Every business for sale arrives with an add-back schedule. The broker's package takes the earnings the business reported, adds back the owner's salary, a lawsuit, a truck, a sponsorship and a bad year, and prints an adjusted figure larger than anything the tax return has seen. That figure sets the asking price. The buyer's job, and later the lender's, is to find out how much of it survives.
An add-back is an expense taken out of reported earnings because a new owner will not carry it. A Quality of Earnings (QoE) analysis is the discipline of testing each one: is it real, is it over, and can you prove both? This article sets out the four categories, the evidence that survives a lender's read and a QoE analyst's, the items both reject and why, and a worked example where a seller's schedule meets its first serious reader.
The bridge runs in both directions
Normalized EBITDA is reported EBITDA plus the add-backs that hold up, less the deductions the seller left out.
The second half is the one sellers forget. A below-market salary for an owner who works long hours is a deduction: the replacement will cost more. Rent paid to the owner's holding company at a friendly rate is a deduction: the new lease will not be friendly. Non-recurring income, an insurance recovery or a one-off grant, comes out just as a one-time expense goes back in. An analyst who adds and never subtracts is not normalizing. They are advocating.
One more distinction. EBITDA leaves a market-rate salary in the owner's seat; seller's discretionary earnings (SDE) adds the owner's full compensation back, on the theory that a single owner-operator will take it. The categories below are the same on either basis. Only the owner compensation line changes.
The four categories
| Category | Typical items | Evidence that survives | Where the lender pushes back |
|---|---|---|---|
| Owner compensation and perks | Salary above a replacement rate, family on payroll above the work they do, personal vehicles, health and life premiums, travel, club dues | The payroll register and the owner's tax slips, a market salary reference for the role, the buyer's own salary plan, the expense detail in the general ledger | Only the excess over a replacement salary comes back, and the lender uses the greater of the market rate and what the buyer will draw. The full salary is an SDE add-back, not an EBITDA one |
| One-time | Legal settlements, M&A advisory fees, restructuring, relocation, a failed product launch, storm damage | The invoice, the settlement or engagement letter, the GL entry, and three years of statements showing the item does not repeat | "One-time" items that appear every year: legal fees, bad debt, severance, write-offs. Recurring is a pattern, not a label |
| Non-cash | Impairment, write-downs, loss on disposal of assets, unrealized foreign exchange, equity compensation | The fixed-asset register, the GL entry, the accountant's working papers | Inventory write-downs and bad-debt provisions in a business that has them every year are operating costs. A gain on disposal is a deduction |
| Discretionary | Charitable donations, sponsorships, the owner's conferences, a marketing program the owner chose, bonuses paid at the owner's whim | The invoices, the GL detail, and a credible answer to one question: does any revenue depend on this spend? | Marketing is rarely discretionary. A bonus the staff expect is compensation. Anything that keeps a customer stays in the cost base |
The first row is the one lenders argue about, and the test is replacement cost, not the owner's pay. A QoE analyst prices the role the owner fills, the general manager, the lead estimator, the one salesperson who knows every account, at what the market would charge, adds the employer's payroll costs on that salary, and adds back only the excess. A lender compares that figure with the salary the buyer has written into the projections, and the larger of the two stays in the cost base. Family members get the same test: a spouse who does the books part-time is replaced by a part-time bookkeeper, not by nobody. Perks come back only when the paper says they are personal. A vehicle with a mileage log is an add-back for the personal share. A vehicle with no log is an argument.
What evidence survives
Two readers will go through the schedule, the buyer's analyst and the lender's, and they look for the same things in the same order. First, the general ledger detail that ties the item to the statements: account, date, amount. An add-back that cannot be found in the ledger does not exist. Second, the document that created the cost: the invoice, the contract, the payroll record, the settlement. Third, where the add-back rests on a judgment, a third-party reference that supports it: a salary survey for the replacement rate, an appraisal or comparable leases for market rent. Management's explanation is the fourth item, and on its own it is a note, not evidence: it tells the analyst where to look.
A schedule that survives also shows what was considered and left out, so a reader comparing it with the broker's version can see that the gap is a decision rather than an omission.
The add-backs lenders reject, and why
- The whole owner salary, on an EBITDA basis. Someone will run the business, and that person's pay stays in the cost base, whether it is the buyer's draw or a manager's salary. Only the excess is an add-back.
- "One-time" items that recur. Bad debt, severance, legal fees, inventory write-offs and equipment repairs that show up every year are the cost of being in that business. A lender spreads three years side by side for this reason.
- Run-rate and pro forma items. A contract signed but not yet delivered, a cost cut planned but not made, a price increase announced for next year. These are forecasts. They belong on a separate run-rate line, never in the historical figure, and a lender that may not rely on projections to meet its coverage test reads them the same way.
- Personal expenses with no paper trail. Meals, travel and vehicle costs are an add-back for the share a log, a receipt or a card statement supports, and nothing for the rest.
- Capital spending added back as if it freed cash. When the owner expensed a new roof and the schedule adds it back, the add-back is fair, but the lender moves the cost to maintenance capex and takes it out of cash flow there. The add-back relocates the cost; it does not remove it.
- A schedule with no deductions. Below-market owner pay, related-party rent, deferred maintenance and non-recurring income are items a lender's own report is required to look for. A schedule that found none of them is read as incomplete, not clean.
On an SBA 7(a) acquisition the rejection is a rule rather than an opinion. Under SOP 50 10 8.1, Appendix 15, every adjustment the lender makes to the coverage ratio needs a written justification in the credit memorandum, and an adjustment without one "will be ineligible". On an Initial Acquisition or a Business Expansion at a Business Purchase Price of $3 million or more the lender must obtain its own Quality of Earnings, with a cash proof, performed for the lender and "not prepared by or for the seller", and must use its earnings in the coverage test. The report "must identify and document all add-backs and adjustments to the seller's reported earnings", naming non-recurring revenue or expenses, above- or below-market owner compensation, related-party transactions, deferred maintenance and cash-basis versus accrual differences. The rule is set out in our explainer on SBA 7(a) acquisitions after 1 October 2026. Below the threshold the report is optional. The questions are not.
A worked example
Every figure in this section is a worked example. A buyer is looking at a commercial HVAC service business with reported EBITDA of $640K. The broker's schedule lists eight add-backs totalling $566K and prints adjusted EBITDA of $1,206K:
- The owner's salary, $310K, added back in full
- The owner's spouse on payroll, $60K, added back in full
- A legal settlement with a former supplier, $45K
- A bad-debt write-off, $28K, labelled one-time
- Vehicles and travel, $22K
- Sponsorship of the owner's golf club, $12K
- Loss on disposal of a service truck, $9K
- A signed maintenance contract starting next quarter, $80K of annual EBITDA
The buyer's analyst reads the ledger. The owner is the general manager and lead estimator, a role the market fills at $150K including the employer's payroll costs, and the buyer's projections carry the same figure. The spouse does the bookkeeping, part-time work a bookkeeper would do for $25K. Bad debt appears in each of the last three years at between $20K and $30K, so it is not one-time. The mileage logs support $14K of personal vehicle use; $8K of travel has no documentation. The settlement, the sponsorship and the truck all tie to the ledger and the paper behind it. The maintenance contract has not produced a dollar yet. Then the deduction the schedule left out: the premises are rented from the owner's holding company at $96K a year, and two comparable leases put market rent at $132K.
+ owner salary above replacement ($310K less $150K) $160K
+ spouse payroll above a part-time bookkeeper ($60K less $25K) $35K
+ legal settlement, matter closed $45K
+ personal share of vehicles and travel, per the logs $14K
+ golf club sponsorship $12K
+ loss on disposal of a service truck $9K
− related-party rent below market ($132K less $96K) $36K
= Normalized EBITDA $879K
Left out: $28K of bad debt that recurs, $8K of travel nobody can document, and $80K of contract earnings that have not happened yet. The seller's $1,206K becomes $879K, a gap of $327K. Every dollar of that gap is priced at the deal's multiple, and on a 7(a) loan it is also the difference in the earnings the lender divides by the debt service. The contract is not lost: it is carried on a run-rate line and argued separately, not as a line in the history.
Educational guide, not advice. This guide is general information, not investment, legal, tax, accounting, lending or valuation advice. Rules, ratios, prices and worked figures are stated as of the date shown, are drawn from public sources and our own experience, and can be out of date or wrong; they are not a substitute for a lender's, regulator's or program's current rules. Verify any figure with its primary source and a qualified professional before acting on it.
How Acquiror AI does this
The Quality of Earnings workbench in Acquiror AI is where the schedule above is built, one adjustment at a time. Each adjustment carries a category, a metric (EBITDA or SDE), a recurrence (one-time, ongoing or run-rate) and a confidence level, and the evidence sits on the adjustment: a linked document, a general-ledger reference, an invoice or payroll record. A support status on each line says whether that evidence is sufficient. Adjustments are approved or rejected one by one, a run-rate item is carried separately from the normalized figure, and the period is locked. Locking makes the adjusted figure the authoritative one for Valuation, Financing and the investment memo, and the memo lists the adjustments in three groups, approved, run-rate and excluded, with an evidence appendix of the source documents and GL references behind every approved one, so a reader can reconcile it to the seller's schedule line by line. On a 7(a) Initial Acquisition or Business Expansion at a Business Purchase Price of $3 million or more, the cash proof is the lender's QoE firm's work; the workbench is where you build and lock the bridge that report will test. If you are choosing between building the schedule yourself and buying a per-deal report, Acquiror AI vs. per-deal QoE reports sets out where each one fits.
The takeaway
Add back only what the ledger and the paper support. Subtract what the seller left out. Keep forecasts out of the history, and write each decision down where the next reader can see it. The lender's analyst will do the same work on the same schedule; the only question is whether they confirm your figure or replace it. Every term in this article is defined in the glossary.