An LBO model sounds like something investment bankers build at 2 a.m. It is not. It is a short list of questions about one business, answered in a spreadsheet you build yourself: what you pay, how you pay for it, what the business earns, and what happens to you when it earns less. This lesson walks through the model in words first, then in numbers, using the Coastal Comfort HVAC case. Your job here is judgment, not Excel. The arithmetic is simple. Deciding which assumptions to believe is the hard part, and nobody else can do it for you.
Principle
LBO stands for leveraged buyout. Leverage means borrowed money. An LBO is buying a business mostly with loans, paying the loans down out of the business's own cash flow, and owning what is left. That is exactly what a self-funded search is: one buyer, an SBA 7(a) loan, a seller note, a small equity injection, and a personal guarantee.
Three layers of money sit in that deal, and you need to know each one by name:
- The senior loan. In self-funded search this is usually an SBA 7(a) loan: a bank loan partly guaranteed by the U.S. Small Business Administration. It gets paid first, every month, no matter what kind of month the business had.
- The seller note. Part of the price the seller agrees to receive later, as a loan to you. It lowers the cash you need at closing and keeps the seller invested in the transition.
- The equity injection. The cash that goes in ahead of the debt: your own money, plus any investors'. It is the first money lost if the business fails.
This school's teaching baseline for that stack is 65 percent SBA debt, 15 percent seller note, 15 percent investor equity and 5 percent your own capital, and its standing warning is that leverage above 90 percent will kill you. The SBA's own rules set the edges of the stack. Under SOP 50 10 8.1, effective October 1, 2026 (Y25), an initial acquisition requires an equity injection of at least 10 percent of total project cost (all costs required to become operational, regardless of the source of funds, except for lines of credit and 504 loans), and that minimum cannot be reduced. At least half must come from unlimited sources such as cash that is not borrowed. The limited sources, which include seller debt on full standby (no payments of principal or interest for the term of the SBA loan) and equity from minority investors who own under 20 percent and have no control, can together provide no more than half. (Exception: from Veterans Day 2026 through fiscal year 2027, an eligible veteran buyer may raise the full 10 percent from permitted outside sources, such as a full-standby seller note or investor capital (Y29). SBA Procedural Notice 5000-883564 requires that veterans own at least 50 percent of the borrower and that the loan receive its SBA loan number by September 30, 2027; confirm with your lender. The injection, the 1.25 coverage floor and the full-standby rule still apply.) Change-of-ownership loans amortize over no more than 10 years; the exceptions are a real estate portion and, under the technical updates effective October 1, 2026, a loan of up to 25 years where the purchase includes an owner-occupied special purpose property integral to the business and 85 percent or more of total project cost is for real estate (Y26). And seller earnouts are prohibited inside a 7(a) acquisition.
The three numbers you read
Every model ends in three numbers. Define them once and keep them straight:
- IRR (internal rate of return): your annual return on the cash you put in, accounting for when each dollar comes back. Think of it as the interest rate your equity earned.
- MOIC (multiple on invested capital): how many times your money you got back. Put in $250,000, take out $750,000 over the hold, and your MOIC is 3.0x.
- DSCR (debt service coverage ratio): the cash the business generates in a year (here, EBITDA, which is earnings before interest, taxes, depreciation and amortization) divided by that year's loan payments. At 1.0x the business exactly covers its payments with nothing left over. Below 1.0x, you are covering them.
This school's rules of thumb are an IRR of 25 percent or more, a MOIC of 3.0x or more over the hold, and a DSCR of 1.25x or more in every year. Treat those as the curriculum's floors, not as market data or a forecast of what you will earn. The coverage floor now has the force of underwriting rules: SOP 50 10 8.1 requires 1.25x for initial acquisitions, measured on the last fiscal year or the average of the last two, not on projections.
Why the bear case is the case that matters
In a self-funded deal, the IRR almost never decides the deal. Coverage and the bad year do.
Here is why. When you put in 10 percent equity and borrow the rest, any business that survives produces a spectacular IRR on paper, because the denominator is tiny. A 2021 survey of first-time self-funded searchers found they buy smaller companies with more debt than traditional search funds, and report higher returns (V1, 2021). The debt explains both. It multiplies the good years, and it multiplies the bad ones by exactly the same lever.
The difference is who absorbs the bad ones. A traditional search fund's investors lose their money if a deal fails. You lose yours, and then the personal guarantee reaches past the business into your savings and your house. That asymmetry decides how you read the model: the base case tells you whether you want the business. The bear case tells you whether you can survive owning it. Read them in that order of importance, not the order the spreadsheet prints them.
There is one more reason to distrust your own base case. A 2025 Yale analysis of 44 traditional search fund companies from one investor compared the margin forecasts made at acquisition with what happened: the forecasts expected margins to rise by about 6 percent of sales, margins actually fell by about 10 percent, and only 12 of the 44 companies improved them (M6, 2025). That is a traditional-search sample, not a self-funded one, but the lesson about forecasts transfers cleanly. Your base case should not assume you will run the business better than the founder did. If you do, that is upside you did not pay for.
Case
Coastal Comfort HVAC is the fictional company that runs through this curriculum. The bridge in lessons 6.2.1 and 6.6.1 took the broker's $1,710,000 of claimed adjusted EBITDA down to $1,017,000 of recurring adjusted EBITDA. That number is the engine of the model. Everything below scales off it, which is why the margin check in lesson 6.2.1 matters: this $1,017,000 implies a recurring margin almost four points above the prior year's, and the documents do not say why. The model is built on it anyway, because it is the best number the seller's documents support before the letter of intent; the quality of earnings report later moves it (lesson 7.3.1 shows what it found).
Here is a first-pass model at a purchase price of $3,600,000, financed the simplest way the SBA allows: a 10 percent equity injection and an SBA loan for the rest, no seller note. The interest rate, capital spending and tax figures are illustrative assumptions for teaching, not quotes. Most 7(a) loans carry a variable rate, so the 10.0 percent used here will move, and your lender quotes the real one.
Sources and uses
| Uses | Amount | Sources | Amount |
|---|---|---|---|
| Purchase price | $3,600,000 | Equity injection (10%) | $430,000 |
| Closing costs and SBA fees | $250,000 | SBA 7(a) loan (90%) | $3,870,000 |
| Working capital | $450,000 | ||
| Total project | $4,300,000 | Total | $4,300,000 |
Key assumptions (illustrative)
| Input | Value | Why |
|---|---|---|
| Year-0 EBITDA | $1,017,000 | Your recurring number, not the broker's |
| SBA loan | 10.0%, variable, 10-year amortization | About $614,000 a year of payments at that rate |
| Capital spending | $90,000 a year | Trucks and equipment wear out |
| Taxes | 25% of EBITDA less interest and capital spending | A placeholder; your CPA will refine it |
| Working capital | $450,000 | About five weeks of operating costs |
| Hold and exit | 5 years, sold at 3.75x EBITDA | The midpoint of the value range in lesson 6.4.1, and below the 4.2x the whole project cost you |
Notice what is already inside the $1,017,000: a $225,000 market cost for the general manager (salary, bonus and payroll taxes). If you are the general manager, that is your pay. Do not count it twice by treating the cash left after debt service as your paycheck.
Three scenarios
A single forecast is a guess. Three is honesty. Each scenario is a story about the business, and the percentages come from the story, not the other way around.
- Bull: the maintenance-plan base grows and pricing improves. EBITDA grows 6 percent a year.
- Base: the business runs about as it has. EBITDA grows 2 percent a year.
- Bear: Beachline, the homebuilder, goes quiet and the pulled-forward replacements leave a soft year behind them. Revenue falls 15 percent from its recurring level of about $5.6 million (fiscal 2024 revenue plus normal growth, which is where the business runs without the one-time project and the pulled-forward work). Because about 40 cents of each lost revenue dollar was contribution (revenue minus the materials and field labor that go with it), EBITDA falls by roughly $337,000, to about $680,000, and stays there.
| Bull | Base | Bear | |
|---|---|---|---|
| Year-1 EBITDA | $1,078,000 | $1,037,000 | $680,000 |
| Year-1 DSCR | 1.76x | 1.69x | 1.11x |
| Cash to you, year 1 (after debt, capital spending, tax) | about $221,000 | about $191,000 | about -$77,000 |
| IRR on $430,000 | about 83% | about 65% | negative |
| MOIC | about 10x | about 6.6x | money lost |
| Same deal with no debt (unlevered IRR) | about 22% | about 17% | about 3% |
Read that table the way the personal guarantee reads it.
The base case is not a forecast of 65 percent returns. It is what 10 percent equity does to arithmetic. The bottom row is the honest one: bought for all cash, this business earns about 17 percent a year at this price. The loan turns 17 percent into 65 percent. It does the same thing in reverse.
These year-1 figures run on projected earnings, so they are your stress test, not the lender's. The lender's test under SOP 50 10 8.1 must be met "using either the last fiscal year-end or an average of the last two fiscal year-end statements on either a historical or adjusted basis". On this first pass, FY2025's $1,017,000 against about $614,000 of payments is about 1.66x, and the average of the last two fiscal years, about $903,000, gives about 1.47x. Both clear the 1.25x floor.
The bear case is the deal. At 1.11x the business still covers its loan payments, so the bank is fine. You are not. After trucks and taxes you are writing checks of about $77,000 to $106,000 a year to keep the loan current. Over five years you put in about $885,000 counting your original equity, and the sale returns about $143,000. The deal survives. Your balance sheet does not do well.
Two break-even points come out of the same model, and they are worth more than any IRR:
- Payment break-even: EBITDA can fall to about $614,000, roughly 40 percent below today, before the business cannot make its SBA payment.
- Cash break-even: EBITDA can fall to about $783,000, roughly 23 percent below today, before you start funding the business from your own pocket.
What happens below the payment break-even, when a guaranteed SBA loan cannot be paid, is the subject of Part VIII; lesson 8.4.3 covers the bankruptcy chapters a failed business and its guarantor may face.
The sensitivity grid
A sensitivity grid is a table that reruns the model across a range of two inputs at once, so you can see where the deal breaks instead of where it looks good. This one shows year-1 DSCR at different purchase prices and different levels of EBITDA, financed the same way (90 percent SBA, 10 percent equity):
| Price | SBA loan | DSCR at $1,017,000 | DSCR at $810,000 | DSCR at $680,000 | Base-case IRR |
|---|---|---|---|---|---|
| $3,000,000 | $3,330,000 | 1.93x | 1.53x | 1.29x | about 91% |
| $3,600,000 | $3,870,000 | 1.66x | 1.32x | 1.11x | about 65% |
| $4,500,000 | $4,680,000 | 1.37x | 1.09x | 0.92x | about 33% |
| $5,250,000 | $5,355,000 | 1.20x | 0.95x | 0.80x | about 9% |
| $6,450,000 (the ask) | $6,435,000 | 1.00x | 0.79x | 0.67x | negative |
Two rows are there for the arithmetic only. The maximum for any one standard 7(a) loan is $5,000,000 (Y25), so the $5,250,000 and $6,450,000 rows, which need loans of $5,355,000 and $6,435,000, could not be financed this way at all.
Read the columns differently. The $1,017,000 column is the closest to the lender's view, because it is last fiscal year's earnings on an adjusted basis: any price where it shows under 1.25x (here $5,250,000 and $6,450,000) fails the SBA floor for an Initial Acquisition under SOP 50 10 8.1 and cannot be financed as structured. A lender may also test the two-year average or make its own adjustments, so treat the column as your estimate of the lender's number. The $810,000 and $680,000 columns are your stress test, not the lender's; the lender may not rely on projections, and you must.
At the $6,450,000 asking price, the broker's $1,710,000 would show a comfortable 1.68x. Your $1,017,000 shows 1.00x before anything goes wrong. Same deal, same loan. The only difference is whose earnings number you believed.
Look down the IRR column too: it clears this school's 25 percent floor all the way up to about $4,750,000. If IRR were your only test, you would pay about $1,100,000 more than the bear case can carry. That is the whole argument of this lesson in one column.
This is a first pass, not the deal. It treats all $430,000 of equity as yours; in the closed deal half of the equity comes from three minority investors, so read these returns as returns on total equity, not on your half. Lesson 6.5.2 turns it into a walk-away price, shows the letter-of-intent structure, where two seller notes, one of them on full standby, change the debt service this model has to carry, and re-runs the walk-away on the quality of earnings figure the case actually closes on.
Steps
You build this yourself, in any spreadsheet. A blank workbook and an afternoon are enough. Do not borrow a model you cannot explain cell by cell; you will be defending every assumption to a lender, and to yourself at month eleven.
1. Start from your number
The first input is your recurring adjusted EBITDA from lessons 6.2.1 and 6.3.1, never the broker's adjusted figure. If you have not finished the normalization, you are not ready to model. A model built on the seller's earnings is the seller's model.
2. Build sources and uses
Two columns that must balance. Uses: purchase price, closing costs and fees, and working capital. Sources: SBA loan, seller note, equity. Check the SBA rules as you go (SOP 50 10 8.1): at least 10 percent equity injection for an initial acquisition, limited sources such as full-standby seller debt or minority investor equity providing no more than half of it (all of it for an eligible veteran under the 2026 exception above), and no single standard 7(a) loan above $5,000,000. Then apply this school's own rule, which is stricter than the SBA's: total leverage that does not cross 90 percent of the project.
3. Build the debt schedule
One row per year for each loan: payment, interest, principal, ending balance. Your spreadsheet's payment function (PMT) gives the level payment from the amount, rate and term. A seller note goes in one of two ways. On full standby it has no payments of principal or interest for the whole term of the SBA loan, so it adds nothing to debt service; its interest accrues and the balance is still owed at exit. Paying, it gets its own row from the first payment to the last, and its payments count in coverage for every year they run.
4. Build the annual cash flow
For each year of the hold:
- EBITDA
- minus capital spending
- minus taxes (a placeholder rate until your CPA gives you a better one)
- minus total debt service, all loans
- equals cash to you
Then DSCR for each year: EBITDA divided by total debt service. Record the lowest year, not the average.
5. Build the exit
At the end of the hold: exit multiple times final-year EBITDA, minus every loan balance still outstanding, equals your equity at exit. Use an exit multiple no higher than the one you paid on the whole project (total project cost divided by recurring EBITDA). Multiple expansion is a hope, and you are not allowed to finance hope.
6. Calculate IRR and MOIC
Lay out one row of cash flows: your equity as a negative number in year 0, the cash to you in years 1 through 5, and the exit equity added to year 5. IRR is your spreadsheet's IRR function on that row. MOIC is everything you got back divided by everything you put in. If you had to put more cash in during a bad year, that counts as money in.
7. Write three stories, then three sets of numbers
Write the bull, base and bear cases as one paragraph each, about this business specifically: which customers, which channel, which employees, which external event. Then translate each story into growth and margin. Build the bear case from the deal's actual red flags (lesson 5.2.1), and use contribution margin, not a proportional cut. A 15 percent revenue drop with fixed overhead is a much larger EBITDA drop, and a model that shrinks every cost line in proportion flatters you.
Put the three scenarios behind a single selector cell (1 for bull, 2 for base, 3 for bear) so the whole model recalculates when you change it.
8. Build at least two sensitivity grids
- Price by EBITDA, showing DSCR, like the one above.
- Price by exit multiple, showing IRR.
Shade each cell green (meets the floor), yellow (close) and red (fails). The edge between green and red on the price axis is where the deal stops working. Lesson 6.5.2 turns that edge into a number.
9. Read the model in the right order
- Bear-case DSCR. Does the business still make its payments in the bad year, with a cushion? This curriculum's floor for the bear year is 1.10x (lesson 6.5.2). If it falls below 1.0x, stop.
- Bear-case cash to you. How much would you have to write checks for, and for how long? Compare it to the reserve you actually have.
- Base-case DSCR, every year, including any year a seller note's payments start or stop.
- Base-case IRR and MOIC. Only now.
A deal that only works in the bull case works in your imagination. A deal that pays its loans in the bear case, from your own number, is a deal you can take to the next gate.
You can sanity-check your coverage math against the site's SBA Deal Stress-Tester, but build your own model first. A tool can check your arithmetic. It cannot check your assumptions.
Deliverable
A one-page LBO summary for your deal, with:
- Sources and uses, balanced, with the equity injection percentage and total leverage stated.
- The three scenario stories, one paragraph each, and the growth and margin each one implies.
- A results table for all three scenarios: year-1 DSCR, lowest DSCR, cash to you in the worst year, IRR, MOIC, and the unlevered IRR.
- The payment break-even and cash break-even EBITDA.
- Two sensitivity grids, shaded.
- A one-word verdict with one sentence of reasoning: PURSUE, RESTRUCTURE or WALK.
Bring it to lesson 6.5.2, where the edge of the grid becomes the walk-away price.
This lesson is education, not investment, lending, tax or legal advice. The figures are illustrative and the company is fictional. Have a CPA and an SBA lender review the model for any business you intend to buy.