Most restaurant statistics get quoted by people who have never checked them. The 90% first-year failure rate is the famous one, and it came from a television advert. So this is a set of 2026 numbers with the source attached and, more importantly, with what each one actually implies for a restaurant that has to open tomorrow. Where the figures are American, and most of the good ones are, they are labelled as such. The useful benchmark in the end is not the industry average anyway: it is the same number measured in your own building, which is why the reports that matter most come out of your restaurant POS rather than out of a press release.
Sales, margins, failure rates, costs since 2019, off-premise share, labour, technology adoption and what diners say they want, in that order. If you only read two sections, read the one on profitability and the one on costs, because together they explain why a year of record industry sales still felt like a bad year on the ground. Anyone reconciling these benchmarks against their own accounts will want restaurant accounting that reports on the same lines the surveys use, otherwise you are comparing your gross to somebody else's net and drawing the wrong conclusion.
How big the industry is in 2026
The National Restaurant Association's 2026 State of the Restaurant Industry report forecasts total US restaurant and foodservice sales of $1.55 trillion, a 4.8% increase on 2025. Adjusted for inflation, real growth is 1.3%.
Those two numbers next to each other are the entire year in miniature. Almost three quarters of the increase is menu pricing rather than more people eating out. If your own sales are up 4% this year, you are running flat.
The rest of the scale, for context. Just under a million restaurant and foodservice outlets operate across the US. Employment is projected to reach 15.8 million in 2026, up by more than 100,000 jobs, which makes the industry the country's second largest private sector employer behind healthcare. Restaurants take 53% of the household food dollar, a split that has been drifting away from grocery for decades. Quick service and fast casual together account for roughly 40% of all sales.
European operators should read all of that as direction rather than benchmark. The pattern of pricing-led growth, thin margins and off-premise expansion holds across most western European markets. The absolute figures do not transfer, and neither do the labour costs.
Profitability: the numbers that explain the mood
Here is the statistic that should be quoted more often than the sales forecast. 42% of operators reported that their restaurants were not profitable in 2025. Six in ten saw traffic decline.
Median margins, from the Association's 2025 Operations Data Abstract, covering 2024: 2.8% for full-service restaurants and 4% for limited-service. In 2019 the same figures were 4% and 6%.
Sit with 2.8% for a second. A full-service restaurant turning over $1.2 million a year keeps about $34,000 of it. That is a single equipment failure, one bad month, or three weeks of a road closure outside your door. It is also why the debate about whether to accept a 30% delivery commission is not really a debate: at a 2.8% margin, commission at that level has to be priced into the menu or it is a donation.
The other thing margin data hides is timing. Profitable restaurants close because cash and profit are not the same thing, and a business can be making 4% on paper while unable to pay a VAT bill because the money went out in stock before it came in through the till. Watch both.
The failure rate everybody gets wrong
The claim that 90% of restaurants fail in year one appeared in a 2003 American Express television advertisement. Researchers asked the company for the underlying data. It replied, in writing, that it could not provide any. That is the entire provenance of the most repeated statistic in the industry.
What the research shows instead. H.G. Parsa's longitudinal study of restaurants in Columbus, Ohio found first-year failure at 26%, second year 19%, third year 14%, for a three-year cumulative rate of 59%, split 57% for franchised chains and 61% for independents. A separate study by Luo and Stark, working from around 81,000 restaurants across the western US over two decades, put first-year failure at roughly 17%, which is lower than the 19% they measured for all other service businesses. Median restaurant lifespan came out at 4.5 years, slightly longer than the 4.25 years for other service startups.
Two things follow. Risk is front-loaded and falls every year you survive, so the second anniversary is worth more than it feels. And restaurants are not unusually doomed; they are a normal small business with a long list of ways to go wrong. The myth has a real cost, incidentally, because lenders price risk off perception, and operators have been paying inflated interest on the strength of an advert.
Costs: what has actually happened since 2019
This is the section that explains the profitability numbers above.
Food costs are up 38% since 2019, with the Bureau of Labor Statistics Producer Price Index for All Foods showing average food costs more than 35% above pre-pandemic levels. Labour costs are up 35% over the same period. Credit and debit card processing fees have risen 40% since 2020. Menu prices, meanwhile, are up 32%.
Read those four numbers in order and the margin compression is not a mystery, it is arithmetic. Costs rose faster than the prices operators felt able to charge, and the difference came out of the 2.8%.
The recent detail: in 2025, 82% of operators reported higher average food costs than the year before, and only 6% saw a decline. More than two thirds, 68%, said tariffs on imported goods pushed their food or beverage costs up. More than 90% of operators describe food, labour, insurance and general inflation as significant challenges, and more than 80% report real strain from card processing fees and from energy and utilities.
TouchBistro's 2026 report, based on a Maru/Matchbox survey of over 600 US full-service owners and managers, adds the pricing response. 68% ended up raising prices, up from 47% the previous year, and the average increase was 12%, slightly softer than the 14% before it. More than a quarter, 28%, named food and inventory costs as their single greatest financial strain, rising to 38% among operators in Los Angeles.
What to do with this: stop tracking your food cost percentage against a 2019 memory, and start tracking the rate of change per supplier. Operators who caught the 2025 increases early were the ones whose invoices were matched line by line in stock management, not the ones checking a monthly total.
The costs nobody puts in the business plan
Food and labour get all the attention because they are the biggest lines and the ones you can influence weekly. The quieter three are worth separating out, because more than 80% of operators report significant strain from card processing fees and from energy and utilities, and almost nobody budgets for them properly.
Card fees are the clearest example. Up 40% since 2020, charged on gross rather than on your margin, and largely invisible because they arrive as a netted-off settlement rather than an invoice you have to approve. On a restaurant taking $80,000 a month by card, a difference of half a percentage point in effective rate is $4,800 a year, which is more than most operators recover from a full round of menu engineering. Read one month's statement line by line, check the effective rate rather than the headline rate, and confirm whether your restaurant payments setup is charging you for authorisations and terminals you no longer use.
Insurance belongs in the same category, along with the standing charges on energy contracts that renewed during the 2022 spike and were never revisited. None of it is exciting. All of it is recoverable without touching a recipe or a rota.
What operators are actually doing about it
Surveys usually stop at the complaint. This part is more useful, because it shows where a large sample of operators put their effort when the costs above landed on them.
On the cost side, the leading responses were waste reduction, cited by 42%, supplier diversification at 39%, and technology or AI tools to find inefficiencies at 29%. Notice the order. The cheapest lever came first, and it is the one that needs no negotiation, no new vendor and no capital: throwing away less of what you already bought.
On the revenue side, 33% increased marketing investment, 33% ran new promotions and 30% expanded into catering and events. What is absent from that list is interesting. Nobody's headline strategy was another price rise, after two years in which prices did most of the work.
The expansion plans follow the same logic of using what you already have. 51% intend to take part in local events such as food festivals, 39% are adding private events and 38% are launching catering. All three monetise a kitchen, a team and a room that are already paid for, which is the only kind of growth available to a business clearing 2.8%.
Asked what stands in the way of growth, operators named food costs at 18%, economic uncertainty at 15%, rent at 13% and attracting new customers at 13%. And then, despite all of the above, 86% said they were optimistic about the future of their own business. Make of that what you will. My reading is that operators separate the industry's problems from their own room, which is both a survival trait and occasionally a blind spot.
Off-premise is no longer a channel, it is a third of the business
At full-service restaurants, off-premise accounted for more than 35% of customer traffic in 2025, according to Technomic. In 2019 it was 12.1%.
Nothing else in these statistics has moved that far that fast. A dining room is capped by seats and by turn times; takeaway, delivery and drive-thru are capped only by opening hours and your kitchen's throughput, which is why the operators who built for it have grown without signing a second lease.
On the consumer side: 47% of US adults pick up takeout at least once a week, and the generational split is stark, with 59% of millennials and 57% of Gen Z ordering weekly against 44% of Gen X and 33% of baby boomers. Drive-thru runs at 42% of adults weekly, led by 55% of millennials and 54% of Gen Z.
Worth doing the commission arithmetic once, properly, because the percentage never feels as bad as it is. A $40 delivery order at 30% commission leaves $28. Food at 30% of the original price takes $12, packaging another dollar, and the labour to make it is unchanged. You are left with roughly $15 to cover rent, energy, insurance and everything else, against about $27 on the same order collected at the counter. That does not mean refuse the channel. It means know which items you can afford to send out and price the menu you send accordingly.
The operational warning is that a third of your covers arriving through a different door does not run itself on the same setup. Ticket times diverge, packaging becomes a cost line, and a dining room ticket and a delivery ticket firing at the same second need sequencing that a paper docket cannot give you. This is squarely a takeaway POS problem before it is a marketing one.
Labour: the constraint that is structural, not cyclical
Employment reaches a projected 15.8 million in 2026. Nearly three quarters of operators plan to hire, and most expect real difficulty finding experienced managers and chefs.
The number that deserves more attention is the slowdown. Restaurant payrolls grew by 181,000 in 2025. In 2024 the figure was 1.46 million, and in 2023 it was 2.52 million. The Association flags that deceleration as the biggest single risk to the industry and to the wider economy, which is a striking thing for a trade body to lead with.
Underneath it sits demographics you cannot recruit your way out of: the 16 to 24 year old population is shrinking. There is a partial offset, in that if participation rates among younger workers merely hold flat rather than continuing to fall, the labour force could contain roughly 400,000 more teenagers and 500,000 more young adults than current BLS projections assume. Hardly a rescue.
One more set of figures worth knowing, because it is the industry's best argument for itself. 67% of US adults have worked in restaurants or foodservice at some point, more than any other sector: 78% of Gen Z, 74% of millennials, 69% of Gen X and 52% of baby boomers. More than half of all adults had their first regular job in foodservice, rising to 67% of Gen Z.
Practical reading: with the entry-level pool contracting, retention is cheaper than recruitment and the gap is widening every year. A shift you can fill from your existing team is worth more than a job ad, and the maths on that already favoured keeping people before the pool started shrinking.
Technology adoption in 2026
Operators were asked where technology has actually made a difference. Top of the list for both segments is on-premises ordering and payment: 67% of limited-service and 58% of full-service operators.
Artificial intelligence sits at 26% of operators using AI-related tools, with marketing the leading use at 19% of full-service and 15% of limited-service operators, and only 6% using AI for customer orders. Worth noting that TouchBistro's survey of full-service operators returned 87% using "some form of AI", a gap that comes from how the question was framed rather than from a real disagreement, and one worth understanding before you quote either figure. There is more on that in our piece on AI in restaurants.
Plain automation, as opposed to AI, is where full-service operators report the strongest uptake: online ordering at 68%, up from 57% a year earlier, payroll at 54%, invoicing at 52%, kitchen order routing at 52% and email marketing at 51%. Their stated reason is speed, with 52% citing faster service, up from 37% the year before.
Guest appetite has clear edges to it. A solid majority of consumers are comfortable ordering and paying by smartphone, website, kiosk or tablet at the table. 62% would order by speaking to a live person on a video screen, but only 39% would do so with an AI-generated persona. Nearly half would use an AI chatbot on a website or app, rising to six in ten among millennials and Gen Z.
And overall sentiment stays genuinely split: 41% of consumers say technology improves hospitality, 38% say it detracts, 21% say it makes no difference. Just over half of Gen Z adults and millennials say it enhances the experience, against 23% of baby boomers. Which is the real lesson of every technology statistic here. The average tells you nothing; your guest mix tells you what to install, and a mobile ordering rollout that delights one room can irritate another.
What diners say they want
61% of adults describe restaurants as an essential part of their lifestyle, holding up across generations and dipping only to 56% among the oldest group. Even consumers who report struggling with monthly expenses mostly keep dining out, ordering takeaway or using delivery in some form.
A solid majority also say membership of a loyalty or rewards programme influences where they choose to eat, on-premise and off. Value offers do the same work.
The backdrop for all of it is an economy the Association expects to keep growing, with real GDP up 2.7% in 2026 against 2.3% in 2025, and about three quarters of consumers expecting their own finances to improve or hold steady over the coming year. Consumer confidence at that level is why a soft-traffic year still produced record sales.
Spending itself remains split along income lines, the pattern the Association keeps calling a K-shaped economy: higher-income households spending comfortably, lower and middle-income diners becoming more selective as prices bite. For an operator that means the same menu is being read two ways, and the answer is rarely a blanket discount. It is a clear entry price, one or two visible value items, and a reason to come back that is not purely money, which is what a working loyalty program buys you.
How to use these numbers without fooling yourself
Five habits keep benchmarks honest.
Check whether you are reading a median or a mean. Restaurant averages are dragged around by outliers, and a median margin of 2.8% describes the middle restaurant, not a typical one, since half of them are below it.
Match the segment. Full-service at 2.8% and limited-service at 4% are different businesses with different rent-to-revenue ratios. Comparing a 60-seat bistro to a national quick-service average produces conclusions that feel rigorous and are not.
Separate opinion from record. Almost every figure on this page from an operator survey is what operators say. Cost indices and payroll data are measurements. Both are useful; only one is evidence about your own building.
Watch the base year. Numbers quoted against 2019 look dramatic and are, but a supplier who raised prices 8% last year and 2% this year is not the same problem as one who did the reverse, even if the cumulative figure matches.
Then, and this is the part that pays: build the five numbers you will actually track. Prime cost as a percentage of sales, covers, average spend per cover, off-premise share of revenue, and labour hours per cover. Pull them weekly, write them in the same place, and the industry statistics become context instead of a scoreboard you cannot influence. Our guide to restaurant KPIs covers how to get each one out of a normal week's trading.
Start with three lines this week. Your margin next to 2.8%. Your off-premise share next to 35%. Your food cost change since 2019 next to 38%. Wherever your number is worse, you have found this quarter's project, and wherever it is better, you have found something worth protecting.
Read next: restaurant profit margins, the KPIs worth tracking, and AI in restaurants.
