to the age of the business, not the age of the owner, so someone who buys a ten-year-old business steps into the flatter part of it. Buying brings its own risks, including the price paid and what the numbers do not show, and the start, buy or franchise comparison and The Franchise Trade-Off Nobody Puts in the Brochure cover them.

Timing also sits within your control. I argued in the piece on leaving corporate that the people who start before they have to are better placed than those who start because a retrenchment letter arrived. The preparation guide covers how to use the salary while you still have it.

What this means practically

The failure statistic cannot be changed. Five decisions made before you resign can be.

Set your income number before you pick the business. Work out what the business must pay you each month, and by when, to beat the salary you are giving up. If the answer needs more than the business can plausibly deliver, you have learned that cheaply.

Write your stop rule while you are calm. Decide the date or cash balance at which you return to employment. The runway calculator shows how many months you have, and can I afford to leave my job works through the personal budget behind it. If the plan needs outside money, read Small Business Funding in South Africa: What Nobody Tells You Before You Apply before you need it.

Start close to your industry. Test the idea against what you already know. How to Test a Business Idea in South Africa explains how to tell real validation from polite enthusiasm.

Test demand while you are still employed. Your contract and seniority decide what you are allowed to do, so read Starting a Business in South Africa While Still Employed first.

Cost the customer, not the company. Registration is cheap. Reaching someone who has never heard of you is what breaks the budget.

If you are still deciding whether to leave at all, start with should I leave corporate and start my own business. If you already have an idea, will my business idea work sets out which evidence counts.

Frequently asked questions

What percentage of new businesses fail in South Africa?

South Africa has no reliable national figure. The often-quoted 70% to 80% failing within five years traces to a 2003 estimate by Brink, Cant and Ligthelm, not to a tracked cohort. The closest hard data is American: 49.2% of new US employer establishments survived five years, averaged across 1994 to 2022 (US Small Business Administration, Office of Advocacy, 2026).

Do 9 out of 10 startups fail?

Not businesses in general. Harvard's Shikhar Ghosh found about 75% of venture-backed companies never return cash to investors, and 90% appears only when failure means missing your own projection. For all new US employer establishments, 49.2% survive five years and 33.9% survive ten (US Small Business Administration, Office of Advocacy, 2026).

Do small business owners earn less than employees?

At the median, often yes. Barton Hamilton's 2000 study in the Journal of Political Economy found that the median US entrepreneur in business for ten years earned about 35% less than a comparable employee. In South Africa, 27.1% of informal business owners netted R0 to R100 in the month before Stats SA's 2023 survey.