EBITDA add-backs: which ones hold up, and the record that shows it
By Derwin Lucas ·
EBITDA add-backs are expenses added back to reported EBITDA because they are not expected to recur, or are not part of the business being valued or lent against. The result is adjusted, or normalized, EBITDA, the basis a valuation multiple is applied to. A leverage covenant is tested on EBITDA as the credit agreement defines it, which can differ. In a home-services roll-up, many add-backs arrive with acquisitions: owner compensation above what the role would cost to fill, one-time legal settlements, deal costs, and an owner’s personal expenses run through the business.
Whether an add-back is legitimate is rarely the hard question. The hard question is what supports it, and for which figure. One add-back can be accepted by management, excluded from governed Adjusted EBITDA and not permitted under the credit agreement, all at once and all correctly. Each of those answers is what a quality-of-earnings review will ask about. This guide works through one twelve-month add-back register from Goverisyn’s demonstration portfolio, with synthetic data, and the classification that separates what holds up from what doesn’t.
Adjusted EBITDA add-backs: one register, three answers
Exhibit 1. Add-backs by kind, sum of brands, twelve months to June 2026, $ thousands (demonstration portfolio, synthetic data)
| Kind | Count | Accepted by management | Governed | Covenant basis |
|---|---|---|---|---|
| Unusual or nonrecurring charges | 2 | 24 | 24 | 24 |
| Litigation settlements | 3 | 36 | 36 | 25 |
| Permitted acquisition costs | 1 | 38 | 38 | 38 |
| Owner compensation | 4 | 76 | – | – |
| All add-backs | 10 | 174 | 98 | 87 |
Governed counts only the add-backs a named reviewer classified supported nonrecurring: the evidence establishes the amount and that it does not recur. The covenant column applies the credit agreement’s own terms, here a synthetic agreement that caps litigation settlements at $25k a test period and does not permit owner compensation. The covenant-basis figure, computed on the agreement’s terms, also adds $79k for one acquired brand’s months before acquisition, which is not an add-back and sits outside this register.
Each column is correct on its own definition. The $174k management accepts, the $98k that is governed and the $87k the agreement permits answer three different questions. The $76k between management’s figure and the governed figure is owner compensation, all of it. A board pack that shows one of these figures without saying which invites the wrong comparison.
What every EBITDA add-back has to show
1That it happened, and is what it says it is
The expense traces to the general ledger and to the invoice, payroll record or settlement agreement behind it. Most add-back schedules pass this test, and it settles the least. Reconciling an amount to the ledger shows where it came from. It doesn’t show that it qualifies.
2That it will not recur
A settlement can be nonrecurring. A brand that settles a dispute every year has a recurring cost of doing business. The evidence is the history: the brand’s prior years, and whether the same kind of charge appears again.
3That the amount is right
This is where owner compensation sits. The add-back is the owner’s pay less what the role would cost to fill, and that second number is a judgment someone has to make and someone else will test.
Run-rate savings are a different kind of adjustment. They are not past costs that won’t recur; they are forecasts, and they need their own support. In the demonstration, $90k of claimed run-rate savings is left out of the covenant-basis figure because no officer’s certificate, which the agreement requires, is held in the record.
Owner compensation add-backs: the one that rests on judgment
Owner compensation is among the most common add-backs in an acquired, founder-run contractor, and among the first a quality-of-earnings review tests. In the demonstration register, four owner compensation add-backs total $76k. Three of them, $54k, are classified supported with judgment: the support exists, but the amount depends on an assumption a different reviewer could set differently. The fourth, $22k, is unsupported.
Supported with judgment and unsupported are different findings, and they shouldn’t be reported as one. A buyer’s quality-of-earnings team will test the first and has nothing to test in the second. At an illustrative 8.0× multiple, up to $432k of enterprise value rests on a judgment a buyer may set differently, and $176k rests on nothing.
None of this makes management’s $174k wrong. It is management’s figure, on management’s definition. What matters is that each add-back carries its classification, so that the figure everyone else will use can be stated beside management’s.The record behind governed Adjusted EBITDA
An add-back holds up only as well as the support recorded when it was taken.
Classify each add-back when it is taken, not at diligence
Add-back schedules are usually assembled in the months before a refinancing or a sale, from spreadsheets, email and the memory of whoever is still there. By then the brand that took the add-back may have changed accounting systems, and the controller who accepted the benchmark may have left. The reviewer’s question, why this amount and on what support, arrives long after the person who could answer it.
In Goverisyn, each add-back enters the register when it is taken, with its kind, amount and support, and a named reviewer classifies it against criteria the client approved. Governed Adjusted EBITDA counts only what is classified supported nonrecurring. An add-back still awaiting a decision is left out of governed Adjusted EBITDA and shown as an exception with an owner until it is classified. The classification is the reviewer’s; Goverisyn records it, and doesn’t make it.
What this guide does not claim
It does not say that owner compensation is never a valid add-back, or that any platform’s adjusted EBITDA is overstated. Whether an add-back is permitted is a matter of the credit agreement’s own definition, and real agreements vary. Whether a buyer accepts it is a matter for the buyer’s own review. Goverisyn declares neither. The figures come from a demonstration portfolio with synthetic data, and the credit agreement is synthetic.
Start with the largest add-backs
Start with the add-backs that are largest relative to each brand’s EBITDA, and put the three tests to each one in order. Where the answer to the third depends on an assumption, record the assumption, who accepted it and when. That one line is what a reviewer asks for first.
The questions a lender or buyer asks of the whole figure are set out in Adjusted EBITDA diligence questions for home-services roll-ups, and how revised add-backs show up as growth in EBITDA growth since acquisition. Goverisyn’s Portfolio Measurement Review tests a documented basis across selected brands, after a free 45-minute Portfolio Review.
More guides on financials
EBITDA comparability across portfolio companies
EBITDA growth since acquisition: separating performance from changes in basis
Adjusted EBITDA diligence questions for home-services roll-ups
Covenant EBITDA vs adjusted EBITDA: what the credit agreement counts
Sell-side quality of earnings: what to have on record before the review begins
Start with a Portfolio Review.
45 minutes. No data needed in advance.
By Derwin Lucas, founder of Goverisyn, with a career in management and information systems in home services. Published . About Goverisyn