How to Calculate Prime Cost for a UK Restaurant
Prime cost is your cost of sales plus your total labour cost, measured against sales excluding VAT.
prime cost % = (cost of sales + total labour) ÷ ex-VAT sales × 100
Those are the two lines that move every week and the two you can actually do something about before the month closes. Rent does not change because service was quiet. Food and labour do.
Get the denominator right first
This is where most UK prime cost calculations go wrong, and it is not a rounding error.
Restaurant food and drink consumed on the premises is standard-rated, and the standard rate of VAT is 20%. The money in the till is not your revenue. One fifth of it — strictly, one sixth of the gross — was never yours.
Take a week's takings of £12,000 including VAT. Ex-VAT sales are £12,000 ÷ 1.2 = £10,000. Suppose cost of sales for that week came out at £3,300 and total labour at £3,100.
- Prime cost = £3,300 + £3,100 = £6,400
- On ex-VAT sales: £6,400 ÷ £10,000 = 64%
- On the till figure: £6,400 ÷ £12,000 = 53.3%
The same week reads either as a business with 36% left for everything else, or as one with 46.7% left. The second number is fiction and it is nearly eleven points of fiction. Every benchmark you will read is quoted against net turnover — the UKHospitality and Christie & Co benchmarking survey collects operating costs "by reference to percentage of net turnover" — so if you compute yours on gross takings you are not measuring badly, you are measuring something else.
What actually goes in the labour line
Labour is not the wage bill. The benchmarking survey defines payroll as "gross payroll costs including staff wages, PAYE, NICs and manager's salary", and the gap between the hourly rate you quote at interview and what that hour costs you is large.
Work it through for a full-time employee aged 21 or over, 38 hours a week, on the National Living Wage.
- The National Living Wage from 1 April 2026 is £12.71 an hour for workers aged 21 and over (£10.85 for 18 to 20, £8.00 under 18 and for apprentices).
- Gross pay: 38 × £12.71 × 52 = £25,115 a year.
- Employer's National Insurance: 15% above a secondary threshold of £96 a week for 2026–27. That is £4,992 a year of earnings below the threshold, so 15% × (£25,115 − £4,992) = £3,018.
- Employer's pension: the automatic enrolment minimum employer contribution is 3% of qualifying earnings, banded between £6,240 and £50,270. 3% × (£25,115 − £6,240) = £566.
- Total employer cost: £25,115 + £3,018 + £566 = £28,699.
Now divide by the hours you actually get. Statutory holiday is 5.6 weeks, so a full-timer works 46.4 weeks and is paid for 52. That is 38 × 46.4 = 1,763 hours on the floor.
£28,699 ÷ 1,763 = £16.28 an hour actually worked.
A £12.71 rate costs £16.28 on the pass — 28% more than the number in your head — and that is before sick pay, training time, recruitment, or a rota that overstaffs a wet Tuesday. Any rota model built on the headline rate is understating labour by roughly a quarter.
The benchmarks worth using
Food cost: 30–35% of ex-VAT sales, which is a food gross profit of 65–70%. Be careful with what circulates online. The "25–35%" and "28–35%" bands you will find on most costing blogs are recycled from US restaurant-association material — the tell is dollar figures and no VAT discussion anywhere on the page. UK primary data does not support a whole-menu 75% food GP.
Labour: expect payroll in the high twenties to mid thirties as a percentage of net turnover. The most detailed UK primary source is the UKHospitality and Christie & Co Benchmarking Report. Its last full edition — 40 companies covering 3,548 managed outlets, for the year ending September 2017 — put typical payroll at 29.4% of turnover, 1.5 percentage points more than the year before, with the average manager's salary alone at 7.4% of turnover. Casual dining ran highest of the licensed segments at 33.5% of revenue. The same report records payroll having "risen from 17% to over 29% since the introduction of the National Minimum Wage Act in 1998".
Two health warnings on that. It is licensed-sector data spanning food and wet sales, so it is a direction of travel rather than a target for your site. And it is old — the wage floor has risen every year since, so 29.4% is a floor to reason up from, not a benchmark to hit. Work out your own using the fully-loaded hourly figure above; that is the only number that describes your rota.
Put the two halves together and a typical UK operator's prime cost lands somewhere in the low-to-high sixties as a percentage of net turnover. The American "keep prime cost under 60%" rule does not transfer, because it is quoted on sales that never had 20% VAT inside them and against a labour market with different statutory on-costs.
Run it weekly, not monthly
Prime cost is only useful at the frequency you can still act on it.
- Fix a period and use the same one for both halves. If payroll runs weekly and the stock take runs monthly, your prime cost is comparing four weeks of labour with a month of food and the answer will bounce for no operational reason.
- Cost of sales comes from the count, not from invoices alone: opening stock + purchases − closing stock. Purchases on their own tell you what you bought, not what you used.
- Put the owner's labour in at a real rate. If you work sixty hours on the line and take a small draw, your prime cost is flattered by the difference and it will not survive the day you hire a replacement.
- Include employer NI, pension and holiday accrual, using something like the £16.28 figure above rather than the contracted rate.
- Split the number after you have it. A 66% prime cost made of 31% food and 35% labour is a rota problem. The same 66% made of 37% food and 29% labour is a pricing or a yield problem. The headline tells you there is one; only the split tells you which.
Mistakes that make the number lie
- Dividing by gross takings. Covered above. It is the single most common error and it makes a struggling site look healthy.
- Comparing a VAT-inclusive menu price with an ingredient cost that is net. Most food bought for a kitchen is zero-rated, so the invoice is already net; the menu price is not. Bring the selling price to ex-VAT before you divide.
- Leaving delivery commission in cost of sales. Aggregator commission is a cost of that sale, but it is not food and it is not labour. Track it separately or your food cost percentage will move every time the delivery mix shifts.
- Averaging the whole week. A prime cost that is fine across seven days can hide two sessions running at 90% and five at 55%.
- Treating the target as the goal. Prime cost is a diagnostic. The goal is the cash left after it, and a site with a 68% prime cost and cheap rent can be in better shape than one at 61% paying a high-street lease.
Where to start
Cost the dishes first — a prime cost built on guessed portion costs is a guess with more decimal places. Run a dish through the free food cost calculator, scale a batch recipe in the recipe costing tool, and if you want the food and labour halves kept in one worksheet week after week, the PrepSheet templates include a weekly GP tracker and a labour cost calculator built for UK rates.
Run the free calculator → · See the templates → · Save your recipes free →