A restaurant is a manufacturing business with a shopfront lease and a same-day expiry date on everything it makes. The failure is almost always arithmetic, and the arithmetic is visible before you sign.
The short answer: most US restaurants fail on three numbers, not on cooking: prime cost above 65% of sales, rent above 10% of sales, and too few covers per service to cover fixed costs. Commonly cited labour data puts first-year closures around one in five, and roughly half within five years. All three of those numbers can be modelled before opening.
The three numbers that decide it
1. Prime cost
Food plus hourly labour, as a percentage of sales. Below 60% is comfortable, 65% is the ceiling most operators work to, and above 70% the business is funding customers.
Why it drifts: portion creep, waste, staff meals nobody counts, and a delivery mix that adds packaging and commission without raising the menu price.
2. Rent
Rent should stay under 10% of sales, and under 8% if you are new. The trap is signing a lease based on projected sales that assume a full house every night in month two.
The check is simple and takes ten minutes: take the rent, divide by your average ticket, and see how many covers a month you must serve before rent alone is covered. If that number already looks like a good Friday every night, the lease is the problem.
3. Covers per service
Seats multiplied by turns. A 40-seat room doing two turns a night serves 80 covers. At a $28 average ticket that is $2,240 a night, about $67,000 a month before you subtract anything.
Most new operators overestimate turns. Two turns on a Tuesday is an achievement, not a baseline.
A worked break-even
Fixed monthly costs for a small sit-down restaurant: rent $6,000, utilities $1,400, insurance and licenses $700, software and card fees $600, salaried management $5,000. That is $13,700 before a single plate.
At an average ticket of $28 and a 65% prime cost, each cover leaves $9.80 to put against fixed costs. So break-even is about 1,400 covers a month, or roughly 47 covers a day, every day, including the quiet ones.
That is the number to carry into the lease negotiation. Not "will people like the food", but "can this room do 47 covers a day in February".
Where the money quietly leaves
- Delivery commissions. A third-party order at a 30% commission can turn a 65% prime cost into a loss on that plate. Price delivery as a separate menu, or do not offer it.
- Menu size. Every extra dish adds prep, waste and a slower ticket time. Small menus make money.
- Staffing to the schedule, not to the sales. Two people standing in an empty room is a decision, not a necessity.
- Discounting instead of filling. A 20% discount needs 25% more volume to stand still.
What the survivors do differently
- They know their break-even covers number and check it weekly, not yearly.
- They keep the menu short and the prep predictable.
- They negotiate rent with a step-up instead of a fixed high number from day one.
- They keep three months of fixed costs in reserve before opening, not after.
Lease traps that end restaurants
Three clauses do the damage. A personal guarantee turns a failed business into personal debt. A fixed escalation raises rent on a schedule that has nothing to do with your sales. A long term with no exit locks you into a space that the neighbourhood may outgrow.
Ask for a percentage-rent structure in the first year, a tenant improvement allowance, and a break clause at eighteen months. Landlords say no to all three less often than new owners expect.
Menu engineering, without the jargon
Every dish has a cost and a popularity. The ones that sell well and cost little are the ones to push: position on the page, describe better, price slightly up. The ones that sell well and cost a lot need a price rise or a smaller portion, not removal, because they may be the reason people come.
Dishes that sell badly and cost a lot leave the menu. Dishes that sell badly but cost little stay as variety, if the prep is shared. Four columns, ten minutes a quarter, and the food cost moves two points.
Labour is a schedule, not a number
Labour cost is decided on Tuesday afternoon when the rota goes up, not on Friday when the numbers arrive. Staff to covers, not to habit: forecast the day, set hours against it, and cut an hour before service if the bookings are not there.
The single most expensive habit is keeping a full team through a dead service because it feels fair. Fair is paying people for work that exists.
The delivery maths most owners skip
A $30 order through a third-party app with a 30% commission returns $21 before food cost. If food cost is 32%, that is $9.60, leaving $11.40 for labour, packaging, rent and profit on an order that still occupied your kitchen at your busiest hour.
Two fixes work: a separate delivery menu with fewer, higher-margin items, or your own ordering link promoted at the table. Both are unglamorous and both protect the margin.
What a turnaround looks like
When a restaurant is losing money, the order is fixed: prime cost first, rent second, volume last. Cut the menu to the twenty dishes that sell. Re-price the losers. Renegotiate or sublet space. Only then spend on marketing, because marketing a business with broken margins just buys a faster loss.
The operators who survive a bad year are the ones who read the weekly numbers and act in week three, not the ones who wait for a quarter to confirm what the Friday nights already told them.
Opening month is not the test
New restaurants are busy because they are new. Friends come, the neighbourhood tries it, the local paper writes a line. The business you actually have appears in month four, when the curiosity has been satisfied and only the regulars remain.
That is why a soft opening beats a launch. Two weeks of quiet service teaches the kitchen the menu, the front of house the flow, and the owner the real ticket time, without a full room watching it go wrong.
Cash flow has a calendar
Suppliers want payment weekly, staff want it fortnightly, card processors settle in days, and guests pay at the end of the meal. The gap between paying for food and receiving the money for it is real, and it is why a profitable restaurant can still run out of cash.
Keep a simple weekly cash calendar for eight weeks ahead. If a payroll week and a supplier week land together and the card settlements are thin, you need to know that on Monday, not on Thursday night.
The numbers to look at weekly
- Sales per labour hour, by day part
- Food cost as a percentage, by category
- Covers and average ticket, separately
- Delivery share of sales, and its margin after commission
- Cash position, eight weeks forward
Five lines. They tell you whether to change a price, a portion, a schedule or a channel, and they do it while the change is still cheap.
Why good food is not enough
Cooking is the part most owners love and the part customers judge last. They notice the wait, the noise, the bill and whether anyone looked after them. A restaurant with average food and excellent operations outlives one with exceptional food and a chaotic room, and the closures data reflects that every year.
None of this means quality does not matter. It means quality is the entry ticket, and the arithmetic decides whether you get to keep playing.
What to do before signing anything
Stand outside the space at the times you plan to trade. Count the people walking past, at lunch and at dinner, on a weekday and a weekend. Multiply by a conservative capture rate and compare it with the covers you need.
Most failed restaurants chose a location on rent and hope. An hour of counting on two evenings is the cheapest research available, and it is the one that would have saved a large share of them.
Verdict: restaurants are not a talent business first, they are a margin business with a talent component. If the lease, the prime cost and the covers do not work on paper, no kitchen will fix it. Run the 47-covers arithmetic on any space before you sign, and walk away from the ones that only work in a good month.
Questions people ask
Is a food truck safer than a restaurant?
Cheaper to test, not automatically safer. You trade rent for permits, commissary fees and the uncertainty of a location you do not control.
How much cash should a new restaurant hold?
Three months of fixed costs at minimum, plus the fit-out overruns, which are almost guaranteed. Opening with no reserve means the first slow month becomes the last one.
Related: what $10,000 actually buys in four business models.