Car Wash Underwriting Method: Membership Demand, Tunnel Throughput and the Class's SBA Record

An express car wash sells two things through one machine. It sells a monthly membership, which a driver pays whether or not the car is dirty that month, and it sells a single wash to whoever turns in from the road. A lender financing a new tunnel is relying on both revenue lines and on a building with few other uses. The sections below give the class's method: how the member count is reached from public counts of vehicles and residents, how throughput is checked against the speed of the tunnel in its busiest hour, which metrics a credit officer reads and what anchors each, what the SBA loan record for car washes shows and what it cannot, and which rules decide whether and how a wash is financed under 7(a), 504 or a USDA guarantee. A short model closes the article. The full version of that model, with a sourced market and a complete cost stack, is the worked example this series pairs with the class, a tunnel in Maryland (Worked Example: SBA 504 Express Car Wash Feasibility Study in Maryland).

The general methods live elsewhere in this series, and this article does not repeat them. The article on trade areas and demand covers drive-time boundaries and penetration rates in any class, the supply article the competitor census, the ramp-up article stabilization, and the coverage article the ratio itself. The site's archived car wash posts, linked at the foot of this page, describe the market and its operators. This one is the method behind the numbers, written so that a lender's reader can trace each line.

The Code Behind the Class: What 811192 Holds

Every loan figure in this article comes from loans SBA filed under NAICS 811192. The Census Bureau's 2022 manual defines that industry as establishments primarily engaged in cleaning, washing or waxing automotive vehicles, among them passenger cars, trucks, vans and trailers ^[1]. Its illustrative examples are automotive detail shops, mobile car and truck washes, car washes and self-service car washes ^[2]. The code describes an activity, not a building. An express tunnel on two acres, a row of self-service bays, a wash for tractor-trailers, a detail shop in a strip center and a van that cleans cars in office parking lots can all carry the same six digits.

The Loan Analytics asset map records this as the class's limitation: truck washes and mobile washes sit in the same code as car washes, and nothing in the loan file lets them be split out. The way loans are assigned to classes, and the other places where two industries share a code, are covered in the NAICS mapping article. For this article the consequence is simple to state and easy to forget. Wherever a car wash loan count, dollar total or rate appears below, it is the record of the whole code, with express tunnels and every other format in it counted together. The SBA file does not say which format a loan financed. The spread of loan sizes shows how mixed the record is: the middle half of 7(a) car wash loans runs from $150,000 ^[3] to $2,360,000 ^[3], and the file cannot say which of those loans built a tunnel.

The breadth of the code follows a study into its supply count. County Business Patterns, the Census series that counts establishments by county and industry, covers establishments with paid employees ^[4]. Businesses with no paid employees or payroll are counted in a separate series, Nonemployer Statistics ^[5]. A count of 811192 establishments from either series mixes tunnels with detail shops and mobile operators. A study should therefore treat the published count as a ceiling on the code in the county, then build the list of competitors it actually tests from a site by site survey of the trade area, recording each wash's format, number of tunnels or bays, posted prices and membership plans. A mobile operator without employees is invisible in County Business Patterns, and an express tunnel with one manager and a part-time crew is counted beside a detail shop with ten employees. Neither series ranks the competitors that matter to a new tunnel; the survey does.

Building the Member Count From Public Counts

Membership is the revenue line a lender most wants to see, and none of the public series this article draws on counts it: they count vehicles, residents, establishments and traffic. So the model reaches the member count in five steps, each an equation whose inputs are labelled as sourced or assumed:

1. Vehicles per resident: r = R / N, where R is the state's registered motor vehicles from FHWA Table MV-1 and N is the state's resident population from the Census Bureau's estimates for the same year.
2. Vehicles in the trade area: V = P × r, where P is the resident population inside the drive-time boundary.
3. Memberships held in the trade area at any wash: M = V × m, where m is the share of those vehicles that carry a membership somewhere.
4. Fair share of the subject tunnel: f = t / (t + c), where t is the number of tunnels at the subject site and c the number of competing tunnels open or permitted inside the boundary.
5. Members at the subject site at stabilization: S = M × f × k, where k is a share index that sets the new site against a full fair share.

Two of the inputs are public. FHWA counts 23,606,401 registered motor vehicles in Texas in 2024 ^[6], and the Census Bureau puts the state's resident population at 31,318,578 on July 1, 2024 ^[7]. Table 2 turns the pair into 0.75 vehicles per resident. Texas stands in for whichever state holds the site; the arithmetic does not change.

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

The registration count is broader than the cars a tunnel washes. Table MV-1 counts buses, trucks and motorcycles beside automobiles, and in Texas the 2024 table shows 15,933,517 trucks against 7,281,613 automobiles ^[6]. A study should not strip the trucks out, because the table does not separate pickups and vans from heavy trucks and because the class washes both. It should instead say that r is a rate of all registered vehicles and let the penetration input m absorb the difference, since m is estimated against the same base.

Two cautions travel with the series. FHWA notes that the table reflects State-supplied data without its own estimation procedures, except where noted ^[13], and that its 2024 figures use actual State-reported data after estimated data for 2022 and 2023 ^[14]. A change in registrations from 2023 to 2024 therefore mixes a change of method with any real growth. A study that wants a registration trend for its market should take it from the state's motor vehicle agency on one consistent basis, by county where the agency publishes counties, and should not read growth into the switch.

The trade-area population P is an assumption in the example. In a study it is a sum of Census tracts inside a drive-time boundary drawn around the site, with the boundary pulled in where a competitor or a barrier cuts the drive; how tracts are summed, and how their margins of error combine, is in the trade area article. The example uses 85,000 residents, which yields 64,069 vehicles.

The penetration input m carries the most judgment and the least public support. A study should estimate it from its own survey of competing washes: the share of cars in member lanes at observed peaks, the plan prices on each menu, and any member counts a competing wash has made public. Where the subject is an operating wash, the seller's membership ledger replaces the estimate for the subject site but not for the market. The example assumes that 14.0 percent of trade-area vehicles hold a membership somewhere, which gives 8,970 memberships across all washes inside the boundary.

The fair share counts tunnels, not sites, because a site with two tunnels has roughly twice the hourly capacity of a site with one. One subject tunnel against two competing tunnels gives a fair share of 33.3 percent. The share index k then discounts the new site, since its members have to be won from washes that already hold them; the example sets k at 90.0 percent. A study should defend k from the evidence it has: how the competing tunnels compare on location, access and age, and whether any of them is newer than the subject would be. The model reaches 2,691 members at stabilization (Table 2).

The model is a chain of multiplications, and that is its main hazard. An error of a tenth in any link passes straight into the member count, and the shock table in the short model shows what a member count 15 percent lower does to the loan. Stabilization is also not the opening month: the path from the first member to the stabilized count, and how a study proves its length, is in the ramp-up article. The model states the destination; it does not state how quickly the site gets there.

Throughput: Passing Traffic, Tunnel Speed and the Busiest Hour

Retail washes are the part of car wash demand that still moves with the road. The model takes retail washes per average day as D = A × q, where A is the annual average daily traffic on the road the site fronts and q is the share of passing vehicles that buy a single wash on an average day. The example assumes a count of 32,000 vehicles and a capture of 0.25 percent, which gives 80 retail washes a day and 28,800 a year over 360 operating days.

A comes from the state transportation department's count program, and a study should be clear about what that figure is. FHWA's Traffic Monitoring Guide says the AADT statistic represents a volume for an average day of a calendar year and is not meant to be a precise count of the vehicles on any exact day ^[15]. Where the count at the site is a short count, it reaches an annual average through factors computed from continuous count stations ^[16], and FHWA sorts those factors into four kinds, by hour of the day, by weekday, by month and by year ^[17]. A study should print the count's year, whether it came from a continuous station or a short count, and the factor group the agency used. It should also not borrow a capture rate measured on one kind of road, such as a commuter arterial, and apply it to a count on another, such as a retail corridor with weekend peaks, because the two carry different mixes of trips and different hourly shapes.

The tunnel caps the business. Its rated speed in cars per hour, times the hours it is open and the days it operates, gives annual capacity: K = s × h × d. The example's tunnel, rated at 120 cars an hour and open 12 hours a day, could wash 518,400 cars a year. Member washes are the member count times the washes each member takes a month, times twelve; with 3.0 washes a member each month the example's members take 96,872 washes a year, and all washes total 125,672, or 24.2 percent of annual capacity.

That annual figure flatters the site, because washes do not arrive evenly. The load that binds is the busiest hour of a peak day: B = (W / d) × p × u, where W is all washes in a year, p is the peak day's volume against an average day and u is the share of a day's washes that arrive in its busiest hour. With a peak day 1.8 times an average day and 14.0 percent of a day's washes in its busiest hour, the example brings 88 cars to the tunnel in that hour, 73.3 percent of its rated speed (Table 2).

The peak check tells a lender whether the member forecast can physically be served. If the forecast implies a busiest hour above the rated speed, the extra cars queue, leave, or come back on a quieter day, and only the last of the three keeps the revenue. A study should test the queue the site plan provides against the peak-hour figure, and it should take the rated speed from the equipment maker's specification for the tunnel length proposed rather than from the fastest tunnel on the market. Where a member forecast is raised to make a loan work, the peak-hour row is the first place to look, because a higher member count at the same peak shape loads the busiest hour first.

Members are a flow as well as a stock. With monthly churn of 7.0 percent, holding 2,691 members takes 188 new members a month (Table 2). The model expresses that need against retail traffic, 7.8 sign-ups for every 100 retail washes, because the retail customer is the nearest pool from which a member can be drawn. The two demand lines are linked through that ratio: a site with thin passing traffic must convert its retail customers at a higher rate to hold the same base. A study should print the conversion its member count requires beside the member count itself, so that a credit officer can judge whether the site's traffic can supply it.

Revenue follows from the counts. At a net price of $24.00 a member each month, after discounts, the membership line is $774,976 a year; at an average retail ticket of $15.00, the retail line is $432,000. The membership share of wash revenue, 64.2 percent, is an output of the model, not an input to it. A study that starts from a target share and works back to a member count has reversed the method.

Seven Metrics on the Credit Officer's Desk, and Their Anchors

A car wash credit file turns on a short list of measurable things. Each has an evidence source and, where one exists, a published series or rule against which it can be checked.

Members, churn and net price are three numbers, and a study should keep them apart. The member count is a balance at a date. Churn is the monthly loss from that balance, which for a recurring card charge includes both cancellations and renewals that fail. Net price is what a member pays after introductory offers, family plans and discounts, not the posted price on the menu board. For an operating wash a study should take all three from the point of sale system month by month, for at least the period the seller has run a membership plan, and reconcile the membership receipts to bank deposits. For a new wash the same three numbers come from the competitor survey and the model above, and the study should say so plainly rather than present them as observed.

Buying an operating wash brings in the acquisition test. In this series a purchase is tested on the seller's historical earnings, not on what a buyer plans to do with the membership plan, and the series sets out why in the acquisition article. Coverage, the measure every other metric feeds, is tested in this article under the conventions this series shares with its worked examples:

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.

The site's SBA program page gives 504 a floor of 1.15x, which may be shown on two-year projections as well as on history ^[12]. The worked examples also test against a 1.25x lender assumption, and whichever is higher governs. The coverage test then gives the lender a member count to check rather than a ratio: the number of members at the stated net price that the binding minimum requires, which the mini-model below prints as a row.

Equipment life decides how a 504 project is drawn, and it is a schedule, not a single figure. A tunnel's package includes a conveyor, wash arches and brushes, high-pressure pumps, dryers, a water reclaim system, vacuum stations, pay stations, gates and software. Those items do not share a life, and the rule that admits fixed assets to a 504 project turns on life, as the eligibility section below sets out. A study should list each component with the life the maker states and carry that list into the cost stack.

The wash water route decides whether the site can open at all. A tunnel discharges wash water that carries soap, grit and oil, and which authority governs depends on where that water goes: a connection to a municipal sewer brings the sewer authority's rules, while a discharge, an impoundment or a tank system on site can bring a state permit, as the Oklahoma example below shows ^[23]. A study should obtain the authority's written acceptance or the state authorization before the loan closes and note any renewal date that falls inside the loan term.

Size matters most for an operator with several sites, because the affiliates' size is counted with the applicant's ^[24]. A study should total the receipts of every affiliated site before relying on the standard.

The Car Wash Record in SBA 7(a) and 504

Between approval fiscal years 2017 and 2026, code 811192 carries 2,713 ^[3] 7(a) approvals and 997 ^[25] 504 approvals in the dataset. Every truck wash, mobile wash and detail shop in the code is counted beside the tunnels, and no figure in this section can separate them. The 7(a) approvals total $3,840,167,000 ^[3]; the 504 approvals total $1,209,738,000 ^[25], and for 504 that figure is the debenture alone, so the bank's first lien on those projects is not in it.

Loan sizes differ by program. The median 7(a) car wash loan is $789,000 ^[3]. The median 504 debenture is $1,094,000 ^[25], with the middle half of debentures running from $564,000 ^[25] to $1,655,000 ^[25]. Because the debenture is only one layer of a 504 project, the projects behind these debentures are larger than the figures shown, by the bank loan and the borrower's contribution, neither of which this field records.

Franchised borrowers are a small part of the record. Approvals carrying a franchise code are 3.06 percent ^[3] of 7(a) car wash approvals and 5.02 percent ^[25] of 504 car wash approvals. A study of a franchised wash should still read the franchise disclosure document, but it should not expect the SBA record to say much about franchised washes as a group.

The 7(a) record reaches most of the country, and three states lead it. California holds the most 7(a) car wash approvals in the window, 406 ^[3], followed by Texas with 336 ^[3] and Florida with 208 ^[3]. A study of a site in a state with a thin record should use the national class figures and say why, rather than lean on a handful of local loans; the state cells in the same dataset carry their own counts, and any rate in them on fewer than 30 loans is withheld.

The 7(a) count peaked in FY2021 at 391 ^[3] approvals and has been lower in every year since; the 504 count stayed between 74 ^[25] and 121 ^[25] in each year from FY2017 to FY2025. FY2026 is a short year in the file, with 192 ^[3] 7(a) approvals, because the file holds nothing from the first six weeks of that fiscal year, October 1 to November 12, 2025, as the dataset's report records; the year covers about ten and a half months. The report offers no cause for the class's trend, and none is offered here. What a study can say is where the class stands: these counts belong to one code among the many SBA lends to, and the program totals they sit inside are in FIGURES, which the block below cites with the rest (Table 5).

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 Car Wash Floors Against the Program Floors

Every rate below is a floor on an approval cohort, computed by the method the charge-off floor article sets out. The charge-off floor divides charge-offs so far by the disbursed loans of the FY2017 to FY2021 approval cohort; the cancellation floor divides cancellations by all approvals of FY2017 to FY2024. Loans still open stay in the denominator, but their status is exempt from disclosure ^[26], so a later charge-off can only raise a rate. Each rate is the record of code 811192 as a whole, truck washes, mobile washes and detail shops included.

The 7(a) charge-off floor for car washes is 2.47 percent ^[3], on 33 ^[3] charge-offs among 1,337 ^[3] disbursed loans, where the 7(a) program's whole cohort shows 6.33 percent ^[27]. On the fixed five-year horizon, which measures every cohort year over the same stretch after approval, the class shows 1.50 percent ^[3], set beside 3.72 percent ^[28] for the program entire. In dollar terms the class's charge-offs equal 0.27 percent ^[3] of its approved dollars, against 1.95 percent ^[29] across the whole program. Of the class's denominator, 395 ^[3] loans are still open, each of which could still end in a charge-off.

The 504 record runs the other way. The car wash charge-off floor is 1.00 percent ^[25], against 0.35 percent ^[30] for every 504 loan in the cohort; within five years of approval it is 0.75 percent ^[25] against 0.17 percent ^[31], and by dollars 1.28 percent ^[25] against 0.29 percent ^[32]. The class's 504 rate rests on 4 ^[25] charge-offs among 402 ^[25] disbursed debentures, 199 ^[25] of them still open. It is printed because its denominator clears the 30-loan rule set out in the suppression article; its numerator is small, and one more charge-off would move it visibly. A reader should weigh it as a count of a few loans, not as a measured tendency of the class.

Cancellation is a different event from charge-off, and the cancellation article explains why the two must be read apart. The 7(a) car wash cancellation floor is 10.84 percent ^[3], with a ceiling of 11.59 percent ^[3] if every approval still undisbursed were to cancel; the whole 7(a) program's pair is 11.24 percent ^[33] as floor and 12.24 percent ^[34] as ceiling. In 504 the car wash floor is 16.46 percent ^[25] and its ceiling 21.91 percent ^[25], beside a program pair of 13.60 percent ^[35] as floor and 16.90 percent ^[36] as ceiling; 45 ^[25] car wash 504 approvals in the cohort are still undisbursed, which is why the 504 range between floor and ceiling is wide.

The floors cannot say why any loan ended as it did. The status field records an outcome such as cancelled, charged off or undisbursed ^[37], and nothing in the file records the reason. Nor can the floors pull the tunnels out of the code. They cannot rank one car wash format against another, and they cannot be turned into a probability that a given tunnel will fail, since the loans behind them were approved in other years, at other costs and in other markets. A class floor below its program's floor does not show that the class is safe, and a class floor above its program's does not show that it is risky; both describe a cohort whose open loans have not finished their lives.

What the floors can do in a study is modest and useful. They place the class within its program, so a credit officer reading a car wash file knows the record that program has with the code. They show which event, cancellation or charge-off, is the larger part of the class's record in each program. And they give the study's downside section a public, keyed reference point that a reader can check, which is more than a sentence about industry risk can offer. A study should print them as floors, with their keys and cohorts, and never as a forecast for the subject.

The Rules That Decide How a Car Wash Is Financed

An operating business, not a landlord

SBA finances the company that runs the wash; an owner who only collects rent from it qualifies through one route alone. Under 13 CFR 120.110(c), a developer's or landlord's passive business is ineligible when it does not itself occupy or use the assets financed, the Eligible Passive Company being the exception ^[38], and SBA's 504 page names passive activity among the kinds of business it will not lend to ^[39]. Where an investor owns the tunnel and a separate company runs it, the deal has to fit the Eligible Passive Company mold, and three of that structure's conditions shape a car wash file. First, the lease, counting any renewal only the tenant can trigger, may not end before the loan does ^[40]. Second, the operating company is on the hook for the loan, as co-borrower or guarantor ^[41], and if the 7(a) money also covers the operator's working capital or other assets, a guarantee falls short and the operator must sign as co-borrower ^[42]. Third, rent is limited to the loan payment plus the owning entity's direct costs of carrying the property, which include maintenance, insurance and property taxes ^[43]. That third condition decides where coverage is tested: with rent set to meet debt and carrying costs and nothing more, the owning entity has no surplus to test, so the study tests the operator.

A wash building with space the operator does not use, a detail bay or a lube bay let to someone else, brings in the occupancy rules. In a new building the operator's own use has to reach 60 percent of the rentable space, while up to 20 percent may sit with permanent tenants ^[44]. Beyond that, the rule asks for a plan: part of whatever is neither occupied nor permanently leased has to come into the operator's own use within three years ^[45], and every unleased foot of it inside ten years ^[46]. For an existing building, the operator's share can fall to 51 percent, with as much as 49 percent let ^[47]. A study should measure rentable area from the plans, not assume the wash fills its building.

The 504 contribution and the limited purpose question

In a 504 project the borrower's cash turns on two questions: how long the business has been operating, and whether the tunnel counts as a limited or single purpose building. The two answers set both the borrower's minimum and the bank's, as Table 6 lays out.

A tunnel built by a newly formed company meets the first condition. The second is a classification of the building, and a study should not present its own reading as settled. It should state which reading its structure assumes and show the other beside it, so that the lender sees what a change in classification does to the borrower's cash and to the bank's exposure. The 504 method article covers the project split and the job test in general.

Equipment against real estate

The split between equipment and real estate decides what may sit inside a 504 project, and it is the point at which car wash projects differ most from an office or a warehouse. Under 13 CFR 120.120(a)(6), the fixed assets in a 504 project "must have a useful life of at least 10 years and be at a fixed location," and the rule leaves a narrow opening for short-term financing of equipment, furniture and furnishings, which has to be both essential and a minor share of the whole project ^[18]. SBA's own 504 page makes the same cut from the other side: it admits machinery and equipment when a decade or more of useful life remains ^[19], it bars working capital and inventory ^[51], and it offers maturities of 10, 20 and 25 years ^[11].

For a tunnel the rule is applied component by component. A study should place inside the project only the items whose stated life reaches the threshold, keep the rest outside, and say how the outside items are paid for: in cash, by the bank on its own terms, or by a separate 7(a) loan. In the example below, the items assumed to last at least 10 years (conveyor, arches, dryers and water reclaim) sit in the project, and pay stations, software, signage, opening chemicals and a reserve sit outside it (Table 7).

7(a) reaches the same split through the term instead of through eligibility. Its proceeds can pay for working capital, inventory, supplies and raw materials ^[52], the first two of which a 504 loan cannot fund ^[51]. A 7(a) term runs ten years at most, and only real estate or equipment with more than ten years of useful life earns a longer one ^[53]; a loan for equipment may add up to 12 months where they are needed to finish the installation ^[54], a provision a tunnel package set inside a new building may well use; the outer limit is 25 years once extensions are counted ^[55]; and the rule asks for the "shortest appropriate term" given what the borrower can repay ^[56]. The component schedule is thus also what supports any 7(a) term past ten years on a wash's equipment. On a large tunnel the size cap can bind before coverage does, because the program caps any one 7(a) loan at $5,000,000 ^[57]; the 7(a) method article covers the program's other tests.

Size, and the code that sets it

SBA's size standard for 811192 is $9.0 million in annual receipts ^[21], in a table whose standards are stated in employees or in millions of dollars of receipts ^[58]. For 7(a) and 504 there is a second route to small status, measured with affiliates included: tangible net worth of $20 million or less, together with net income after federal income taxes, taken as the mean of the last two finished fiscal years, of $6.5 million or less ^[22]. The code that sets the standard is the borrower's primary industry, which for a car wash operator is 811192 whatever format it runs.

USDA Business and Industry guarantees

OneRD's rural area is, among other conditions, land "not in a city or town that has a population of more than 50,000 inhabitants", nor inside the urbanized ring around one ^[59]. The rule then takes back into the rural area any piece of that ring judged "rural in character" ^[60]; who judges is the Under Secretary of Rural Development, and each judgment attaches to one project only ^[61]. A site that passes is eligible ground for a B&I guarantee, the other route for a car wash, and a study should quote the definition in full and say, for a site in an urbanized fringe, whether a determination exists for this project. What else the Agency tests is covered in the B&I method article. The Agency can call for an independent study when the lender's package, read through the five feasibility study components, does not show how the loan will be repaid, or when the project would weigh heavily on operations already running ^[62]. What such a study contains may change with the project's scope, and a qualified independent third party should prepare it ^[63]. 13 CFR 120.160(b) gives SBA the same lever, naming a feasibility study beside appraisals and a survey among the conditions it may impose ^[64].

The state's wash water file

The state gate this article documents for the class is the wash water file rather than a business license. Oklahoma's general permit for vehicle wash facilities is one state's version. It authorizes vehicle wash operators to discharge wastewater, run surface impoundments or install and operate subsurface tank systems under its conditions ^[23], and a new facility must apply for and obtain an Authorization before it starts any activity the permit regulates ^[20].

The permit's edges follow the shared code closely. It excludes steam cleaning and mobile washing ^[65], sends a facility that washes interiors to an individual permit ^[66], and excludes washes without grit traps or oil and water separators ^[67]. Its definition of vehicles leaves out buses, recreational vehicles, motor homes, tank trucks, garbage trucks, livestock trucks and trailers, farm and oil field equipment, cement mixers and construction equipment ^[68], and washes in the excluded categories need an individual wastewater permit or another general permit ^[69]. An express tunnel, a truck wash and a mobile washer filed under one NAICS code can therefore sit under three different environmental files in a single state. The permit itself expires at midnight on March 31, 2028 ^[70], inside the term of any loan made now, so a study should record the expiry and the conditions a renewal may change.

A Short Model: One Express Tunnel Under 504

Here the demand model feeds a 504 structure for a ground-up tunnel that a new company builds. On the reading that the tunnel is a limited or single purpose building, the borrower puts in 20 percent of what the 504 project costs ^[8], and the bank's first lien carries 50 percent ^[10]. Every other input is printed, with its basis, in the inputs table above. Table 7 puts $5,600,000 inside the 504 project and $400,000 of short-life items outside it, paid in cash, so the borrower brings $1,520,000 in all against a contribution of $1,120,000.

At the base case the tunnel earns $506,930 of net operating income against annual debt service of $375,852, which is 1.35x coverage (Table 8). The site's SBA program page sets the 504 floor lower, at 1.15x ^[12], than the 1.25x this example assumes for the lender, so the assumption binds, and the base case clears it with $37,115 of income to spare.

Three shocks show where the margin sits. With members 15 percent lower, coverage falls to 1.22x, below the binding minimum by $11,709 of income. A net price 10 percent lower leaves coverage at 1.26x, just above it. Rates one point higher on both loans take coverage to 1.23x. The member shock does the most damage, in part because it is the biggest of the three and in part because the member count drives the larger of the two revenue lines, which is what the chain of multiplications in the demand model leads a reader to expect.

The last rows turn the test into the figure a lender can check against a ledger. At the base net price the site needs 2,384 members to meet the binding minimum, 307 fewer than the model reaches, a margin of 1.6 months of sign-ups at the churn assumed. If the net price falls 10 percent, the requirement rises to 2,649 and the margin shrinks to 42 members. A credit officer can hold either figure against the member counts that comparable washes in the trade area support, and against the ramp the ramp-up article asks a study to prove, without rebuilding the model.

The mini-model leaves things out on purpose. It has no ramp, no expense build line by line (the expense article sets one out), no fees on either loan, no construction contingency (the cost evidence article covers it) and no grid of joint shocks (the stress testing article shows how to size one). The worked example for a Maryland tunnel carries the market evidence, the cost stack, the debt layers and a program comparison in full, and it is the place to see the method applied to a named market.

Before the File Goes to Credit: Car Wash Items a Study Should Settle

A study for a car wash loan should close with the items a credit officer will test first, each traceable to a page of the study:

• The breadth of code 811192 stated wherever an SBA loan figure appears, since the code holds detail shops and mobile washes as well as tunnels ^[2].
• A vehicle base drawn from one year and one method, with the 2024 change in FHWA's registration basis noted where a trend is shown ^[14].
• The traffic count's year, station type and factor group, and a capture rate calibrated on the same kind of road ^[16].
• A competitor survey by format and tunnel, with the published counts for the code used only as a ceiling ^[4].
• Members, churn and net price kept as three separate figures, each with its evidence, and the member count the binding minimum needs printed beside the forecast (Table 8).
• The busiest hour of a peak day checked against the tunnel's rated speed and the queue on the site plan (Table 2).
• A component schedule of equipment life, with only qualifying items inside a 504 project ^[18].
• The 504 contribution and the bank's share shown under both readings of the building ^[8].
• Where an investor owns the real estate, each Eligible Passive Company condition met in the lease and the guarantees ^[40].
• The wash water route, with the sewer authority's acceptance or the state authorization and its expiry ^[20].
• The class's floors printed as floors, with their keys and cohorts, and no cause asserted for them (Table 5).

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
• Car Wash Feasibility Study
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: SBA 504 Express Car Wash Feasibility Study in Maryland
• Car Wash Feasibility Study: A Comprehensive Guide for U.S. Developers, Investors, and Lenders
• SBA Loans for Car Washes: 7(a) and 504 Eligibility, the Equipment-Real Estate Split, and Membership Revenue in the Credit Memo
• The Express Car Wash Membership Monetization Benchmark: 2026 Unit Economics for SBA-Financed Operators

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