Why Do Restaurants Fail? The Real Reasons and the Early Warning Signs

Restaurants rarely fail for the reason the owner gives afterwards. “The market was tough” and “the landlord raised the rent” are usually the final event, not the cause. The cause was typically set months or years earlier, in a lease signed at the wrong rent, a menu priced from competitors instead of from cost, or a capital plan that funded the build and nothing else.

This post is about the diagnosis: what the real closure rates are, which causes the evidence actually supports, and (the useful part) which warning signs appear while there is still time to act on them.

The Bottom Line

• The “90% of restaurants fail in year one” figure is a myth with no traceable source behind it. A University of California, Berkeley study found about 17% of full-service restaurants closed within their first year.
• The National Restaurant Association puts the industry’s average failure rate nearer 30%, and public summaries put five-year survival at roughly half.
• The pressure is real even for survivors: 42% of operators reported their restaurant was not profitable and 60% reported softer customer traffic, with more than 9 in 10 citing food, labour, insurance, energy and swipe fees as significant challenges (National Restaurant Association).
• Most closures trace to a handful of causes: undercapitalisation, location economics, pricing that does not cover cost, weak differentiation, inconsistent execution, loose inventory and labour control, and an owner who never built systems.
• Almost all of them announce themselves early. Falling covers at a steady average check, a prime cost drifting above 65%, and a cash balance that shrinks in a good month are the three to watch.

What percentage of restaurants actually fail?

Less than you have been told, and the real numbers are worth knowing because the mythical one is demoralising for no reason.

Figures compiled in MenuTiger’s restaurant failure rate statistics.

So roughly four in five full-service restaurants make it through year one, and about half are still trading at five years. That is a demanding business, not a lottery. And the distinction matters practically: a 90% failure rate implies the outcome is random and preparation is pointless, while a 17-30% rate implies the opposite; that the difference between surviving and not is mostly made up of decisions.

Where the 90% failure rate myth came from

It appears to have originated in an American Express television commercial in 2003, and there is no study, dataset or survey anybody has been able to point to behind it: a provenance traced in MenuTiger’s review of the failure rate data. It has been repeated ever since because it is a memorable number and because it flatters survivors.

It is worth actively discarding, because believing it leads to two bad behaviours: treating failure as bad luck rather than something with identifiable causes, and using it to justify not planning properly on the grounds that the odds are hopeless anyway.

The seven causes behind most restaurant closures

1. Undercapitalisation

The most common single cause, and the most preventable. The pattern is always the same: the owner raises enough to build the restaurant and open it, and nothing to operate it. Then rent, payroll and supplier invoices arrive for eight or ten weeks before revenue stabilises, and a business that would have been fine at month six dies at month three.

Three to six months of full operating expenses, raised as a separate and protected pot, is what prevents it. The temptation to spend that on a better fit-out is exactly the temptation to resist, and the discipline for this belongs in the cash flow forecast of a restaurant business plan written before the lease is signed rather than after.

2. The location economics never worked

Not “a bad location” in the abstract: a location whose rent was always too high for the covers the space could physically turn at the check average the neighbourhood supports.

This is arithmetic, and it is knowable before signing, and there is a repeatable method for working through the site itself (traffic, access, catchment, competition) set out in this guide to choosing a restaurant location. Take the front-of-house square footage, divide by square feet per seat for your format, apply a realistic turn rate, multiply by your target check, and compare the result to total occupancy cost including common area maintenance, taxes and insurance. If the rent needs more covers than the room holds, no amount of good cooking fixes it. The other half of the location question is demand: traffic counted at your actual trading hours, the surrounding income profile, access and parking, and why the previous tenant of that unit left.

3. Menu pricing that does not cover cost

Many restaurants price by looking at what the place down the road charges, which embeds a competitor’s cost structure into your business. Then costs move and the prices do not.

Food and beverage costs rose 21.8% and labour costs 18.3% over a recent three-year period. A menu last costed eighteen months ago is losing margin on every single cover sold, invisibly, and the owner experiences it as “we’re busy but there’s no money”: which is the most common sentence in a failing restaurant.

The fix is unglamorous: cost every dish to the gram including trim and garnish, price from a target food cost percentage, and re-cost quarterly against actual invoices.

4. No real reason to choose it

A restaurant that is merely competent in a market with options does not get chosen often enough. “Good food and friendly service” is what every competitor also claims, so it cannot be a differentiator.

This is a concept problem rather than an operations problem, which is why it is so hard to fix after opening. It has to be settled before the lease: a specific answer to who eats here, how often, and what they would otherwise have done. Format is part of that answer too, since the staffing model, check average and space requirements all change with it; working through the types of restaurants and what each demands operationally is a useful way to check the concept you have chosen is the one your market and your capital actually support.

5. Inconsistent execution

The restaurant that is excellent on Thursday and poor on Saturday loses customers faster than one that is reliably average, because inconsistency breaks the reason to return.

Inconsistency is nearly always a systems problem rather than a staff problem: no recipe cards, no prep lists, no opening and closing checklists, no documented service standards, and training that happened once. It also tends to show up in the first fortnight of trading, which is the argument for a soft opening; running real service at reduced capacity surfaces the dishes that collapse at volume and the stations that break under load while the audience is still small and the fixes are cheap.

A related and underrated version is physical: a layout that forces servers to take ten extra steps per table, or a dish pit that backs up into the dining room at peak. That is permanent labour cost and permanent service drag for the length of the lease, and the restaurant floor plan is where it is either prevented or locked in.

6. Loose inventory and labour control

The two largest controllable costs are food and labour, and in a struggling restaurant both are usually leaking rather than overspent by decision. Industry estimates attribute around 20% of sales lost to a mix of theft, spillage and unrecorded expenses where inventory and staff management are weak.

What it looks like in practice: no weekly inventory count, so food cost percentage is a guess until the accountant produces it months later. No par levels, so over-ordering and spoilage. Schedules built from habit rather than from forecast covers, so labour is paid for hours with no sales in them. Comps and voids nobody reviews.

Most of this is a reporting problem before it is a discipline problem, and it is usually fixable with the system you already own. A properly configured till gives you hourly sales, item-level mix, comps and void reporting and labour against sales without anyone compiling a spreadsheet, which is why the choice of POS is worth more scrutiny than it normally gets: a comparison of the best POS systems for restaurants is a reasonable place to start if yours cannot produce those numbers on demand.

Prime cost (food plus labour as a percentage of sales) is the number to manage weekly. Most full-service restaurants need it at or below about 60-65% to be viable. Above that, the business is consuming itself and the monthly accounts will confirm it too late to act.

7. The owner never built a business

A restaurant that only works when the owner is in the building is not an asset, it is a job with unlimited liability. It cannot survive illness, it cannot be sold for much, and it cannot grow.

The symptoms are familiar: no manager with real authority, no documented processes, every decision routed through one person, and an owner working so many hours on the line that nobody is working on cost control, pricing or marketing. The restaurants that last build systems in year one; documented recipes, trained managers, weekly numbers reviewed on a schedule. Those systems are also the precondition for anything beyond one site; a business that depends on the founder’s presence is not one you can replicate, which is why most attempts to franchise a restaurant fail at the documentation stage rather than the demand stage.

The early warning signs

Most of the causes above are visible in the numbers long before they are visible in the bank balance. These are the indicators that something is wrong while it is still fixable.

Financial signals:

• Prime cost drifting upward, week over week. A two-point move sustained over a month is a real problem, not noise.
• Cash falling in a month that looked good. The clearest sign that profit and cash have separated, usually through supplier terms or debt service.
• Food cost percentage you cannot explain. If you do not know why it moved, you are not controlling it.
• Rising comps, voids and discounts as a share of sales, which often means service problems being papered over at the till.
• Payables stretching. Paying suppliers later each month is the last stage before the end, and it is often the first thing an owner normalises.

Demand signals:

• Covers falling while average check holds. Fewer people, spending the same. This is a demand problem, and it is the single most reliable early indicator.
• Repeat visits thinning out. Harder to see without data, which is an argument for capturing it, but a drop in recognisable regulars precedes a drop in revenue.
• Reviews shifting from the food to the service and the wait. Almost always a capacity or consistency problem showing up before it hits the numbers.
• Falling RevPASH while sales look flat. Revenue per available seat hour catches the room emptying out at the edges of service (the hours you are paying rent and staff for and not selling) which daily sales totals hide completely. It is the metric that tells you a problem is structural rather than seasonal.

Operational signals:

• Staff turnover accelerating, particularly among people who were good. Staff see the decline before customers do.
• Ticket times stretching without a volume increase to explain them.
• Deferred maintenance. Equipment that is not repaired because the cash is not there is both a symptom and a future cost.

Any one of these can be a bad fortnight. Two or three together, sustained for a month, is a business telling you something. The owners who recover are the ones watching weekly rather than waiting for the quarterly accounts, because almost every item on this list is recoverable if caught early and almost none of it is recoverable once payables have been stretched for two quarters.

What this means if you have not opened yet

Nearly every cause on this list is decided before service begins. The capital plan, the lease arithmetic, the menu costings, the concept’s reason to exist, the layout, the systems: all of them are set in the planning stage, and all of them are enormously cheaper to get right than to fix. That is the real argument for doing the planning sequence properly rather than quickly, and how to start a restaurant covers that sequence in the order the decisions actually have to be made.

Frequently asked questions

What percentage of restaurants fail in the first year?

About 17% of full-service restaurants, according to a University of California, Berkeley study. The National Restaurant Association puts the industry-wide average failure rate closer to 30%. The widely repeated 90% figure has no identifiable source and should be disregarded.

What is the number one reason restaurants fail?

Undercapitalisation: opening without enough working capital to trade through the period before revenue stabilises. It is also the cause most fully within the owner’s control before opening, since it is a planning decision rather than a market condition.

Why do restaurants fail even when they are busy?

Because busy is not the same as profitable. A restaurant can run full every night and still lose money if the menu is priced below its true cost, prime cost is running above 65%, or occupancy cost is too high for the covers the room can turn. “Busy but no money” nearly always points to pricing or cost control rather than demand.

How long does it take to know if a restaurant will survive?

The first ninety days reveal whether the core assumptions were right: check average, covers, food and labour cost percentages. The first full year reveals whether the business can absorb a slow season. Most of the outcome, though, was determined before opening by the lease, the capital plan and the concept.

Can a failing restaurant be turned around?

Often yes, if it is caught before payables are badly stretched. The usual sequence is to re-cost and reprice the menu, cut the item count, rebuild the schedule from forecast covers, renegotiate supplier terms, and address the one or two service problems the reviews keep naming. Turnarounds that fail usually do so because the underlying problem was the lease, which no operational fix can reach.