The Feasibility Study Consultant's Role in Restaurant Feasibility Studies

Two numbers govern restaurant lending and both are usually stated wrongly. The first is that ninety per cent of restaurants fail in their first year, which is false — it originated in a 2003 American Express commercial and has no data behind it. The second is that restaurants are no riskier than any other small business, which is also false. Restaurants default at roughly two to three times the SBA portfolio average. A feasibility study on a restaurant deal exists to reconcile those two facts: the concept is not doomed, and the credit is genuinely difficult.

What changed in June 2025

Any restaurant feasibility study written now is written against a materially tighter lending standard than one written two years ago, and a consultant who does not reflect that is preparing a document for a regime that no longer exists.

SOP 50 10 8 took effect on 1 June 2025 and replaced a principles-based regime with a prescriptive one. The changes that bear directly on restaurant credits:

A mandatory 10% equity injection for all startups operating one year or less and for all complete changes of ownership, calculated against total project costs. ESOPs are exempt.

Seller standby notes count toward that injection only on full standby — zero principal and interest for the entire life of the loan, typically ten years — and can satisfy no more than half the required injection.

A 1.10x debt service coverage floor for 7(a) Small Loans at or below $350,000.

Collateral requirements above $50,000, which means a consultant must now address liquidation value alongside going-concern cash flow. For restaurant equipment, those are very different numbers.

Tax transcript verification restored.

The policy reversal was a response to portfolio performance rather than a philosophical shift. As Phillips Lytle LLP summarised it, "For fiscal year 2024, the SBA 7a program saw negative cash flow… the first year of negative cash flow for the program in over a decade," which SBA officials attributed to "the substantial rise in defaults from underqualified buyers."

The effect on volume was immediate. The programme had originated a record $37.3 billion in FY2025. iBusiness Funding measured a 17% to 22% quarter-over-quarter decline in Q3 FY2025; broader industry estimates put the June to August 2025 decline in the range of 40% to 50%.

One September 2025 amendment is worth knowing. Procedural Notice 5000-872764 broadened the expansion-acquisition exception — same six-digit NAICS code, identical ownership — by removing the prior requirement that the acquisition be in the same geographic area. Some multi-unit restaurant acquisitions can now avoid the 10% injection entirely.

Restaurants are the biggest category in SBA lending and the worst-performing

Full-service restaurants under NAICS 722511 are consistently the single largest category by SBA 7(a) approval dollars.

On a cumulative basis across the SBA's disclosed loan records from 1992 to 2025, as compiled by PeerSense and updated March 2026:

Full-service (722511) — 41,841 approved loans, $20.2 billion, average $483,000. That average sits 42% above the $340,000 national SBA average. Average term 148 months. Roughly 11% ran through the 504 programme where real estate was included.

Limited-service (722513) — 28,029 loans, $12.1 billion, average $431,000, 27% above the national average. Average term 138 months, roughly 8% via 504.

Single-year trend has been strong. iBusiness Funding's analysis of SBA disclosure data showed full-service restaurant approvals up 24.7% year over year in Q1 FY2025, from 729 loans to 909, and $355.6 million to $401.7 million. Limited-service grew 40.5% by count and 62.7% by dollar volume, from 450 loans to 632 and $176.4 million to $286.9 million.

And the loss experience is the worst of any sector. Accommodation and food services is repeatedly identified as the highest charge-off industry group. The magnitude depends on the measure:

On a resolved-loan basis, the sector runs roughly 12% to 15% in normal conditions per Crestmont Capital's 2026 analysis. PeerSense's July 2026 benchmark puts the broader limited-service restaurant group at 18.9% against an all-industry average near 15.8%.

On an annual cohort basis, restaurants run approximately 6% to 9% against 3% to 5% for the broader portfolio.

The methodologies disagree on the number. They agree entirely on the direction.

This is the fact that justifies the engagement. A lender asking for an independent feasibility study on a restaurant is not being difficult; they are responding to the loss data in front of them.

The failure statistic, corrected

Because it comes up in every restaurant conversation and because getting it right is part of the consultant's credibility, it is worth stating what the research actually shows.

The "90% fail in year one" figure has no source. It entered circulation through a 2003 American Express commercial.

The most rigorous study says the opposite. Luo and Stark's 2014 UC Berkeley analysis of 81,000 western US restaurants found that only 17% of independently owned full-service restaurant startups failed in their first year, compared with 19% for all other service-providing startups. Median restaurant lifespan was about 4.5 years.

The Ohio State study is harsher but still nowhere near 90%. Parsa's 2005 research in Columbus, Ohio found 26% first-year failure and roughly 59% to 60% cumulative failure at three years.

Recent segment data is lower still. Datassential reported first-year failure rates in 2025 of just over 1% for quick service and casual dining, 0.6% for midscale and 0.5% for fast casual, against a seven-year low in overall closures.

The distinction that matters analytically is between first-year failure and cumulative failure. Year one is survivable for most. Cumulative closure rises toward roughly half by year five. A projection that models five years of operation is modelling a period in which a substantial minority of comparable businesses will not survive, and the sensitivity analysis should reflect that rather than treating year five as a straight-line extension of year one.

The 2026 operating environment

The National Restaurant Association's 2026 State of the Industry report describes a market growing on price rather than on volume, with margins under sustained pressure.

Sales. $1.55 trillion projected for 2026, a 4.8% increase on 2025. But Chief Economist Chad Moutray puts real sales growth at 1.3%. Nearly $1.2 trillion of that sits in eating and drinking places, and restaurants now command 53% of the household food dollar. The industry expects to add roughly 100,000 jobs to reach 15.8 million.

Profitability. 42% of operators reported their businesses were not profitable in 2025. Only 42% reported that they were.

Costs have reset permanently. Per the NRA, average food costs are more than 35% above pre-pandemic levels, average hourly restaurant earnings have jumped 41% from pre-pandemic levels, and total average restaurant expenses rose 36% between 2019 and 2026.

Traffic is flat to negative. Circana reported US foodservice traffic down 0.3% year over year in 2025, with Q4 2025 average spend per visit up 3% — confirming that revenue growth is price-driven. Circana projects less than 1% traffic growth in 2026. As advisor David Portalatin put it, "Growth in foodservice is about winning the battle for market share."

Only about a third of brands tracked by Black Box Intelligence posted positive comparable sales in 2025.

Breakfast has been the weak daypart. Revenue Management Solutions measured quick-service breakfast traffic down 8.7% in Q2 2025, recovering to negative 3.3% by September. McDonald's chief executive Chris Kempczinski called breakfast "absolutely the weakest daypart" in August 2025.

The menu-versus-grocery gap has begun to close. For 2025, the Bureau of Labor Statistics reported food away from home up 4.1% December to December against food at home up 2.4% — a gap that pushed some consumers back to cooking. By June 2026 menu prices were up 3.4% year over year, the slowest in seventeen months, and grocery prices had outpaced menu growth year to date. That is a genuine tailwind and it is recent enough that many projections have not caught up with it.

For the consultant, the practical consequence is that a projection built on pre-2020 cost ratios is not conservative. It is wrong. A lender's analyst checks prime cost first, and a model showing food at 28% and labour at 28% in a market where comparable operators run materially higher invites the question the sponsor least wants asked.

The demand base has moved off-premise

This is the structural shift that most changes how a restaurant trade area should be analysed, and it is larger than most sponsors realise.

Nearly 75% of all restaurant traffic now happens off-premises, per the National Restaurant Association's 2025 Off-Premises Restaurant Trends report.

The composition of that shift since 2019: delivery rose from 4% to 9% of traffic, drive-thru from 41% to 43%, with takeout remaining the most-used method. Off-premise now represents a larger share of sales than in 2019 for 58% of limited-service operators and 41% of full-service operators.

Three analytical consequences follow.

The trade area is drawn differently. A dining room draws from where people are willing to travel and sit. A drive-thru draws from a traffic stream. A delivery radius is set by platform geography and driver economics. These are three different maps and a single drive-time ring describes none of them properly.

Physical capacity stops being the constraint on revenue. Seats and turns cap on-premise volume. Kitchen throughput, pickup staging and drive-thru queue capacity cap off-premise volume. A concept whose kitchen cannot produce beyond its dining room's seat count has capped itself at a quarter of the addressable demand.

Delivery revenue is not equivalent to dine-in revenue and must be modelled separately. More on that below.

What the consultant actually does

Defines the trade area by daypart and by channel

Restaurant trade areas are drawn by drive time, and the drive time differs by segment and by daypart. Quick service and fast casual draw from tight convenience radii and depend heavily on daytime population and pass-by traffic. Full-service and destination concepts draw wider. Breakfast and lunch follow where people work; dinner follows where they live.

The method combines Census and American Community Survey demographic bands, drive-time analysis, state Department of Transportation traffic counts, and a physical competitive census — walked or driven rather than pulled from a listing service.

Traffic counts are the most misused input in restaurant feasibility. A high count on an adjacent road is not demand unless the site is accessible from it, visible early enough to react, and on the correct side for the relevant daypart. Access, egress, signalisation and turning movements belong in the demand analysis, not in the site description.

Sizes demand from spending rather than population

The defensible chain runs: households within the drive-time bands, household income, food-away-from-home spending per household, the share addressable at this concept's price point and format, then competitive capture.

Each link needs external evidence. Spending comes from consumer expenditure data adjusted for local income. Addressable share depends on whether the average check fits the market's capacity to pay.

Price point discipline is where studies most often fail. A $34 average check in a trade area whose median household income supports $19 is a mismatch, and the consultant's job is to say so before the loan is made.

On saturation, Parsa's Ohio State research found restaurant density and ownership turnover correlated at 0.99 — an unusually strong relationship, and a caution that a saturated market cannibalises even competent operators. There is no universal restaurants-per-capita benchmark worth applying; the analysis quantifies local leakage or surplus against the actual competitive set.

Builds revenue from operational units

For full service: seats, turns by daypart, days of operation, average check by daypart, with off-premise contribution modelled separately.

For limited service: transactions by daypart, average ticket, and the drive-thru versus in-store split.

Turn benchmarks in current use: quick service 3 to 5 per day, fast casual 3 to 4, family and casual around 3 at dinner, fine dining 1 to 2, full-service overall 1.5 to 2.5.

Sales per square foot is the fastest sanity check available, and the thresholds are well established. For full service: below $150 per square foot leaves little chance of profit; $150 to $250 is break-even to 5%; $250 to $325 supports 5% to 10% of sales. For limited service: below $200 is a likely loss; $200 to $300 is break-even to 5%; $300 to $400 supports 5% to 10% pre-tax.

A projection that lands below those floors has answered the feasibility question already, and no amount of downstream modelling changes it.

Models the ramp honestly

Restaurants do not open at stabilised volume, and they do not ramp in a straight line. The pattern the industry consistently reports:

Months one to three — a novelty or honeymoon period running at roughly 90% to 110% of projected revenue

Months four to six — a decline of 15% to 25% as novelty fades

Months seven to twelve — rebuild through operations and marketing to roughly 75% to 85% of mature run-rate by the end of year one

Months twelve to eighteen — stabilisation

Break-even commonly falls between six and eighteen months. Consistent profitability typically arrives at two to three years.

A model showing a straight line from opening to stabilisation signals inexperience to anyone who has financed a restaurant. The dip is real, it is predictable, and the working capital has to survive it.

These decay percentages are consultancy convention rather than a published dataset, and should be treated as directional — but the shape is not in dispute.

Builds the cost structure to current reality

Prime cost — cost of goods plus labour — is the metric the industry manages against and the first thing a lender tests. Full service typically runs 60% to 65% of sales; limited service 55% to 60%; the industry generally targets no more than 60%.

Food cost typically 28% to 32%, with an industry average around 32.4%. Built from the proposed menu at current input pricing, not from a percentage assumption.

Labour built from a staffing schedule by daypart at local prevailing wages. The median full-service restaurant spent 36.5% of sales on labour in 2024 per NRA operations data; operators reporting losses ran as high as 42.9%.

Turnover is a recurring operating cost, not a one-time opening expense. Industry turnover exceeds 75%. Black Box Intelligence tracks limited-service hourly turnover falling from 133% in 2019 to 110% by Q3 2025, full-service hourly holding around 92%, and management churn easing but still elevated at 44% to 47% in limited service and 35% in full service. Replacement costs run approximately $2,706 per hourly employee, $11,940 per non-general manager and $17,651 per general manager. A fifty-employee restaurant at 80% turnover can face over $400,000 annually.

Occupancy cost at contracted rent plus triple-net charges, with escalations modelled across the projection.

The lines operators consistently understate: insurance, card processing fees which scale directly with sales, repairs and maintenance which arrive unevenly and are frequently omitted in year one, and marketing, which cannot be zero for a new concept.

Replacement reserves. Kitchen equipment, HVAC, furniture and smallwares have finite lives. Where the financing is USDA-guaranteed this is not discretionary — 7 CFR Part 5001 defines debt service coverage as EBITDA less reasonably expected replacement capital expenditures over annual debt service, so an unfunded reserve reduces the coverage ratio the lender computes.

Treats delivery as a separate business

Third-party marketplace commissions run 15% to 30% per order, with pickup orders around 6%. Once packaging, payment processing, promotional participation and refunds are included, the effective cost frequently reaches 30% to 40% of the ticket.

Against independent restaurant net margins of 3% to 5%, a 25% to 30% commission erases order-level margin entirely.

And delivery is partly cannibalistic. Existing dine-in guests shift into a thinner-margin channel rather than adding incremental demand.

The 2026 operating consensus is hybrid: use the marketplaces for discovery, then migrate repeat customers to first-party ordering, which preserves both margin and customer data.

For the consultant, the requirement is channel-level margin analysis rather than a blended assumption. A projection that grows revenue through delivery while holding margin constant has described something that does not happen.

Tests the structure rather than the base case

A single stabilised-year coverage figure tells a credit committee very little. The work that matters:

Coverage at 90%, 85% and 80% of projected revenue, and specifically where coverage crosses 1.00x

What a 200 basis point move in prime cost does to coverage

What a delayed opening does to the ramp and to the interest reserve

What rent escalation does across the loan term

Where the concept depends on a single daypart or a single traffic generator, what happens if that source weakens

What happens if delivery volume arrives as projected but at marketplace rather than first-party margin

The purpose is not to prove survival under every scenario. It is to identify which variables actually govern the outcome and quantify them.

Capital cost, and the second-generation question

Buildout costs, current 2026 ranges:

Second-generation restaurant space: roughly $50 to $150 per square foot

Vanilla or cold shell: $150 to $350 per square foot

Ground-up full service: $250 to $450 or more per square foot

National average across all buildouts approximately $404 per square foot, on a range of $150 to $750

Kitchen equipment and infrastructure typically represents 25% to 35% of total cost. A basic quick-service equipment package runs approximately $40,000 to $75,000; a full-service kitchen with multiple cooking lines and walk-ins runs $150,000 to $500,000 or more. Leasehold improvements alone run $50 to $250 per square foot, with hood and fire suppression at $15,000 to $40,000.

Total project cost by format: the NRA cites median independent full-service startup investment of $275,000 to $425,000. A 2,000 square foot fast-casual buildout in a higher-cost market runs $500,000 to $800,000. Full service in expensive markets reaches $700,000 to $1.1 million. Fine dining can exceed $1.7 million.

The second-generation trap

Second-generation restaurant space looks like the obvious saving. The hood, grease interceptor and kitchen utility rough-in are already installed, and the headline saving against a vanilla shell is $50,000 to $150,000.

The saving is real and it is conditional. Roughly 40% to 60% of the time, the existing electrical service, HVAC tonnage and restroom counts are undersized for the incoming use — because the prior tenant ran a different concept at a different volume with a different equipment load.

When that is discovered after the lease is executed, the cheap deal becomes an expensive one, and the tenant has no leverage left.

The practical instruction is to commission an engineering assessment of electrical capacity, HVAC tonnage and restroom fixture counts before signing the letter of intent. That is a small cost against a six-figure exposure, and it is the single most actionable thing in this article.

Working capital

Budget three to six months of operating expenses. Lenders typically want at least three months in reserve, and first-time operators consistently underbudget — a 15% to 20% contingency above initial estimates is standard practice.

Used equipment sells at auction for 20% to 40% of retail, which lowers initial capital and raises reliability risk, particularly in refrigeration.

Franchise restaurants

Restaurants represent 26.7% of US franchise businesses per the International Franchise Association, and franchised units are heavily represented in SBA restaurant lending.

A franchise feasibility study is not a shorter version of an independent one.

Item 19 financial performance representations are optional under the FTC Franchise Rule at 16 CFR 436.5(s). FRANdata estimates roughly 66% of franchise systems now include Item 19 data, up from 52% in 2014. A Franchise Update survey of more than 110 franchisors found 91% included financial performance representations, of whom 94% listed revenue projections, 63% expenses, 49% profitability and 24% a full profit and loss statement. Roughly 40% of franchisors provide no Item 19 at all — and that absence is itself informative.

Build from the median, not the average. Jack in the Box's 2026 Franchise Disclosure Document reports FY2025 system-average unit volume of $1,913,335 across 1,754 continental US franchise units — but only 40.8% of units exceeded that average, and the median sat $83,000 below the mean. Average EBITDAR was 17.7%. A projection built on the system average is a projection built above the median unit's actual performance.

Item 20 is the table nobody reads. Franchise openings, closures, terminations, non-renewals and transfers appear there. Net unit growth conceals churn; openings minus closures is the number that matters.

Fee structures — initial fees, ongoing royalties and marketing fund contributions — appear in Items 5, 6 and 11 and enter the pro forma as fixed drags before any operator-level margin.

SBA franchise eligibility changed twice. The SBA discontinued the Franchise Directory in 2023 under SOP 50 10 7. SOP 50 10 8 reinstated it on 1 June 2025 with a new model: franchisors execute a one-time SBA Franchisor Certification, replacing the per-loan franchise addendum, and brands not certified by the recertification deadline are removed and become ineligible. The SBA also eliminated "control" as an affiliation test. Verify the brand is currently listed before the file goes anywhere.

Territory and encroachment provisions in Item 12 determine whether the trade area analysed today remains protected in five years.

On whether franchises actually perform better, the evidence is genuinely split. PeerSense's July 2026 data shows a 1.6 percentage point franchise advantage in limited-service restaurants at 17.3% franchise default, and finds franchises defaulting lower in 41 of 99 industries. Other analyses of SBA data from 2010 to 2021 put franchise default around 9.9%, slightly above the broader portfolio. Brand-level dispersion is enormous — the best systems below 5%, the worst above 40%. "Franchise equals safe" is not supportable. Brand-specific data is.

Where restaurant projects fail

Ranked roughly by frequency in the research and industry consensus:

Undercapitalisation. Insufficient working capital to survive the months four to six dip.

Location and trade-area mismatch. A concept and price point that do not fit the market that surrounds them.

Unsustainable occupancy cost. Occupancy — rent plus taxes, insurance and common area maintenance — generally needs to stay within roughly 6% to 10% of sales. Ratios materially above that predict distress, and the lease is signed before a single customer arrives.

Prime cost mismanagement. Food and labour drifting above the segment band with no path back.

Management and turnover. Parsa's research emphasised that family and quality-of-life pressures on owner-operators were a material factor in closures.

Over-optimistic projections. Capture rates and average checks above what the competitive set supports, and ramp assumptions that ignore the novelty dip.

A note on ghost kitchens, because sponsors still propose them. The US real-estate-heavy model has largely collapsed. CloudKitchens ran facilities at approximately 50% occupancy with a 58% restaurant failure rate at surveyed locations, closed its East Oakland facility in November 2025 and delayed its Middle East listing in December 2025. Kitchen United closed or sold all physical units in 2023 and pivoted to software. Reef lost its Wendy's partnership in 2023. Wonder abandoned its van-based model in January 2023. Venture funding into the category fell roughly 95% year over year, from around $210 million in Q4 2024 to around $10.5 million in Q4 2025. Deloitte's Evert Gruyaert told CNBC in February 2024 that "it is clear that the impact of ghost kitchens was overestimated, and we see that today with the decline in ghost kitchens." The frequently cited Euromonitor projection of a $1 trillion global market by 2030, made in November 2019, has been abandoned in practice rather than formally revised.

How the programmes differ

SBA 7(a). The dominant route. SOP 50 10 8 sets when a third-party feasibility study is expected — startups and businesses under two years old, changes of ownership, new construction or major expansion, and special-purpose property. A restaurant acquisition by a first-time operator triggers two of those simultaneously. Note also that the SOP creates a uniform floor: under the previous SOP a Preferred Lender could accept a study meeting its own internal standards, and that discretion is gone.

SBA 504. Applies where the transaction includes real estate — roughly 11% of full-service and 8% of limited-service SBA restaurant loans. Special-purpose and hospitality property carries heightened scrutiny.

USDA B&I. Available for restaurants in communities of 50,000 or fewer under 7 CFR Part 5001, the consolidated OneRD regulation effective 1 October 2020. Loans to $25 million standard, and to $40 million for rural cooperatives with Secretary approval. The guarantee is currently up to 80% — percentages are set by fiscal year and by loan size, and should be confirmed against the current notice. There is no credit-elsewhere test, but borrowers must show sufficient equity. At least 5% of B&I funds are set aside annually for local and regional food projects, and the borrower's headquarters may sit in a larger city provided the financed project is in an eligible rural area. FY2025 saw a record $3.5 billion appropriated at a federal cost of roughly $438 per job created, though USDA Rural Development has absorbed a 36% staffing reduction — which shows up as timeline risk rather than eligibility risk.

USDA's coverage definition is stricter than commercial practice. EBITDA less reasonably expected replacement capital expenditures, over total borrower debt service. On a restaurant with a full kitchen approaching replacement, that deduction moves the ratio materially.

Conventional. No prescribed scope, but lenders frequently require a study on restaurant credits precisely because the loss data is well known to them. The absence of a regulatory template means the consultant writes to the lender's credit policy, and asking for it in advance is worth doing.

In all cases the study must be prepared by an independent third party with no financial interest in the transaction — which excludes the broker, the seller, the franchisor and the equipment supplier.

What a lender is reading for

Is the revenue reachable? Not whether the concept is appealing, but whether the covers, checks and channel mix assumed are achievable in this specific trade area against this specific competitive set.

Is the cost structure current? Prime cost against the segment band, a real staffing schedule, and the lines that get omitted. This is checked first and fastest.

Does the sponsor clear the injection? 10% for a startup or full change of ownership, with seller notes only on full standby and capped at half.

Does occupancy cost work? Within roughly 6% to 10% of projected sales, or the lease itself is a failure predictor.

Does it cover through the ramp, not only at stabilisation? Including the months four to six dip.

Is delivery modelled at channel margin? Or is it blended in at dine-in economics.

Can this operator execute? In a business with median margins in the low single digits, this frequently decides the credit.

Frequently asked questions

Sources

SBA Standard Operating Procedure 50 10 8, effective 1 June 2025, and Procedural Notice 5000-872764, September 2025.

SBA 7(a) loan disclosure data, as compiled by PeerSense (updated March 2026) and iBusiness Funding.

National Restaurant Association, 2026 State of the Industry report, and 2025 Off-Premises Restaurant Trends report.

Circana US foodservice traffic data, 2025 and 2026 projections.

Black Box Intelligence comparable sales, turnover and at-risk unit data, 2025.

Revenue Management Solutions quick-service daypart traffic data, 2025.

US Bureau of Labor Statistics, Consumer Price Index, food away from home and food at home, 2025 and 2026.

Luo, T. and Stark, P.B., University of California Berkeley, 2014, analysis of 81,000 restaurants.

Parsa, H.G. et al., Ohio State University, 2005, restaurant failure study.

Datassential restaurant openings, closures and failure rate data, 2025.

Crestmont Capital, SBA loan default rates by industry, 2026.

FRANdata and Franchise Update Item 19 disclosure surveys; International Franchise Association.

Jack in the Box Franchise Disclosure Document, 2026.

7 CFR Part 5001, USDA OneRD Guarantee Loan Initiative.

Bloom Intelligence restaurant benchmark data.

Prepared by feasibility-study-consultant.com. Industry data reflects published sources at the date below and moves quickly in this sector. Programme requirements including equity injection, coverage floors and guarantee percentages are set by SBA and USDA and are periodically revised; confirm current requirements with the participating lender before relying on any detail. Ramp and decay percentages are industry convention rather than published datasets and are directional. This is not legal, tax or lending advice. Last updated: August 5, 2026.