Gas Station Underwriting Method: Fuel Gallons, Inside Sales and the Class's SBA Record

A gas station fuel and inside sales underwriting model has to answer two questions that most other asset classes fold into one. How many gallons will the forecourt sell, and at what margin per gallon? And how much will the people standing at the pumps, plus the ones who never buy fuel, spend inside the store? The two answers rest on different public data, fail in different ways and carry different weight in the cash flow. What follows is the station method in order: how each answer is built, which published series and rules anchor its figures, what the federal tank rules add to the credit file, which program rules decide whether the borrower qualifies at all, and what SBA's own loan file reveals about stations, along with what it leaves out. A short worked model under the series conventions closes it.

The method here is the class-specific layer. The general tools it borrows, from trade area capture to stress testing, have their own articles in this series and are linked where they are used. The full worked version of a station study is the Tennessee 7(a) example; the archived post on SBA loans for gas stations covers program choice for the same class. Nothing below values a property: a study tests whether a project's market and cash flow can carry the debt asked for.

Why a Station Is Underwritten as Two Businesses

The industry code itself describes the split. NAICS 457110, gasoline stations with convenience stores, covers businesses that retail automotive fuels together with a limited line of groceries ^[1]. Its sibling, NAICS 457120, covers stations with no convenience store and also truck stops ^[2], and the examples the classification lists for that code include marine service stations ^[3]. A food mart that operates no fuel pumps is classified elsewhere, as a convenience retailer ^[4]. Every gas station in the loan record therefore sells fuel, and those coded 457110 also run a small grocery business on the same lot.

The fuel side is a commodity sold against the station across the road. EIA notes that a retailer's cost of doing business varies greatly with where the station sits ^[5], and that even nearby stations can differ in traffic, rent and supply source, with the count and position of local competitors also bearing on price ^[6]. Ownership varies too: some outlets are owned and run by refiners, others are independent businesses buying from refiners and marketers for resale ^[7]. The method treats the inside store as a different business, driven by footfall and basket size rather than by gallons, with its evidence in different records.

A study keeps the two apart until gross profit, for three reasons. First, the drivers differ: gallons follow traffic, resident fuel use and the station's share of local pumps, while inside sales follow transactions, the share of fuel customers who walk in, and the average ticket. Second, the margins differ, so a single blended margin hides which half carries the debt. Third, the sales dollars of the fuel side are a poor measure of volume. Census's annual survey shows gasoline station sales rising 28.5% from $572.7 billion in 2021 to $736.1 billion in 2022 ^[8], over a stretch in which EIA's monthly retail price for regular went from $3.064 a gallon in June 2021 ^[9] to $4.929 in June 2022 ^[10]. Sales equal price times gallons, so a jump in dollars can arrive with flat or falling volume. Census's monthly estimates put 2025 sales of gasoline stations at $633,659 million ^[11]; that total says how much money crossed the counter nationally, not how many gallons any one forecourt sold. The underwriting unit for fuel is the gallon, priced in cents of margin, and the underwriting unit for the store is the customer.

The Gallons Equation, Built Twice

A study for a new or substantially changed station should estimate gallons by two independent routes and carry the lower one unless site evidence supports the higher. The first route works from the road; the second works from the people who live nearby. Both are equations whose every input is either a published figure or a labelled assumption, and Table 1 lists the inputs of the illustration used throughout this article. The illustration is a made-up station, not a real project.

Route one, traffic. Gallons per day equal the annual average daily traffic past the site, times the share of passing vehicles that stop for fuel, times the gallons sold per stop. The traffic figure is the published one: a state transportation department's count for the road segment. Readers of those counts should know what they measure. Minnesota's transportation department describes its AADT as an estimate of all vehicles using a length of road, in both directions, on any given day of the year ^[12], built with factors that adjust for season, day of the week and vehicle type, and it warns that the values vary ^[13]. FHWA's Traffic Monitoring Guide describes the same machinery from the federal side: adjustments for the hour, the weekday, the month and the year, derived from continuously operating count stations, convert a short count into an AADT estimate ^[14], and the result is influenced by how long the count ran, how well the chosen factor fits the site's monthly and weekly pattern, and how variable the traffic is ^[15].

Two practical consequences follow. A count can be old: Minnesota schedules counts on its trunk highways every three years and on local system roads every six ^[16], so a study should print the count year beside the figure and say whether anything has opened or closed on the corridor since. And a count can be thin: Minnesota's 48-hour short counts are typically set on a Monday, Tuesday or Wednesday ^[17], so a station whose trade leans on weekend or seasonal traffic sits on a factored estimate of exactly the days it depends on. Because the count covers both directions, a study should also say which direction the site serves well. A curb cut on the far side of a median, or a left turn across heavy oncoming traffic, can take much of one direction out of the reachable stream. The stopping share is the assumption that carries this route, and it is the one a study should defend with something other than judgment: the gallons and counts of comparable stations, or the subject's own history where there is one.

Route two, residents. Annual gallons equal the trade area's residents, times highway gasoline use per resident, times the site's share of the trade area's fueling positions, grossed up for the share of the site's sales made to people who live elsewhere. Per resident use comes from two federal series. FHWA's Table MF-21 reports 2024 U.S. highway use of 127,278,127 thousand gallons of gasoline and 44,533,935 thousand gallons of special fuel ^[18]. Census's Vintage 2025 estimates put the population at 341.8 million on July 1, 2025, after growth of 1.8 million over the year ^[19], which places the July 2024 population at 340.0 million and per resident highway gasoline at 374.3 gallons for the year, as Table 2 computes. The illustration uses the national row only to show the arithmetic. A study should use its own state's MF-21 row and the Census estimates for the trade area's own geographies, because nothing guarantees that the national average holds in any one state.

The share term is a fair share: the site's fueling positions over all positions in the trade area, the station's own included. The trade area and demand methods article covers fair share, capture and penetration in general; what is particular here is that fueling positions are countable in the field, and a study should count them, competitor by competitor, rather than infer them. The inflow term carries sales to people who live outside the trade area. At a site that leans on through traffic it can be large, and because no public series measures it for one site, a study should say what supports the figure it uses.

Rounded as shown; lines may not foot by one unit in the last place.

In the illustration the traffic route gives 3,564 gallons a day and the resident route 2,930, so the study carries 2,930. That works out to 366 gallons per fueling position per day, a figure a reviewer can set beside the positions and the volume claimed for any comparable. When the two routes disagree widely, the gap is information. A traffic figure far above the resident figure suggests the site depends on through traffic, so the inflow assumption and the access questions deserve more scrutiny. A resident figure far above the traffic figure suggests the stopping share is too low, or that the count missed the road most residents use.

Diesel needs its own line. MF-21 counts 44,533,935 thousand gallons of special fuel used on highways in 2024 ^[18], and EIA's weekly survey priced on-highway diesel at $6.127 a gallon on November 9, 2026, $2.290 above a year earlier ^[31]. The class in the loan file includes truck stops. A study of a truck stop, or of any site where diesel is a material share of the gallons, should model diesel as a third route, from truck counts by vehicle class, and carry its margin and price separately from gasoline.

Inside Sales: Customers, Conversion and the Ticket

Inside sales are built from people, not from gallons. Fuel transactions per day equal carried gallons divided by gallons per stop. Some share of those customers walk inside and buy something; others drive off. A second group never buys fuel at all: the neighbour who stops for coffee, the worker who buys lunch. Annual inside sales equal fuel customers who convert plus inside-only customers, times the average ticket, times the days the store is open. Table 3 runs that chain for the illustration: 325.6 fuel transactions a day, 130.2 of them converting at the assumed 40 percent, plus 250 inside-only customers, for 380.2 inside customers and inside sales of $1,179,683 a year.

No public series reports a station's conversion rate, its count of inside-only customers or its ticket, and a study should say so in its inputs table rather than borrow a national figure that may not fit the site. For an existing store the evidence is the point-of-sale record: daily transaction counts, sales by category and the store's own gross margin by category, read over at least the trailing year so seasonality shows. For a purchase, the series article on acquisitions sets out why the test rests on the seller's historical earnings. For a new store the evidence is comparable stores, chosen as the supply and comparable analysis article describes, and the result is still an assumption, labelled as one.

Inside sales per gallon is a useful cross-check because it ties the two halves together. The illustration produces $1.10 of inside sales for every gallon sold, and a reviewer can compare that ratio with the subject's history or with the comparables the study relies on. Where the records show categories earning different margins, the study should carry each category at its own margin rather than one average, so that a shift in mix shows up in gross profit. Lottery commissions and any gaming revenue belong on their own lines for a second reason: the program rules described below test gambling income against a share of gross revenue, and the lender cannot apply the test if the study has folded those receipts into other income.

Inside conversion is also where a new store's ramp sits, since a new station starts with no regular customers at the pumps or inside; the ramp-up and absorption article covers how a study should show the years before stabilisation instead of starting at the stabilised volume.

Margins: Cents per Gallon and What the Pump Price Contains

In the illustration the assumed fuel margin is 8.0% of the November pump price, so the margin assumption carries more weight than the price. EIA's weekly survey priced regular gasoline at $3.995 a gallon nationally on November 9, 2026, $0.939 above a year earlier ^[21], and its retail prices include all taxes ^[20]. EIA quotes the motor fuel taxes in cents per gallon. It puts the federal tax at 18.40 cents per gallon, an 18.30 cent excise tax plus the 0.1 cent federal Leaking Underground Storage Tank fee, and state taxes and fees on gasoline at an average of 33.55 cents per gallon as of January 2026 ^[30]. Together that is 51.95 cents a gallon on those averages, a part of the pump price that is tax, not margin.

The public series nearest to a retail margin is EIA's distribution and marketing component. EIA derives it as a residual: the weekly survey's average retail price less the crude oil, refining and tax components ^[32]. It is broader than the station's margin, because distribution, marketing and retail dealer costs and profits all sit inside it ^[33]; EIA's account of that stage runs from the pipelines that carry gasoline to terminals near where it is used to the tanker trucks that deliver it to each station ^[34]. It also swings. EIA explains that retail prices lag spot prices, so the refining and the distribution and marketing components can vary from month to month ^[35]. In August 2026 the component was 12.5 percent of a $4.058 retail price ^[27]; in March 2026 it was 7.6 percent of $3.638 ^[28]; in December 2025 it was 23.9 percent of $2.894 ^[29]. Table 4 converts those shares to cents.

The component moved by 41.5 cents a gallon between December and March, against an assumed station margin of 32.0 cents in the illustration. A study cannot treat the component as the margin, and it should not treat a single month's margin as normal either. For an existing station the margin evidence should be the operator's own fuel invoices set against its pump price records, month by month, over enough months to include a falling market and a rising one. For a new station the margin is an assumption, and the honest presentation is a base figure with a test of what a lower one does, as the worked model below shows. The card cost deserves the same care. In the illustration card processing is charged on every sales dollar, taxes included, which makes it rise with the pump price even when the margin per gallon does not.

Station Measures and the Records That Support Them

A station study should put a short list of numbers in front of the credit officer and, beside each, the record that supports it. The worked figures in this series follow one set of conventions, which apply wherever this article tests debt service coverage.

Conventions. This article follows the conventions of the Loan Analytics worked examples. Coverage is tested against the program floor the site's program page states, where the program has one, and against a stated 1.25x lender assumption, and the higher of the two binds. Equity is measured at the loan the cash flow supports, with fees recomputed on that loan. An acquisition is tested on the seller's historical earnings, not on projections.

Table 5 lists the measures this article treats as the core of a station file, the evidence a study should produce for each, and the published series or rule that anchors it.

The first four rows build the numerator. Gallons and the margin per gallon give fuel gross profit; inside sales and their margins give inside gross profit; the price and its taxes drive the card cost. A lender reading the result should look at the fuel share of gross profit, which tells it how much of the cash flow depends on the forecourt: in the illustration fuel provides 45.6% of gross profit at the base case. The next three rows are the tank file, covered in the next section, and the brand row is covered with the program rules. The last row is the test itself. It is run twice, first at the minimum ratio named on the site's SBA program page for an ordinary 7(a) loan ^[26] and then at the 1.25x figure assumed for the lender, and the stricter of the two governs; the series piece on coverage and program floors explains that rule and how the supported loan follows from it.

The Tank File Under 40 CFR 280

Underground storage tanks bring a station under a body of federal regulation of its own. 40 CFR part 280 sets the standards, and the agency that enforces them is EPA or, where a state runs an approved program, the designated state or local agency ^[41]. The federal floor reaches every regulated petroleum tank, and a study should know at least the parts that create cost, records or risk.

Release detection never pauses. A tank put in the ground on or before April 11, 2016 needs leak monitoring at intervals of 30 days or less, by one of the methods the rule lists and subject to its stated exceptions ^[36]; tanks and piping installed or replaced after that date need secondary containment, monitored in the interstitial space, with certain suction piping excepted ^[42]. The installation date of each tank is therefore a line in the file, not a footnote. Under the walkthrough option, the inspection checks spill prevention and release detection equipment every 30 days, although spill equipment at a system that takes deliveries less often may be checked before each delivery instead ^[37], and spill prevention equipment and containment sumps used for interstitial monitoring are tested at least once every three years unless double walled and monitored ^[43]. A suspected release must be reported to the implementing agency within 24 hours or another reasonable period the agency sets ^[44]. Closing a tank for good, or converting it to store something the rule does not regulate, first requires sampling for a release at the spots where contamination is most probable ^[45], which matters to any plan that replaces old tanks. Each facility needs at least one designated Class A and one Class B operator for each tank or group of tanks ^[46], so the study's staffing assumptions should name who holds those roles.

Financial responsibility is a cost line and, in some states, a condition of receiving fuel. Owners or operators of petroleum tanks must demonstrate financial responsibility for corrective action and for compensating third parties for bodily injury and property damage from accidental releases ^[47]. A facility selling petroleum to the public is a petroleum marketing facility under the rule's definitions ^[48], and for tanks at such facilities, or handling more than 10,000 gallons a month on average over the prior calendar year, the per occurrence amount is $1 million ^[38]. The annual aggregate is $1 million for an owner or operator of 1 to 100 tanks and $2 million for 101 or more ^[49]. States add their own gates. In Texas, owners or operators of regulated tanks must renew the facility's delivery certificate each year to keep receiving fuel ^[39], must show proof of current financial assurance at renewal ^[50], and must post the certificate where the people delivering fuel can see it ^[51]. Where a state ties deliveries to a certificate in this way, a lapse interrupts supply, so a study should record the certificate's expiry date beside the loan's first payment date.

The national record shows why the file deserves a close reading. EPA counted 534,189 active petroleum tanks under its program at approximately 190,224 facilities as of March 2026 ^[52]. Its technical compliance rate for April 2025 to March 2026 was 60.9% ^[53], and over the same months 90.4 percent of facilities complied with financial responsibility requirements and 83.2 percent with the walkthrough requirement ^[54]. EPA recorded 1,997 confirmed releases between October 2025 and March 2026 ^[55] and 52,859 releases still awaiting cleanup as of March 2026 ^[56]. None of those figures describes a particular site. They do say that a compliant tank file cannot be presumed, and that a study should state which records it has seen and which it has not.

The rule also speaks to lenders directly. A lender with a security interest in the tanks, or in the land they sit on, falls outside owner status for the purposes of the technical, corrective action and financial responsibility requirements, so long as it takes no part in managing the tanks and stays out of petroleum production, refining and marketing ^[57]. Before foreclosure, the holder is not an operator as long as it does not control or carry responsibility for the tank's daily operation ^[58]. After foreclosure, the holder is not an operator if another party runs the tanks day to day and can be held responsible ^[59]; with no such operator, it keeps that position only if, among other conditions, it empties its known tanks within 60 calendar days or another period the implementing agency sets ^[60]. Those provisions tell a lender what it would face in a workout, and a study that names the operator, the tank count and the state program gives the lender what it needs to plan for one.

The SBA and USDA rules point the same way. Except where a supplemental guarantee agreement or a CDC agreement provides otherwise, a lender or CDC needs SBA's prior written consent before taking certain servicing and liquidation actions ^[61], and the list includes taking title to environmentally contaminated property or taking over a business that handles hazardous substances or wastes ^[62]. In the 504 program, environmental studies are among the professional fees that count as eligible project costs ^[63]. Under the B&I rules, when the agency will take a real estate lien the appraisal must weigh how a release of hazardous substances or petroleum products could affect the collateral's market value ^[64], borrower and lender must consider the project's environmental impacts from its earliest planning stages ^[65], and the lender must alert the agency to any environmental issue or item that may need extensive review ^[66]. A feasibility study is not an environmental site assessment and should not pose as one. What it can do is carry the tank file's facts into the cash flow: the compliance testing and pollution insurance lines in the expense budget, any tank replacement in the capital plan, and the dates on which the state's certificate and the insurance renew.

Program Rules That Decide Whether a Station Is Eligible

Every SBA loan starts from one credit standard. The applicant must pass a creditworthiness test that also reaches any Operating Company, and each loan must look safe enough, on the evidence, that repayment is reasonably assured ^[67]. Among the conditions SBA can attach to a loan is a feasibility study ^[68]. Past that common ground, three rules bear on stations in particular, and a fourth issue, the brand, sits beside them.

Passive owners. Where the owner of a station's real estate is not the business that runs it, the passive business rule applies. SBA treats a passive business owned by a landlord or developer as ineligible when it neither uses nor occupies what the loan pays for, the exception being an Eligible Passive Company ^[69]. That route carries conditions, among them the four below, and a study should test each against the signed documents rather than the summary the sponsor offers:

• the lease between the property company and the operating company is in writing and ranks behind SBA's lien ^[70];
• the rent is capped: it may pay the lender's installment plus what the property company spends directly to hold the site, for example upkeep, insurance and property tax ^[71];
• counting only renewal options the operating company alone controls, the lease does not expire before the loan matures ^[72];
• the operating company signs as guarantor or co-borrower, with co-borrowing compulsory if the 7(a) proceeds also buy the operating company working capital or other assets ^[73].

Where a dealer who is not part of the ownership runs the station, each of those conditions reaches the dealer's lease and the dealer's signature on the loan, so a study should establish who runs the site before it models whose cash flow repays the debt.

Leased space. Some stations rent part of the building to a food tenant or another business. Where SBA money builds a new station building, the owner's own business needs 60 percent or more of the space that can be rented, and no more than a fifth may go to permanent tenants ^[74]; the owner must also expect to move into part of the unleased remainder within three years and into all of it within ten ^[75]. In an existing building the owner needs a 51 percent share at minimum, with as much as 49 percent let to others ^[76]. A study should measure the rentable property and print the split, tenant by tenant.

Gaming revenue. SBA's own line sits at one-third: legal gambling above that share of gross annual revenue makes a business ineligible ^[77]. B&I draws its line lower and counts more receipts. Gambling income may not pass 15 percent of the entity's annual gross revenue, the count takes in "any lease income from space or machines", and the rule goes on "excluding State-authorized lottery proceeds or Tribal-authorized gaming proceeds, as approved by the Agency" for the purpose it then states ^[78]. A station with video gaming terminals, a gaming room let to an operator, or heavy lottery sales should present those receipts on their own lines and compute each program's share on that program's basis, so the lender can see the test rather than take it on trust. B&I layers a location test on top. As a starting point, no place with more than 50,000 inhabitants counts as rural for the program, nor does the urbanized area adjoining it ^[79], but the definition carries further conditions: part of that adjoining urbanized area can still qualify once it has been found to be rural in character ^[80], a finding the Under Secretary of Rural Development makes for each project on its own facts ^[81]. A study for a station near a large town should therefore say which test it relies on, and the series article on the B&I program sets out the full definition.

Brands and supply agreements. SBA keeps a Franchise Directory of the brands it treats as eligible for its financing, and a brand appears there only once SBA has reviewed its business model ^[82]. The directory can hold a fuel brand whose retailers are tied to it by a dealer, jobber or licence arrangement, where SBA finds that the tie meets the FTC's franchise definition ^[40]. When a brand is sent for review, SBA wants full copies of every such agreement ^[83], and its instructions name the fuel supplier among the parties whose contact must be given ^[84]. Brands that meet the FTC's definition of a franchise must be listed to obtain SBA financing ^[85], and SBA adds that a listing neither endorses nor approves the brand and does not ensure that the business succeeds ^[86]. For a branded station the study should identify the supply agreement, set its term against the loan's term, and confirm the brand's directory status. The program articles for 7(a) loans, the 504 program and B&I guarantees cover each program's full set of tests.

The Class in the SBA Loan File

The Loan Analytics dataset treats gas stations and truck stops as one class built from NAICS 457110 and 457120, and because 457120 holds truck stops, every figure in this section includes them; the dataset cannot split them out. How loans filed under older code vintages reach the class is set out in the series piece on assigning loans to classes by code. The dataset's notes for the class add that EV charging has no entry of its own in the 2022 index, so a charging business does not surface as a class of its own.

From FY2017 to FY2026 the 7(a) program approved 6,403 ^[22] gas station and truck stop loans for $9,344,305,700 ^[22]. The median approval was $1,095,000 ^[22], with the middle half of approvals between $510,000 ^[22] and $2,069,450 ^[22]. The 504 program approved 954 ^[87] debentures in the class, truck stops again included, for $1,063,613,000 ^[87], at a median of $860,000 ^[87]; the 504 amount is the debenture, not the project, since the bank's first lien sits outside it. Approvals carrying a franchise code made up 46.68 percent ^[22] of the class's 7(a) count and 58.91 percent ^[87] of its 504 count. The share counts approvals with a franchise code entered; the dataset's figure does not separate fuel brands from store brands.

The count did not move in a straight line. Class 7(a) approvals, truck stops included, stood at 690 ^[22] in FY2017, peaked at 983 ^[22] in FY2021, fell to 533 ^[22] in FY2022 and were 581 ^[22] in FY2026. Dollars followed a different path: $776,512,300 ^[22] in FY2017 against $1,118,171,800 ^[22] in FY2026, so FY2026 approved fewer class loans than FY2017 for more money. On the 504 side the class went from 99 ^[87] debentures in FY2017 to 116 ^[87] in FY2021 and 80 ^[87] in FY2026. The dataset records these counts without reasons, and this article offers none.

What the record holds is approvals. It does not show applications declined, the gallons or inside sales of any borrower, its margins, its tank file or its supply contract. A separate piece maps the gaps in the public loan file field by field. For a station study the record is useful in two ways: it sizes a typical request, which is why the worked model below borrows the class median, and it gives the outcome floors read in the next section. Table 6 gathers the figures this article relies on, each with its key.

Loan Data Behind This Article

Every figure in this table comes from the Loan Analytics loan dataset, analytics-loan-data freeze-3 (commit 360c856), and carries its key. SBA rates are approval-cohort floors and are withheld when fewer than 30 loans stand behind them.

Reading the Class's Floors Against the Program's

Two outcomes in the file can end an approval without repayment, and the series treats them separately: a cancellation ends the approval before any money moves, while a charge-off is a loss on a loan that was disbursed. The article on cancellation against charge-off explains the difference. Both rates are floors on approval cohorts, and the cohort floor article explains why each can only understate the final figure.

Cancellation. Of class 7(a) approvals made from FY2017 to FY2024, truck stops included, 793 ^[22] of 5,142 ^[22] were cancelled, a floor of 15.42 percent ^[22]; for 7(a) as a whole, the same approval years produce 11.24 percent ^[89]. Treating the 13 ^[22] class approvals that remain undisbursed as cancellations lifts that to a ceiling of 15.67 percent ^[22], against a program ceiling of 12.24 percent ^[90]. In 504, the class floor is 19.82 percent ^[87] against a program figure of 13.60 percent ^[91], and the class ceiling 24.64 percent ^[87] against 16.90 percent ^[92]. In both programs, then, a larger share of station approvals ended before funding than across the program. The file does not record why an approval was cancelled, and the dataset's report gives no reason for the class pattern, so this article does not supply one.

Charge-off. Of class 7(a) loans approved from FY2017 to FY2021 and disbursed, 34 ^[22] of 2,939 ^[22] carried a charge-off status at the September 30, 2026 cut, a floor of 1.16 percent ^[22] against 6.33 percent ^[88] for the program. Counting only losses booked within five years of approval, the class figure is 0.51 percent ^[22] against 3.72 percent ^[93], and by dollars it is 0.28 percent ^[22] against 1.95 percent ^[94]. In 504 the class floor, truck stops included, is 0.85 percent ^[87], which is 3 ^[87] charge-offs among 355 ^[87] disbursed debentures, against 0.35 percent ^[95] for the program; within five years the class shows 0.00 percent ^[87] against 0.17 percent ^[96]. The denominator clears the 30-loan rule, so the rate prints, but with three losses in the numerator one more would lift it by more than a quarter of a point; the article on why rates under 30 loans are withheld explains how small counts behave.

Why are these only floors? A large part of each cohort is still open, and SBA's FOIA data dictionary keeps back the status of any disbursed loan not yet cancelled, paid in full or charged off ^[97]. Inside the class's 7(a) charge-off denominator, truck stops included, 1,033 ^[22] loans are open, and inside the 504 one, 248 ^[87]. Any of them may still be charged off.

What, then, can a study take from this? Three things, and no more. First, the class's 7(a) charge-off floor sits well below the program's, and its 504 floor above the program's on very few losses; neither says how one station will perform. Second, cancellations run higher than the program in both programs, which a study should note without explaining. Third, the floors pool every kind of fuel retailer in the codes, truck stops and stations without stores included, across all states. A lender should read them as context for the class, not as a probability for the project, and the reasons behind any of them stay outside the file.

A Mini-Example Under the Series Conventions

The illustration carried through this article now becomes a coverage test. The loan request is set at the class's median 7(a) approval, a median that includes truck stops, and the inputs table carries it under its cell key. Prime, the base, stood at 7.00 percent in each daily reading of the week to November 6, 2026 ^[23]. To it the model adds a spread of 2.25 points, an assumption that keeps the price below SBA's ceiling for a variable rate on a loan of this size, which is the base plus 3.0 percentage points for anything above $350,000 ^[24]. Amortization runs 25 years, the maximum 13 CFR 120.212(c) allows ^[25], which assumes a loan led by the real estate. The cash flow is the stabilised year from Table 2 and Table 3, less card costs, operating expenses, a market wage for the owner-manager and a replacement reserve. Table 7 runs the base case and four single shocks.

At the base case the station produces cash flow of $166,612 against debt service of $112,529, a coverage of 1.48x. The binding minimum is the 1.25x lender assumption, since it sits above the 1.15x floor on the site's page, and at that minimum the supportable loan comes to $1,297,025, which is $202,025 more than requested. The case looks comfortable, and each of the four columns moves one assumption to see how comfortable it is.

A margin 5 cents lower drops coverage to 1.01x. Volume fifteen percent lower drops it to 0.96x, because lost gallons also take inside customers with them. A ticket one tenth smaller leaves 1.16x, which clears the site's floor but fails the 1.25x lender assumption, so it fails the test. Prime 2 points higher leaves 1.27x, which passes. In this illustration rate risk is the mildest of the four, and the two assumptions a study can least prove for a new site, the margin and the stopping share behind the gallons, do the most damage.

Table 8 turns the same model into break-even points. Coverage reaches the binding minimum at 2,734 gallons a day, 93.3% of the carried volume, and at a fuel margin of 29.6 cents a gallon, a cushion of 2.4 cents below the assumed 32.0. Set beside Table 4, where the distribution and marketing component moved by 41.5 cents between two months, that cushion is the point of the exercise. A station that passes at 1.48x can still sit a few cents of margin from failing, and a study should print that distance rather than leave the lender to infer it. How large a shock to test, and how to stack several, gets its own treatment in break-even and stress tests. For the full version of a station study, including project costs, the first years of trade and a complete operating statement, see the Tennessee example.

A Review List for a Station's Credit File

A lender or borrower reading a station study can work through a short list. None of it requires data the study should not already hold.

• Two routes to gallons. The study should show a traffic estimate and a resident estimate, say which it carries and why, and print the count year and both directions' access for the traffic figure ^[16].
• A defended stopping share. The share of passing vehicles that stop should rest on comparables or on the site's own history, never on a round number with no source.
• The state's own fuel row. Per resident use should come from the state's MF-21 row and Census estimates for the trade area, not the national average the illustration here uses ^[18].
• Margin over time. The fuel margin should be shown monthly for an existing station and stressed for a new one, with the public component beside it as context only ^[35].
• Inside sales by category. Inside sales should be built from customers, conversion and the ticket, with lottery and gaming receipts on separate lines so each program's gambling test can be applied ^{[77] [78]}.
• The tank file. The study should list each tank's install date and construction, the release detection method, the last walkthrough and three-year test dates, and the financial responsibility mechanism, and say which records it has seen ^{[36] [43] [38]}.
• The state certificate. Where the state issues one, the registration or delivery certificate should be current and its renewal date known ^[39].
• Who operates the station. If a property company owns the site and an operating company runs it, the lease terms should be tested against each passive company condition ^{[71] [72]}.
• The brand. A branded site should show its supply agreement, the agreement's term against the loan, and the brand's directory status ^[82].
• Distance from failure. The study should print the break-even gallons and break-even margin at the binding minimum, not only the base coverage.
• The class record, read as context. Class floors should be quoted with their keys and their limits, truck stops included in every figure, and never as a forecast for the project.

Each item on the list is a place where a station study either shows its evidence or relies on an assumption the reader has to take on faith. The method's purpose is to make that difference visible.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• SBA feasibility studies (7(a) and 504)
• USDA feasibility studies
• Bankable feasibility studies for conventional loans
• Gas Station and Convenience Store Feasibility Study
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: SBA 7(a) Gas Station Feasibility Study in Tennessee
• SBA Loans for Gas Stations: 7(a) vs 504, the Environmental File, and Who Actually Gets Approved

Commission a Feasibility Study

Loan Analytics prepares lender-grade feasibility studies for SBA 7(a), SBA 504, USDA Business and Industry, conventional and EB-5 financing, built on the methods this series sets out. A study is not an appraisal: it tests whether a project's market, costs and cash flow can carry the debt a lender is asked to approve. To discuss a study for your project, contact Loan Analytics.

Daniel Smith, MAI, Loan Analytics