Profitable Pub Food: 12 Dish Types and the GP Maths (2026)

You are looking at a menu you suspect is not paying its way, and you cannot prove it. Nobody ever showed you how to work the numbers on a plate.

This guide fixes that: named dishes, real UK prices, gross profit with VAT stripped out, and the cost nobody puts on the plate, chef minutes. I run The Anchor in Stanwell Moor as a Greene King tenant, so every figure has to survive a rota.

Every worked figure below comes from one illustrative pub: 4,000 pounds a week in net food sales, around 170 mains Monday to Saturday, 80 roasts on a Sunday. A model, not The Anchor's books. Scale it to your site.

Gross profit is not profit

Gross profit pays nothing on its own. Not rent, not rates, not energy, not the chef.

Take that 4,000 pounds a week at a genuine stocktaken 68 percent GP: 2,720 pounds of gross profit, with waste already inside it, because a stocktake counts what left the store rather than what reached a plate. Now take off kitchen labour at 880, front of house at 450, kitchen energy at 160, disposables at 70, card fees at 72, maintenance at 40. About 1,048 pounds is left before a penny of rent or rates lands.

So judge every dish on three things: what it makes in cash, because a percentage never paid a rent bill; how many chef minutes it eats, because on a Saturday that is your binding constraint; and what it pulls alongside it, being honest that tied draught blends out around 55 to 60 percent GP, not the seventy-plus the trade press quotes.

The 12 most profitable types of pub food

Four things decide whether a dish earns money: ingredient cost, prep and service time, perceived value, and how many other dishes it cross-uses. A profitable dish needs at least three. Costs are 2026 prices, so re-cost against your own invoices.

One caution on braising

I have seen this go wrong. Cook in bulk during a quiet daytime shift on a timed cook, then blast-chill within 90 minutes and portion cold. Never leave a pot in a cooling oven overnight: food sitting for hours between 8 and 63 degrees is an environmental health problem, and most pub insurance voids cover on appliances left running in unattended premises.

The quiet winners nobody costs

Chips costing 55 pence sell at four pounds fifty. That is 3.20 pounds of cash gross profit for one minute of work, on a cover you have already paid to acquire.

Be realistic about the base. Most mains already come with chips, so the upsell is the chargeable extra: onion rings, garlic bread, a sauce. Across 250 food covers a week, moving that attach rate from 20 to 35 percent is 38 extra items, worth 6,240 pounds a year. Our guide to upselling scripts that work at the table covers the asking.

The dishes that look profitable and are not

Fresh fish specials at low volume, because three portions a week cannot be bought sensibly. Charcuterie boards, because you add nothing to bought-in premium. And delivery: on a 16.95 pound dish you keep 14.13 net, pay 5.09 commission (30 percent of the gross order value), 60 pence packaging and 5.30 food cost, leaving 3.14. The same dish in the room contributes 8.83, and that cover buys drinks. Check your commission basis: charged on the net value it is 4.24, and contribution rises to 3.99.

What GP should pub food run at?

That bottom row is the honest one: plenty of kitchens believe they sit at 70 percent because they have only ever worked recipe GP, which is always optimistic. If food exists mainly to hold people in the room, chasing 70 percent on a nine-dish menu shrinks portions and costs you the drinks it was selling. See our comparison of wet-led and food-led pub economics.

Working out GP with VAT handled properly

Net selling price = menu price divided by 1.2
Cash gross profit = net selling price minus net plate cost
GP percentage     = cash gross profit divided by net selling price

Both sides must be net. Your menu price includes 20 percent VAT on anything eaten in; your supplier invoice shows a net line, so cost from that.

Work it on the inclusive price instead and you get 15.95 less 4.05, divided by 15.95: 74.6 percent. You have just invented 5.1 points of gross profit.

The error is always 20 percent of your inclusive-price cost ratio. A side overstates by two points, a burger by five, fish and chips by six, a steak by seven. The higher the plate cost, the more the mistake flatters you, which is precisely backwards. A pub proudly running 70 percent on inclusive maths is running 64 percent for real.

Two caveats. Under the 90,000 pound registration threshold you are not VAT registered, so you do not divide by 1.2 at all and your plate cost is the gross invoice figure. On the Flat Rate Scheme, under 150,000 pounds, it differs again. Our pub VAT and accounting guide covers both.

The zero-rating most kitchens miss

Cold food taken away is zero-rated. Hot food is standard-rated wherever it is eaten, so a cold sandwich to go is zero-rated and a toastie never is.

An eight pound cold sandwich with a 1.80 plate cost makes 4.87 pounds eaten in. Taken away it makes 6.20, or 5.95 after packaging: 22 percent more cash for the same plate. You must evidence the split at the till, though, or HMRC treats the lot as standard-rated.

Costing a dish when your till will not do it

1. Weigh every component as it leaves the pass, not as the recipe claims.
2. Divide the case, not the guess. A 2.5kg box of patties at 41.80 pounds gives 22 patties, so 1.90 each. This is where most costing goes wrong.
3. Yield test every protein. A side of salmon loses 10 to 18 percent to trim, pin bones and belly, so raw-weight costing invents around five points of GP. Same for chicken, beef shin and potatoes.
4. Write it on a photo-carded recipe sheet. No card, no menu slot.
5. Re-cost your top 15 dishes quarterly, monthly on beef, dairy, oils and eggs.

For a target GP: divide plate cost by one minus the target, then multiply by 1.2. A 4.05 cost at 70 percent gives 16.20, so round up to 16.50.

The cost nobody puts on the plate: kitchen time

A chef on 14 pounds an hour costs closer to 18.20 once holiday accrual, employer National Insurance and pension land on top. Call it 30 pence a chef minute; our guide to pub wages and labour costs works the on-costs properly.

The homemade pie wins by 30 pence. But the result hangs entirely on that prep figure, so run the break-even rather than argue about it: the two land level at nine total chef minutes a plate, and above nine the bought-in pie makes more money despite 11 points less gross profit. The question is not which pie you prefer, it is whether your prep is genuinely batched.

One honest caveat. A rostered chef costs the same whether the pie is yours or the supplier's. Chef minutes only turn into cash if you take them off the rota, or fill them with covers you are turning away.

My rule: a dish needing more than eight minutes hands-on at the pass has to be one of the highest cash GP dishes in its section, or it comes off the Friday and Saturday menu. Time your ten busiest dishes during a real service. Every operator I have done this with was well under on the estimate.

Then compare sessions, at 9.26 pounds average cash GP a cover. A Sunday doing 80 roasts at 9.54 across 14 chef hours, including the Saturday prep, makes 54.51 pounds per chef hour. A Tuesday doing 25 covers across 6 chef hours makes 38.58. That is why the roast carries the weakest percentage on your menu and is still the best decision in your week.

Menu engineering from your own till data

Plot every dish on two axes: how popular it is within its own section, and how much cash it makes per portion. The grid and the names come from Kasavana and Smith's menu engineering work in 1982.

Build it in five steps.

1. Check your PLU hygiene first. If half your food rings through as "Main 16.95", fix the buttons and come back in 12 weeks. Dirty data breaks everything below.
2. Export item-level sales by PLU for a clean 12-week window, excluding the fortnight after a menu change. Twelve weeks is one season, so a winter dog can be a summer star.
3. Run starters, mains, sides and desserts as separate grids, because a dessert and a steak cannot share a threshold. Attach a costed recipe card to every PLU.
4. Set the popularity threshold at one divided by the number of dishes in the section, times 0.70. With ten mains that is 7.0 percent, and a dish sitting exactly on it counts as popular.
5. Set the margin threshold at the section's weighted average cash gross profit, not the simple average.

An illustrative mains section modelled on a typical mix: ten dishes, 2,040 portions over 12 weeks, weighted average cash GP 9.26 pounds, popularity threshold 7.0 percent.

Stars get protected, never discounted. Do not touch the spec or the supplier. Put them first, photograph them, and test a 3 to 5 percent rise: demand is proven, so they are your least price-sensitive dishes.

Plough-horses get engineered, not shrunk. Attack the plate cost, never the portion. Negotiate the burger patty from 1.90 down to 1.80 in a supplier review, lift the burger from 15.95 to 16.45, and cash GP goes from 9.24 to 9.76. On 19 percent of 170 mains a week that is about 870 pounds a year, and no customer sees a smaller plate.

Fish and chips is trickier, because it is the most price-transparent dish in the trade and your regulars benchmark it against the chippy. Move it in a smaller step, 16.95 to 17.50, and watch the mix for six weeks. Cash GP rises to 9.29, worth about 850 pounds a year.

Now the bit almost every article skips. Raising a 21 percent mix dish also raises the section's weighted average, which is the threshold itself. After both moves the average is 9.45, not 9.26. The burger at 9.76 has crossed into Star territory. Fish and chips at 9.29 has not: a better plough-horse, still a plough-horse. Recompute the average after every move.

Puzzles are a marketing problem, not a kitchen problem. The lamb shank makes 10.23 pounds and nobody orders it. Move it to the top of the section, box it, rewrite the description with provenance and texture, run it as the Friday special, and give it six to eight weeks.

Value that win carefully. If covers stay flat, doubling the shank's mix takes those plates off other dishes, so the gain is the difference in cash GP, not the whole 10.23. Against the 9.26 average that is 97 pence a plate, about 340 pounds a year: a tenth of the naive sum.

Dogs come off, mostly. Delete the scampi. Hold the sea bass one cycle with a rewritten description, because it is your only lighter, non-fried fish, then delete it if the mix does not move. Deletion pays through less waste, less prep and faster tickets, not through the sales line, so do not expect the till to cheer.

The exception is a veto dish. The vegan curry sits just under the section average, but one person can decide where a party of eight eats. Re-engineer it rather than cutting it.

Those three moves are worth roughly 2,060 pounds a year and change two prices. A blanket 4 percent rise draws far more resistance for a similar result. For the whole-ticket version, see our guide on menu engineering to lift average spend.

Pricing without losing your regulars

About 83 pence in every pound of a price rise reaches gross profit once VAT is out, and it carries no extra cost at all, as long as volume holds. Nothing else on this page is that efficient.

Reprice twice a year in small steps. Two 4 percent moves land invisibly; one 8 percent move gets discussed in the car park. Build three or four clear price tiers rather than clustering six mains on the same number, because a flat block removes the customer's ability to trade up. Set a high anchor with one deliberately expensive sharing dish: the one dish allowed to break the Dog rule, because it does not need to sell, it needs to be seen.

What changed at The Anchor

I did all of this in my own kitchen first, in the order above. Costed every dish ex-VAT from delivered prices. Cut the menu back to what genuinely sold. Cross-used ingredients so one braise fed four lines. Priced on true cost, not on what the pub down the road charges.

We grew food revenue by 98% in three months. No rebrand, no gimmick, just knowing what every plate made and acting on it.

Five mistakes that quietly cost pubs money

Working GP on the VAT-inclusive price. Five to seven points of imaginary profit, worst exactly where your costs are highest.

Never stocktaking food. A well-run kitchen holds the gap between theoretical and actual GP under two points. On 4,000 pounds a week, a six-point gap is 12,480 pounds a year, roughly a third of a full-time chef once on-costs land.

Cutting portions to hit a GP target. The plate looks mean, the mix drops, the reputation follows. You have swapped real cash for a better-looking percentage.

Ignoring labour, so hand-made always looks better than it is. Kitchen labour of 20 to 26 percent of net food sales is a fair working band, and most kitchens that have never timed a dish sit well above it.

Not recording staff food, comps and wastage. Without those three columns in a waste book at the pass, your stock gap is unattributable, and unattributable gaps get blamed on theft when the answer is portioning.

Your action plan

Week 1, twelve to sixteen hours. More than anyone else promises, and still the best-paid week of work in your year.

• Monday before service: count the food stock. That is your opening figure, because a single count cannot give you an actual GP.
• Monday: export 12 weeks of item-level sales by PLU, after checking the buttons are clean.
• Tuesday and Wednesday: weigh and cost your top 15 dishes from delivered net prices, yield testing every protein.
• Thursday: work those 15 dishes ex-VAT and record cash gross profit, not just percentage.
• Friday: time your ten busiest dishes with a stopwatch during service.
• Then log every food invoice for four weeks and count again. Opening stock plus purchases minus closing stock is your real cost of sales, and the gap against your recipe figure is what you are after.

Week 2: cut, price and relaunch.

• Build the grid by section, then delete the dogs, keeping any genuine veto dish.
• Move two prices, no more, and give one puzzle the best position on the menu with a single line for the team to say.
• Redo the allergen matrix and any PPDS labelling before printing. Deleting dishes and changing specs invalidates what you have.
• Plan the reprint properly: design, proof, print, plus the website, Google Business Profile, delivery menus and specials boards. Allow a fortnight and 200 to 600 pounds, not an afternoon.

Month 2: make it a routine.

• Weigh 20 random plates against spec and publish the result to the kitchen.
• Open a waste book at the pass: spoilage, prep waste, comps and returns, valued weekly at cost. Under 3 percent of net food sales is a sensible target.
• Consolidate to two or three suppliers, then ask in writing for a 90-day fixed price list on your top 20 lines by spend.
• If you are tied, read your agreement first. Many tenancies carry nominated supplier lists or food purchasing obligations, and a better price you are not permitted to take is not a better price.
• Re-run the grid. If the puzzle did not clear its threshold, delete it.

Common objections

"I do not have time." Week 1 is twelve to sixteen hours, once. Everything after is a monthly routine measured in minutes.

"My regulars will notice if the roast goes up 80 pence." They will notice a mean plate long before they notice 80 pence. Twelve regulars talking about portion size cost you far more than a price move nobody mentions.

"What if I lose covers?" That is why you move two dishes, not the whole menu, and watch the mix for six weeks. If it holds, the cash is banked. If it drops, you have learned something real for the price of one reprint.

The bottom line

Profitable pub food is not a list of trendy dishes. It is three habits: cost everything ex-VAT from delivered prices, judge dishes on cash and chef minutes rather than percentage, and act on your own till data every quarter. Do that and the menu stops being a guess. You know which dishes to protect, reprice, promote and delete, and you can defend every call to your accountant and to the regular at the end of the bar.

Start with Week 1. If you want a second pair of eyes on your costing, menu structure and pricing, we work with licensees across the UK to turn good pubs into profitable ones.