For the 2023–24 year, the ATO's small business benchmarks put cost of sales for Australian restaurants at 32% to 39% of turnover in the smallest band and 31% to 36% above $2 million, with an average of 34% to 35%. For cafes and coffee shops the range is higher: 33% to 42% depending on turnover, averaging 36% to 38%.
Every figure in this guide comes from the ATO's published benchmarks for restaurants and coffee shops, built from actual tax returns for 2023–24 and updated in March 2026 (checked August 2026). This is general information, not tax or accounting advice for your venue.
Why these numbers matter twice
The benchmarks are useful for the obvious reason: they are built from what Australian venues actually reported on their tax returns, not from a survey or an estimate. If your food cost runs ten points above the band for your turnover, the venues you compete with are buying, portioning or wasting differently to you.
They matter a second time because of who publishes them. The ATO's own instruction for a venue outside its key range is: "Check that you have reported all income and accounted for any trading stock used for private purposes." Sitting far outside the band is worth understanding for your own sake before anyone else asks about it.
Restaurants: the 2023–24 benchmarks
| Benchmark | Up to $500k | $500k to $2m | Over $2m |
|---|---|---|---|
| Cost of sales / turnover | 32% – 39% | 32% – 38% | 31% – 36% |
| Average cost of sales | 35% | 35% | 34% |
| Total expenses / turnover | 79% – 89% | 84% – 93% | 88% – 94% |
| Average total expenses | 84% | 88% | 91% |
| Labour / turnover | 18% – 30% | 23% – 32% | 27% – 34% |
| Rent / turnover | 11% – 17% | 8% – 12% | 6% – 9% |
Cost of sales to turnover is the ATO's key benchmark for the industry. The labour rows are labour expenses, so an owner's own unpaid hours never appear in them, which is exactly how a small venue can show 18% labour and still have an exhausted owner.
Cafes and coffee shops: the 2023–24 benchmarks
| Benchmark | Up to $250k | $250k to $600k | Over $600k |
|---|---|---|---|
| Cost of sales / turnover | 34% – 42% | 35% – 41% | 33% – 38% |
| Average cost of sales | 38% | 38% | 36% |
| Total expenses / turnover | 73% – 86% | 81% – 90% | 86% – 93% |
| Average total expenses | 79% | 86% | 89% |
| Labour / turnover | 21% – 32% | 21% – 32% | 27% – 35% |
| Rent / turnover | 10% – 17% | 8% – 14% | 6% – 10% |
Cafe cost of sales runs two to three points higher than restaurants on the averages. Note the turnover bands differ between the two industries, so the tables do not read across like for like. Milk, beans and wholesale food carry thinner mark-ups than a plated menu, and the gap is structural rather than a sign you are doing something wrong.
The arithmetic that decides whether you make money
Take the middle restaurant band and add the two costs you can actually manage week to week:
- Cost of sales at the 35% average, plus labour anywhere in the 23% to 32% band, is 58% to 67% of turnover before rent, power, insurance or a single marketing dollar.
- The ATO's benchmark average for total expenses in that band is 88%, which implies roughly 12% of turnover left before tax, on the benchmark average, and before an owner who draws rather than wages themselves is paid at all.
That is why the two numbers worth checking weekly rather than quarterly are cost of sales and labour as a share of turnover. Everything else is hard to move; those two respond to decisions you make this week. The labour side just got more expensive too: award rates rose 4.75% from 1 July 2026, so a roster that was inside the band last year can drift out of it with no change in hours.
If you are outside the band
- Check both sides of the ratio first. Unrecorded sales shrink turnover, the denominator, and push every ratio up. Stock taken for staff meals or private use inflates cost of sales, the numerator, with the same effect. The ATO's own first suggestion covers exactly these two: report all income, and account for trading stock used privately.
- Cost of sales high: run a two-week stocktake cycle, weigh the five highest-volume lines, and reprice anything whose supplier cost moved more than your menu did. Most venues find the gap in three places: portioning, waste, and quiet supplier price creep.
- Cost of sales well above the band is the one to be able to explain with paperwork, because understated turnover produces exactly that shape whether or not that is the cause. A dated stocktake and supplier price history are the explanation.
- Labour high: compare rostered hours against your actual quiet and busy periods, not against last year's habit, and check the result against where restaurant margins actually land. Fixing the roster shape usually beats cutting heads.
- Keep the working. If you are ever asked about a benchmark gap, a dated stocktake and a supplier price history answer the question in one email.
Where the controllable hours go
Labour is the band with the widest spread, and the spread is mostly about how many paid hours go to work that does not need a person: the phone that interrupts service, the supplier call that eats a manager's afternoon, the booking that rings out and books elsewhere. At current casual Level 1 award rates one of those hours costs $39.66 on a Saturday and $46.27 on a Sunday. The benchmarks tell you where you stand; the roster and the interruptions are where the movable money actually is.