Strip out the adjustments and six times leverage becomes seven.

What is an EBITDA add-back?

It is an amount added back to reported earnings on the argument that it will not exist for the next owner. A founder's above-market salary, a one-off legal settlement, the rent on a building the owner also owns: each is real cash that left the business and, on the argument, will not leave it again.

Every middle-market business has some. Owner-operated companies are run for tax and for the owner's life, not for a buyer's model, and a reported profit figure that has not been adjusted for that is not a useful number either. The exercise is legitimate. What is contested is where it stops.

The reason it is contested is that the add-back schedule is where the entire negotiation over price quietly sits. The multiple gets argued in a meeting. The number the multiple is applied to gets argued line by line in a diligence process, by people with more time and better information than the seller, and every dollar removed from the schedule is removed at the multiple, not at par.

How large has the adjustment layer become?

Large enough that the global financial regulator now measures it. The Financial Stability Board reports that private credit borrowers run leverage of five to six times debt to EBITDA against roughly four times in the leveraged loan market, and that with EBITDA adjustments stripped out, true leverage could be closer to seven times (Financial Stability Board, Report on Vulnerabilities in Private Credit, 6 May 2026). Around seventy-five percent of those borrowers have EBITDA below $100 million, which is to say this is a description of the middle market and not of large-cap credit.

Read that as an implied measurement of the adjustment layer itself. If reported leverage is five to six turns and unadjusted leverage is close to seven, the adjustments are carrying something in the order of a full turn to a turn and a half of the capital structure. That is not a rounding convention. It is a material share of the earnings base on which both the price and the covenants are set.

The documentation confirms the direction. The loan market is still described as broadly borrower-friendly through the second quarter of 2026, with expansive EBITDA adjustments among the features that have persisted (LSTA, Loan Market Covenant Trends 2Q26, 24 July 2026). Definitions have loosened over a decade, and the schedule a seller presents in 2026 is longer than the one a seller presented in 2016 for the same business.

The other side of that has arrived at the same time. Ninety-eight percent of surveyed private credit lenders report that their underwriting standards became notably stricter since the start of 2026, and the share expecting looser documentation fell from thirty-three percent to four percent in a year while the share expecting tighter documentation rose from thirteen percent to fifty-six percent (Houlihan Lokey, Q2 2026 Private Credit Survey). Permissive definitions in the market at large do not mean a permissive reading of your schedule.

Which add-backs survive diligence?

The ones a third party can verify without taking the seller's word for anything. That is the whole test, and it is narrower than most schedules assume. A surviving add-back has a document behind it, a date, a counterparty, and a reason it cannot recur that does not depend on the seller's intention.

Owner compensation above a market rate is the cleanest example, because the replacement cost is observable. It survives when the schedule shows the actual compensation, a sourced market rate for the role, and the difference, rather than a single net number. The same is true of related-party rent, where a third-party appraisal or a comparable lease establishes the market figure, and of discretionary personal expenses that appear in the general ledger with a description.

One-off professional fees survive when they attach to a named, closed event: a completed piece of litigation, a financing that closed, an acquisition that happened. They stop surviving the moment the same line appears in three consecutive years, because a cost that recurs annually is a cost of the business regardless of what caused it each time. A buyer reading three years of monthlies will find that pattern before the seller mentions it, which is the argument for finding it first.

The category that survives least often, and is claimed most often, is the pro forma adjustment: the run-rate effect of a contract signed after the period end, the annualized benefit of a cost programme that is partly implemented, the earnings of a location that opened in month ten. These are not add-backs in the sense the other categories are. They are forecasts, and a buyer will treat them as forecasts, which means underwriting them at a discount or moving them into the contingent part of the consideration.

What does a disallowed add-back actually cost?

The add-back multiplied by whatever the business trades at, which makes it the highest-leverage arithmetic in the entire process. A $400,000 adjustment removed from the schedule is not a $400,000 problem. At six times it is $2.4 million of enterprise value, at eight times $3.2 million, and it is deducted from the equity cheque at the end rather than shared with the debt.

This site does not print a current middle-market multiple by size band, because no current published series exists: the most recent public table by enterprise value band covers the first nine months of 2025. The point survives the uncertainty. Across any multiple a middle-market business plausibly clears at, a disallowed add-back costs several times its own value, which is why an hour spent documenting one is worth more than a day spent negotiating the multiple.

There is a second cost that does not appear in the price. Adjusted EBITDA is also the denominator in the covenants. Leverage tests, coverage tests and the capacity in every basket are measured against a defined EBITDA, and a definition that carries generous adjustments at signing produces headroom that evaporates if the adjustments stop being available. The middle-market borrower population currently sits at median gross leverage of 6.1 times and median interest coverage of 1.6 times (KBRA, Q2 2026 Middle Market Compendium, LTM ended 30 June 2026, published 28 July 2026). At 1.6 times coverage, a turn of phantom EBITDA is the difference between headroom and a conversation with a lender.

The same report carries the reason this matters more in 2026 than it did in 2024. Median EBITDA growth across that population fell to twenty-four percent from twenty-seven percent, the largest quarter-over-quarter decline on record. Growth has been doing the deleveraging work for three years. Where it slows, the adjustment layer is what stands between a reported ratio and a real one, and lenders know it.

How should the schedule be built?

From thirty-six months of clean, normalized monthly financial statements, prepared before a process starts rather than in response to a buyer's questions. That is the stated preparation standard for this market, alongside commissioning a quality of earnings report early; companies that closed successfully were described as having well-prepared financial packages that minimized re-trading risk (Capstone Partners, Capital Markets Update, 4 June 2026).

Three rules make the difference between a schedule that holds and one that gets rewritten. Document before you claim: every line needs the underlying invoice, contract, appraisal or payroll record attached, and an item without one should come off the schedule voluntarily rather than under pressure. Separate the categories: verifiable normalizations belong in one section and forward-looking adjustments in another, clearly labelled, because mixing them invites a buyer to treat all of them as the weakest kind. And test the recurrence honestly: if the same category appears in each of the last three years, it recurs, whatever the individual causes were.

The reason to do this before rather than during is that a re-trade is not a negotiation. Diligence is running one to three months longer than it did for the firms already seeing timelines extend, and seventy-three percent of senior investment bank executives expect it to get more complex still (SRS Acquiom and Mergermarket, published 23 February 2026). A schedule discovered to be soft in month four of a process is discovered when the seller has no alternative buyer, and the price change that follows is not proportional to the error.

The honest summary is that an add-back schedule is a credibility document that happens to contain numbers. Buyers and lenders are already assuming that a portion of the earnings figure is a claim rather than a fact, because the regulator's own arithmetic tells them so. A seller who has removed the weak items themselves is arguing from a different position than one who is defending all of them.