Of the 18,340 Australian accommodation and food services businesses that started in 2022-23, 77.5% were still trading a year later and 50.4% were still trading at June 2026, so about one in five is gone inside twelve months and about half inside three years.
Skai Solutions works with restaurants, cafes and takeaways in Australia and New Zealand, and this is the benchmark operators are most often given a wrong number for. Every figure below comes from one place: the ABS Counts of Australian Businesses, including Entries and Exits, June 2022 to June 2026, released on 18 August 2026. It tracks real businesses on the business register rather than survey opinion, and it is broken out by industry division, so hospitality can be measured against everything else.
The cost that most often empties the opening buffer is the fit-out, and commercial kitchen equipment is the part of it you can stage rather than pay at once. On the ABS entries survival table, 18,340 accommodation and food services businesses entered the market in 2022-23. Here is what happened to them.
| Accommodation and food services, businesses that started in 2022-23 | Still trading | Survival rate |
|---|---|---|
| Entered 2022-23 | 18,340 | 100% |
| June 2024, about one year on | 14,216 | 77.5% |
| June 2025, about two years on | 10,980 | 59.9% |
| June 2026, about three years on | 9,238 | 50.4% |
ABS 8165.0, datacube 8165DC01, Table 3, released 18 August 2026.
So the first-year failure rate is 22.5%, not 60% and not 90%. Roughly one new hospitality business in five does not see its first anniversary. By three years the number is close to half, which is a hard figure, but it is a long way from the one most operators have heard.
Two things to hold onto before you use these. The ABS counts businesses that stop being actively trading on the register, which includes owners who sold up, retired, restructured into a new entity or simply closed a good business by choice. Not every exit is a failure. And "accommodation and food services" is the whole division, so it carries pubs, motels and caterers alongside restaurants and cafes. The ABS does not publish survival at the four-digit level that would isolate cafes on their own.
No. This is the finding that surprises people, and it is the reason this page exists. Rank all nineteen industry divisions by first-year survival and hospitality comes fifth.
| Industry division, businesses that started in 2022-23 | 1 year | 2 years | 3 years |
|---|---|---|---|
| Health care and social assistance | 81.4% | 66.9% | 59.2% |
| Agriculture, forestry and fishing | 80.3% | 67.3% | 60.2% |
| Rental, hiring and real estate services | 78.6% | 64.4% | 58.1% |
| Professional, scientific and technical services | 77.7% | 60.5% | 51.3% |
| Accommodation and food services | 77.5% | 59.9% | 50.4% |
| All industries | 75.9% | 58.1% | 49.2% |
| Construction | 74.8% | 57.2% | 48.0% |
| Retail trade | 74.7% | 56.1% | 46.2% |
| Wholesale trade | 72.0% | 54.0% | 45.6% |
| Transport, postal and warehousing | 71.8% | 48.1% | 37.4% |
Ten of the nineteen divisions shown. ABS 8165.0, Table 3, released 18 August 2026.
Hospitality beats the all-industries average at every one of the three checkpoints, and it beats construction, retail, wholesale and transport at all three. A new cafe was more likely to reach its first birthday than a new building firm, a new shop or a new trucking business. Retail, which nobody describes as a graveyard, sits nearly three points below hospitality in year one and more than four points below at three years.
The gap does narrow. This next part is our own arithmetic off the table above, so the inputs are visible. Hospitality loses 17.6 percentage points in its second year against 17.8 for all industries, then 9.5 points in its third year against 8.9. So the sector starts better than average, holds level through year two and gives a little back in year three. What it does not do is collapse in year one, which is exactly what the folklore says it does.
Nowhere you can check, which is the problem. Search the phrase and the results are qualified to the United States, Canada or the United Kingdom. There is no Australian source that produces a 60% or 90% first-year hospitality failure rate, and the ABS, which is the only body that counts every registered business in the country, produces 22.5%.
Skai Solutions has a standing rule about this kind of number, and it applies to figures that flatter us as well as figures that do not: if a statistic cannot be traced to a named publisher with a method, it does not go on a page. Our page about how this site is sourced works through another example, the widely repeated claim that most callers hang up after 45 seconds, which traces to an unnamed study cited by a vendor and is contradicted by the only large peer-reviewed measurement of call abandonment there is.
The practical cost of the folklore is that it teaches new operators the wrong lesson. If you believe nine in ten fail in year one, opening at all looks like a coin toss and year two looks like safety. The data says the opposite: year one is the part hospitality is comparatively good at, and the attrition is spread across the years after it, when the business is running normally and nobody is watching for a slow leak.
Because they are usually measuring something else, and both things can be true at once.
Australian trade and news coverage of hospitality closures generally runs on insolvency and business failure data from commercial credit bureaus, which counts formal administrations and liquidations in a calendar period and reports them as a rate against the number of businesses in the sector. That is a count of a specific legal event, over twelve months, and it can hit a record in a year when costs and interest rates bite.
The ABS table on this page measures something different: whether a business is still actively trading on the business register, three years after it started, whatever the reason it stopped. It includes orderly closures and sales as well as insolvencies, and it is a cohort followed over time rather than a snapshot of one year.
So a rising insolvency rate and a mid table survival rate are not in conflict. If you want to know how bad this year is, the insolvency numbers are the right tool. If you want to know whether hospitality is structurally riskier than opening a shop or a building firm, the ABS cohort is, and the answer it gives is no.
The ABS does not publish an average lifespan, and any single figure quoted as one is a summary of somebody's assumption rather than a measurement. What it does publish is the survival of the whole existing population, which answers the same question more honestly.
Of the 110,368 accommodation and food services businesses trading in June 2022, 59,294 were still trading in June 2026. That is 53.7% over four years, and it covers established venues, not just new ones.
ABS 8165.0, datacube 8165DC01, Table 2, released 18 August 2026.
Read those two tables together and you get the useful shape. New entrants survive their first year at 77.5%. The standing population, which is mostly businesses that already got through the risky years, still turns over at about 12% a year. Hospitality is not a sector where you survive the opening and then coast. It is a sector with steady annual churn at every age, which is why the operators who last are usually the ones who fixed small recurring leaks rather than the ones who had a good first year.
Fifteen minutes, your numbers, and the calls that never reached a table.
Book a Free DemoThe ABS counts exits, it does not ask why, so anyone giving you a ranked list of causes is guessing. What can be said with sourced numbers is where the money actually goes, and that is enough to show why the margin for error is thin.
On the ATO Small Business Benchmarks, an Australian restaurant turning over A$65,000 to A$500,000 reports cost of sales averaging 35% of turnover, labour of 18% to 30%, and rent of 11% to 17%. Take the midpoints and 73% of turnover is committed before electricity, gas, waste, insurance, accounting, repairs, card fees or delivery commission. Average total expenses across that band are 84%, so everything else in the business has about eleven points to live in. We work that through on what rent should be as a percentage of turnover and restaurant profit margin in Australia.
Inside eleven points, things that look trivial are not. A delivery marketplace headline rate of 30% of order value sits in there, which is why we published what delivery app commissions actually cost. So does an unanswered phone, and that one has no line in any report, because a customer who could not get through simply ate somewhere else.
No, and it is worth saying so plainly rather than reading the rates across. Every figure on this page is Australian, taken from the Australian business register. This page does not carry a New Zealand survival rate, because we have not verified one published to the same industry-division detail, and quoting an Australian rate as a New Zealand one would be exactly the kind of borrowed statistic this page is arguing against.
What we do hold for New Zealand is the size and shape of the market, from Stats NZ: 9,858 cafes and restaurants and 7,422 takeaway food services at February 2025, geographic units, rounded to base 3. Our page on hospitality wages in New Zealand covers the cost side under New Zealand instruments rather than Australian ones.
Nothing dramatic, and it would be dishonest to claim otherwise. No phone system saves a venue with the wrong site or the wrong menu.
But the survival curve says the danger is not the opening, it is the ordinary years afterwards, and the ordinary years are where small recoverable revenue goes missing without appearing anywhere. A booking that rings out during service is turnover you have already paid the rent and the roster on. At the small-band midpoint you carry 14% of turnover in rent whether the table sits full or empty, so a recovered booking lands against a cost you could not avoid either way.
That is the whole argument for an AI receptionist, and it goes live in five days. It is also worth being clear about who should not buy one. If your phone is quiet and your problem is that Tuesday is dead, a phone agent will not fix Tuesday. Our page on the cheapest way to answer the restaurant phone lists the free options first for exactly that reason, and for a lot of venues a good voicemail message and a diverted mobile is the right buy. Before you add it to the overheads, read what the cost of an AI receptionist actually is per month.
Not every exit is a failure, and the difference is usually preparation. How much a restaurant is worth covers the three ways a venue gets valued, and why a short lease cost one fourteen year old cafe more than four fifths of its asking price.
Small venues fail on thin margins and missed trade more than on bad food. Does an AI receptionist work for a coffee shop prices what an unanswered cafe call is actually worth before you decide it is not a problem.
About 22.5%. Of 18,340 accommodation and food services businesses that started in 2022-23, 77.5% were still trading a year later, on ABS Counts of Australian Businesses released 18 August 2026.
The ABS entries table runs to three years for this cohort, and at three years 50.4% were still trading. There is no Australian source that supports a 90% failure claim at any horizon.
Not on first-year survival. Accommodation and food services ranks fifth of nineteen industry divisions at 77.5%, above the all-industries average of 75.9% and above construction, retail, wholesale and transport.
Five days from sign up to live calls with Skai Solutions. You keep your number, and the divert can run during service only, after hours only, or all the time.
More: what labour should cost as a percentage of turnover, restaurant food cost percentage, or every guide on this site.