How to reduce restaurant costs.

Food and labour, because together they're the majority of what you spend and they're the two you can actually move this month. Add commission if you're doing any real delivery volume, because that one is pure percentage off the top.

Put them all on the same footing: as a percentage of sales, measured weekly. Rent and insurance are annoying but fixed, so they're a once-a-year conversation. Food and labour move every single week, which means they can drift for a quarter before anybody notices, and by then it's real money.

The rule of thumb most operators work to is food and labour combined staying under roughly two thirds of sales, with the split between them depending on the kind of restaurant you run. A wet-led place carries a different shape to a chip shop. Work out your own normal from your own numbers over three good months, then treat any week outside that range as a question to answer rather than a bad week to forget about.

More than any other line you can change quickly. Commission commonly runs 25 to 30% of the order value, so a €30 order gives up €7.50 to €9 before you've bought anything.

Do the sum on statements rather than memory. Ninety days, all deductions in, divided by the order count, and you'll have your true cost per order. Then split those customers in two: the ones who found you on the app, and the ones who already knew you and are ordering there out of habit. The first group is what the commission is for. The second is money you're giving away to serve your own regulars.

Moving that second group to your own ordering channel is the single largest cost saving available to most independents, and it doesn't touch the food. The detail is in direct ordering versus third-party apps. Eclyde charges €59 a month plus 5% per order, or €199 a month with no commission at all, €169 if you pay for the year, which is the sum worth running against your own delivery statements.

By finding out what you're actually binning, which almost nobody knows, and which takes one week to learn.

Put a bin liner and a notepad by the pass and log everything thrown out for seven days: what it was, roughly how much, and why. Spoiled, over-prepped, wrong order, returned, staff food. That list tends to surprise people. Most kitchens assume waste is spoilage and find out it's over-prepping, which is a forecasting problem, or remakes, which is an order accuracy problem, and both are cheaper to fix than they are to keep paying for.

Then tighten the boring things. Prep sheets built from what actually sold on the same weekday, not from a feeling. Scoops and a scale for the three items with the worst margin, since eyeballed portions drift upwards all year. Proper stock rotation and date labels so nothing dies at the back of the fridge. Check deliveries against the docket and weigh anything priced by the kilo, because short deliveries are more common than suppliers like to admit. And use the trim: bones, stalks and offcuts are already paid for.

By matching the rota to the shape of the day, rather than to a habit set two summers ago.

Pull sales by hour for a typical week and look at where the money actually is. Most restaurants find their rush is narrower than the rota assumes, with an expensive hour at each end where three people are on for four covers. Stagger the starts and finishes in half hours instead of putting everybody on at five. Watch sales per labour hour rather than headcount, because that's the number that tells you whether an extra body earned their shift.

Then the leaks. Clock-in on the till rather than a sheet, because paper timesheets are always generous. Overtime should need approval before it happens rather than a shrug afterwards. Cross-train enough that a quiet section can help a busy one instead of standing there. Break scheduling that doesn't land in the middle of the rush.

One honest warning. Understaffing a Friday is not a saving. You'll pay for it in ticket times, refunds, reviews and eventually in the people who leave because every shift is a fight. Cut the quiet hours, protect the busy ones.

Yes, and most independents never try, which is why their price list has quietly drifted for two years.

Start by checking the invoices against the delivery dockets for a month. Price creep on individual lines between orders, without any notice, is common, and it's not always deliberate. Then get two other quotes on your ten biggest lines by spend, and take those numbers back to your existing rep. Being an easy customer who pays on time is worth something, so ask for it.

Other levers worth pulling. Fewer, larger deliveries usually get a better rate than daily drops. Buying whole and breaking it down yourself is cheaper if you have the skill and the time. Agreeing a price hold for a quarter on your core lines gives you a costed menu that stays true. And ask what's coming into season, because a rep who likes you will tell you where the value is next month.

Small individually, boring collectively, and mostly fixable in an afternoon.

Kitchens leak energy in predictable places. Fridge and freezer door seals that don't seal any more, and condensers behind them furred with dust. Extraction running full for hours before anybody cooks. Gas rings and grills lit at four for a service that starts at six. A closing routine that nobody wrote down, so what gets switched off depends on who locked up. Write the closing order on the wall and it stops being a debate.

Then the paper bills. Read the standing charges and the unit rates on your energy contract and check whether you're out of a fixed term. Look at what you're renting rather than owning, particularly card terminals, which are often on a rolling agreement nobody has revisited. Check the service contracts that renew automatically. And put your insurance out to the market rather than accepting the renewal quote, which is nearly always higher than what a broker can find.

List them. Actually write them down, from the bank statement rather than from memory, and total the monthly figure.

A typical independent ends up with a till, an online ordering provider, something for email, an accounts package, a rota app, a reviews tool, and a website builder, plus per-till or per-screen fees on top of the ones that charge that way. None of them is expensive on its own. Together they're often a few hundred euro a month for tools that don't talk to each other, which means somebody is also exporting and re-importing data by hand.

Two savings here. The obvious one is the tools nobody has opened since the trial. The bigger one is consolidation: fewer systems means fewer bills, no integration work, and one set of numbers instead of four that disagree.

Usually two or three, and they're often the ones the kitchen is proudest of.

Cost every dish properly. Weigh the portions as they actually go out, not as the recipe says, price the ingredients from a recent invoice, and include the packaging, the sauce pot, the garnish and the trim waste. Then look at cash margin per plate rather than food cost percentage on its own, because a dish at 35% that leaves €9 on the plate is worth more to you than one at 25% that leaves €4.

Do that once and the decisions make themselves: reprice a few, re-portion a few, promote the ones that earn, and take the low-margin, low-selling dishes off entirely. Every dish you cut also gives you back prep time, stock lines and space on the menu for something better.

The things a customer can feel, and the things that get more expensive if you defer them.

Don't shrink a portion that people order specifically. Don't swap the ingredient your signature dish is known for. Don't cut the hours of the closer who runs a clean, calm shift, because you'll pay for that twice over in the next three months. Don't skip cleaning or maintenance: a service on the extraction or the fridges costs a fraction of the breakdown, and an environmental health visit is not the moment to find out.

Cost cutting that reaches the plate isn't cost cutting. It's the start of a slow decline that shows up in reviews about six weeks later and in takings about six weeks after that.

With a short weekly habit, on the same morning, whether or not the week was busy.

Four numbers, half an hour. Food cost as a percentage of sales. Labour as a percentage of sales. Share of orders coming through your own channels rather than the marketplaces. And waste for the week, by reason. Write them somewhere you'll see them again, and compare with the same week last month.

Costs don't jump, they drift. A percentage point here, a bigger portion there, an extra half hour on the rota. The habit is what catches drift while it's still a conversation instead of a crisis.