Why Gas Station Feasibility Studies Get Sent Back

Gas stations are one of only a handful of asset classes SOP 50 10 8 names as special-purpose, one of the largest categories in the history of the 7(a) program, and, on the lender side, one of the most frequently returned feasibility files. The reasons are consistent enough to list. A credit officer or a USDA State Office reviewer reading a fuel and convenience study in 2026 is checking it against a small set of published numbers and a small set of rules, and the studies that come back are the ones that missed one.

What follows is the send-back ledger as it stands in September 2026, ordered by frequency, each with the source the reviewer will use and the fix the consultant should already have applied. The figures are drawn from the NACS State of the Industry release of April 2026 (2025 data), the EIA's 2026 outlooks, the EPA's underground storage tank performance measures for fiscal year 2025, and the SOP text in force, with SOP 50 10 8.1 noted where it changes the acquisition rules from 1 October 2026.

1. Fuel margin taken from the 2025 headline

The single most common return. In 2025 retail fuel margins averaged above 40 cents per gallon, against roughly 22 cents before 2020, and the trade now repeats that "40 is the new 20." A study that capitalizes 40 cents into a 25-year pro forma is capitalizing a cycle. Deflate the series and about half of the nominal doubling since 2019 is inflation. Walk the NACS waterfall and the 40-cent gross lands at 10 to 15 cents of net after card fees, distribution and store-level fuel operating cost; card fees alone, a record $21.3 billion industrywide in 2025 including $4.6 billion charged on the taxes the retailer merely collects, run about 8.4 cents a gallon and rise mechanically with the pump price. NACS's own analysts say the business now requires elevated fuel margins just to break even.

The fix. Base-case a single site in the high 20s to low 30s per gallon, not 40. Deduct card fees as a percentage of the projected pump price, not a fixed cent figure. Stress the worst year back to the low 20s. Treat any trailing margin above the low 30s as cyclical upside the study describes but does not capitalize, and say so in the assumptions list.

2. Gallons without a capture case

The second return is a fuel volume projection with nothing under it: no traffic count from the state DOT, no capture rate, no competitor gallons, no distinction between a legacy corner and a large-format entrant. The national picture is flat to declining. U.S. gasoline product supplied peaked in 2018 at about 9.33 million barrels a day, ran about 8.9 million in 2025, and the EIA projects volume declining through 2026 and 2027 as fleet efficiency improves; NACS reports gallons sold at convenience stores up 0.5% in 2025 on lower prices. Against that, supply is reconstituting rather than shrinking: total stores drifted down to 151,975 at the end of 2025 from a 2018 peak of 154,958, while fuel-selling sites rose to 122,620, an eight-year high, and the chains are building large. Murphy USA's recent new-store classes pump over 310,000 gallons a month; a legacy independent site moves on the order of a million gallons a year. A new large-format store two miles away is a bigger threat to a legacy site's volume than any national demand curve.

The fix. State the ADT on the fronting road and its source and date, distinguish one-direction from two-direction counts, apply a capture rate defended by comparable sites, name every competitor within the trade area with its format, position count and estimated gallons, and identify any permitted or under-construction competitor. For a new build, show the volume the cost basis requires and the market evidence that the site can reach it. For an acquisition, reconcile the seller's gallons to fuel supplier invoices and tank inventory records, not to the seller's summary.

3. Inside sales modeled at chain averages

The store carries the site. In 2025 fuel produced 65% of convenience industry sales dollars and 38.8% of gross profit dollars; the inside of the store produced the rest. Foodservice was 28.5% of in-store sales and 38.9% of in-store gross profit, up from 11.9% of sales in 2005, with prepared food 73.9% of the category and packaged beverages the second category at 18.7%. In-store sales hit a record $341.2 billion, the 23rd consecutive annual increase, on submissions from more than 35,000 stores. The average basket was $7.69. Those are the numbers a study copies, and they are the wrong numbers for a single site, because the dispersion behind them is the whole story: NACS reports the top decile of stores earns 46.6% of inside profit from foodservice and the bottom decile 8.3%, and the gap between top and bottom deciles is tens of thousands of dollars a month per store.

The fix. Build the inside from the subject's own trade area and format, not the industry mean. Model foodservice with a labor line, a waste line of at least 15%, the hood and ventilation capital a hot program needs, any franchise or licensing royalty, and delivery revenue at the margin left after platform commissions. Show cigarettes declining, not flat; the largest operator in the channel reported cigarette volumes down 26% since 2019 when it announced its North American closures. A study that projects chain-average inside sales for a 1,500-square-foot rural box is a study a reviewer stops reading.

4. An environmental file that stops at the Phase I

Gas stations sit on SBA's list of environmentally sensitive industries in Appendix 6 of SOP 50 10 8. The consequence is not only that a Phase I Environmental Site Assessment is required regardless of loan size; the SOP also requires tank and line testing indicating no deficiencies, a Phase I conclusion that either states no further action is needed or recommends a Phase II, and, where a Phase II follows, SBA's required reliance letter. Contamination does not end the loan, but it stops approval and disbursement unless a mitigant the SOP recognizes is documented: coverage by a state trust fund, a no-further-action or closure letter from the regulator, an environmental indemnification, or an escrow sized at 150% of the estimated remediation cost. Under USDA the environmental review under 7 CFR Part 1970 runs on its own track and the Agency will not guarantee a loan until it is complete.

The fix. The study is not the Phase I, but it must reconcile to it. State the Phase I date, its conclusion, whether a Phase II was recommended and completed, the tank and line test results and date, any open release, and which mitigant applies. Where the site relies on a state fund, confirm the fund is open to new releases and adequately funded; 36 states run cleanup funds, three have closed to new releases, several are thin, and California's sunsets in 2036. A fund that is closed is not a mitigant.

5. A tank field with no age and no reserve

The nation operates 533,277 federally regulated underground storage tanks at about 190,000 facilities. The 1988 federal rule forced every pre-existing tank to upgrade or close by December 1998, which created a single giant vintage now 28 to 32 years old; the EPA's own analysis puts roughly 275,000 operating tanks already past 30 years and expects more than 100,000 to come out of the ground by 2034. Fiberglass and protected-steel tanks carry 30-year warranties. When tanks come out, the ground tells the truth: California's forced closure of every single-walled tank by 31 December 2025 revealed previously unknown contamination at 29.7% of single-walled closures against 6.7% for double-walled. Nationally 581,676 releases have been confirmed since the program began and 53,777 remain open. The average completed cleanup costs $142,219, and the number of cleanups exceeding $1 million doubled between 2014 and 2018.

The fix. Tank age, material, wall construction and the date of last testing are underwriting fields, not appendix trivia. A study on a site with tanks past 25 years should carry a replacement plan; past 30 years or single-walled, it should carry a funded reserve equal to replacement cost plus a probability-weighted cleanup, and the pro forma should show the reserve as a use of cash. A 25-year loan closing in 2026 will see the 1998 cohort replaced before maturity. The study should say whose capital does that.

6. Special-purpose rules not reconciled

SOP 50 10 8 defines special-purpose property as "a limited market property with a unique physical design, special construction materials, or a layout that restricts its utility to the specific use for which it was built," and its non-exhaustive list includes gas stations with or without convenience stores. Three consequences follow. Under 504, the borrower contribution rises to 15% for special-purpose collateral and 20% where the business is also a start-up. The appraisal must be a going-concern report from a Certified General appraiser with at least four equivalent going-concern assignments in the prior 36 months, ordered by the lender, allocating value across land, building, equipment and intangibles, and the loan amount is capped at appraised value. And on a change of ownership numbered on or after 1 October 2026, SOP 50 10 8.1 requires 1.25x coverage on trailing EBITDA with no use of projections to meet it, caps total transaction debt at the appraised business value, and requires a lender-ordered quality of earnings report where the business purchase price, net of owner-occupied real estate, is $3 million or more.

The fix. Reconcile the study's cash flows to the appraisal's component allocation, so fuel, inside and rent are consistent with the value assigned to intangibles, equipment and real estate. State the equity injection the structure actually requires and the sources that fund it. On an acquisition, present coverage on the historical basis the lender must code and treat the buyer's improvement plan as the risk case the QoE and valuation do not address.

7. Cash books nobody verified

Fuel and convenience is a cash-heavy, single-operator industry: about 63% of stores belong to operators with ten or fewer locations and roughly 60% are single-store operators. The seller's summary income statement is not evidence. The Office of Inspector General's recurring deficiency list on early-defaulted 7(a) loans reads directly onto this asset class: "unverified seller's financial statements, all liabilities not considered, impact of affiliates not considered, unsupported projected sales," and, on change-of-ownership loans specifically, an OIG judgment that such transactions "continue to be an area of high risk for SBA."

The fix. Reconcile fuel revenue to supplier invoices and gallon reports, inside sales to point-of-sale and lottery and tobacco distributor statements, both to bank deposits, and all of it to IRS transcripts. State which figures were verified to source and which were carried from the seller. Flag the fuel supply agreement: its term, its expiry relative to the loan term, any exclusivity or volume commitment, and any branding incentive that has to be repaid on early termination. Flag a ground lease with no real estate collateral, because the special-purpose rules assume the dirt is in the deal.

8. A state and local layer the study never read

There is no national fuel market. State taxes and fees on gasoline ranged from 9.0 cents in Alaska to 70.9 cents in California as of 1 January 2026, before the unchanged 18.4-cent federal excise; retail prices in mid-2026 spanned roughly $3.35 in Indiana to $5.50 in California. Wisconsin mandates a minimum markup; New Jersey bans self-service. California lost about 17% of its refining capacity with the Los Angeles and Benicia closures, layers a low-carbon fuel standard and cap-and-trade cost of roughly 23 to 27 cents a gallon onto the pump, and forced the closure of about 50,000 single-walled tanks by the end of 2025. At the municipal level, Petaluma became the first U.S. city to ban new gas stations in February 2021 and roughly fifteen jurisdictions, clustered in the Bay Area, have followed; Florida has preempted such bans. Break-even volume, cross-subsidy arithmetic and the legality of the pricing model all change at the state line.

The fix. State the applicable state and local tax stack, pricing law, tank rule and any ban or moratorium in the study's regulatory section. Then read the ban correctly: for an incumbent, an entitled, compliant site inside a jurisdiction that prohibits new stations has a frozen competitive set for the life of the loan, which is a credit positive the study should claim rather than a headwind it should apologize for.

9. The EV assumption in the wrong direction

Studies get this wrong both ways: some ignore electrification, more overstate it. EVs are about 2% of the vehicles on the road, and the fleet's average age is a record 12.8 years, so new-vehicle efficiency reaches gallons slowly. The federal purchase credit expired on 30 September 2025; EV share of new sales spiked to a record 10.5% in the third quarter as buyers beat the deadline and fell to 5.8% in the fourth, where it has stayed into 2026. The same July 2025 law zeroed CAFE penalties, and Congress revoked California's Advanced Clean Cars II waivers in June 2025. The EIA's 2026 long-term outlook now spans an 11% to 23% petroleum decline by 2050 depending on whether standards survive. What a lender can use is geography: EV share of new sales runs roughly 23% in California and under 1% in Mississippi, Wyoming and North Dakota.

The fix. Underwrite gallons by site type and state, not by national headline: roughly minus 0.3% to 0.8% a year at rural, diesel-weighted, low-EV sites; minus 1% to 1.5% on typical suburban corridors; minus 2% to 4%, stressed to 5%, in dense high-EV coastal metros. Credit a charger only where utility make-ready or grants cover the capital and the site can show dwell time converting to basket; national fast-charging utilization ran about 16% against a roughly 20% break-even in 2025, and only about 1.4% of convenience stores offer charging.

What the default record actually says

A study that argues from the folk reputation of the asset class gets returned for a different reason: the lender has the tape. Gasoline stations with convenience stores are the fourth-largest category in the history of the 7(a) program, about $17.7 billion across roughly 21,000 loans since 1995. Measured as dollars charged off against dollars approved, the lifetime rate is about 3.5%, better than restaurants. Measured as defaults against resolved loans, the frequency runs near 15%, several times the program cohort. Both are true. Frequency is high and severity is low, because when these loans fail the land, building and tanks recover most of the balance, and long real-estate-secured structures have defaulted at a fraction of the rate of short goodwill-heavy loans. A study should say which denominator it is using and why the subject's structure sits at the good end of the distribution.

The pre-submission checklist

Before the study leaves the consultant, the reviewer's questions should already be answered on the page.

1. Fuel margin base case in the high 20s to low 30s per gallon, card fees as a percentage of pump price, worst-year stress to the low 20s, and any trailing margin above the low 30s treated as uncapitalized upside.
2. ADT with source, date and direction; capture rate defended by comparables; every competitor named with format and estimated gallons; permitted and under-construction entrants identified.
3. Inside sales built from the subject's trade area and format, with labor, waste, hood capital, royalties and delivery commissions in the model and cigarettes on a declining line.
4. Phase I date and conclusion, tank and line test results, Phase II status and reliance letter, any open release, and the mitigant relied on, with the state fund's status confirmed.
5. Tank age, material, wall type and last test date stated; replacement plan past 25 years; funded reserve past 30 years or single-walled.
6. Cash flows reconciled to the going-concern appraisal's allocation; the equity injection and its sources stated; on acquisitions numbered from 1 October 2026, coverage on trailing EBITDA at 1.25x with the QoE figure carried through.
7. Revenue reconciled to supplier invoices, POS, distributor statements, bank deposits and IRS transcripts, with verified and carried figures labeled; fuel supply agreement term and exit terms stated; ground lease flagged.
8. State tax stack, pricing law, tank rule and any local ban or moratorium stated, with the entitlement treated as the moat it is.
9. Gallon decline by site type and state; charger capital excluded unless externally funded and dwell-to-basket is evidenced.
10. Default denominator stated, and the subject's structure placed against it.