InvestorsFree
EBITDA Normalization and Add-Back Review
Send the financials and the seller's adjustment schedule, and get every add-back tested for evidence, overlap and period, with the supported figure.
River treats the seller's adjustment schedule as an argument rather than a total, and tests every line on it against what the financials and the data room actually show. In the worked example Ardsley Mechanical, a commercial HVAC contractor, reported $4,180,000 of EBITDA on $52,400,000 of revenue and proposed twelve adjustments taking it to $7,334,000. Working through the lines left $5,910,500 a buyer could support. At the 6.5 times the price was built on, that $1,423,500 of earnings is $9,252,750 of purchase price.
A standard add-back schedule asks one question of each line: is there a document behind it. That question is worth asking, and it found $449,500 at Ardsley. It is also less than a third of what was wrong. The schedule format never asks whether another line already removes the same dollar, which was $352,000, or how much of an annualised saving sits inside the period being adjusted at all, which was $304,000. Downward adjustments nobody had written down were another $318,000.
Written for the deal team reading a schedule their banker did not build, for the independent sponsor whose lender will test the same lines against the credit agreement, and for a seller who would rather find the soft items first. The supported figure is what belongs in the investment memo, and the evidence behind each line comes out of the data room review. Owner-specific items usually need their own pass first, which is the owner draw review.
The best-documented line on the schedule was the most wrong
Everybody tells sellers to document the schedule, and the advice is right but it is not the binding constraint. Documentation proves an expense happened and that somebody decided it will not happen again. It cannot prove the dollar is only counted once, and it cannot prove the amount belongs to the period you are adjusting. Those are arithmetic properties of the schedule as a whole, invisible from inside any single line, and they are where the largest corrections live in practice. Perfectly evidenced lines fail both tests routinely.
Ardsley's biggest line was the Rockford branch it closed on 30 September 2025, entered as $420,000 of annualised savings with the closure notice, the lease termination and the final payroll attached. Two problems. A supervisor whose $132,000 sat inside that branch cost moved to the main yard and appears again on the headcount line, so the real annual saving is $288,000. And the branch was open for three of the window's twelve months, making the cost removed $288,000 times three twelfths, or $72,000. That line moved $2,262,000 of price with a clean file.
Two rules public filers work to settle both arguments, and a lender's credit committee tends to borrow them. Item 10(e) of Regulation S-K bars smoothing an item called non-recurring when a similar charge fell in the prior two years. That is how Ardsley's legal fees and bad debt were cut against their own three-year history. Article 11 limits a cost-saving adjustment to its effect on the historical statements as if the saving existed at the start of that period, which is the branch calculation exactly.
How it works
Send the financials
Statements, monthlies, the trial balance and payroll for the period. Say where the window starts and ends.
Add the schedule
The seller's proposed adjustments, however they arrived, plus the multiple and figure the price uses.
Read the three tests
Each line comes back with a supported amount, the test that cut it, and the price effect.
Work the concession list
Ask what a line is worth conceding, or rerun the bridge at a different multiple or window.
What you get
- Every proposed line tested three ways, for evidence, for overlap with another line, and for period
- A bridge from reported to the supported figure that closes, with each test's contribution named
- The recurrence check run against three years, so an item called one-time is checked for being annual
- Annualised run-rate savings recut to the months of cost genuinely inside the measurement window
- Downward adjustments the schedule omitted, priced, including accruals the cash books never carried
- Each line's correction translated into purchase price at the multiple the deal is built on
- A separate list of the lines you would concede, with what conceding each one costs
Common questions
Is this a quality of earnings report?
No, and it should not be presented as one. A QoE is signed work by an accounting firm that carries a name and a liability. This is the analysis you want before you commission one, so the scope you buy targets the lines that move price. Reading the report when it lands is the other half.
What is the overlap test finding that a normal review misses?
The same dollar removed on two lines. At Ardsley, owner compensation was normalised from total cash pay of $1,150,000 down to a market replacement, and a separate line added back $220,000 of owner and spouse bonuses that were already inside that $1,150,000. Both lines had documents. Together they removed the bonus twice.
Why cut an annualised saving that genuinely happened?
Because the add-back removes cost that is sitting in the period you are measuring, and only that much of it is there. A position cut two months before the cut-off leaves ten months of its cost in the trailing twelve, so ten months is the adjustment. Annualising the full year counts two months of savings that no historical period ever paid.
Does it work on the seller's side?
That is one of the better uses. The buyer's accountants will rebuild the schedule whether you hand them one or not, and a line struck in diligence costs more than a line never claimed, because it makes the reviewer reread the ones you kept. Running your own schedule against the three tests tells you which items to drop before anybody asks.
What if the seller will not give us the general ledger?
Then the run says which lines cannot be tested rather than assuming them away. Owner comp usually survives on the payroll register alone. Related-party rent needs the lease and a comparable. Pro-forma savings need the termination or closure date, and without a date the period test cannot be run, which is itself worth stating in writing.
How does the price translation work?
Every corrected dollar is multiplied by the multiple the price is built on, because that is the mechanism the deal uses. At Ardsley the schedule was worth 6.5 times, so the $348,000 struck from one branch line was $2,262,000 of price. Lower the multiple and the arithmetic still holds, it just pays less.
Can it check the debt and working capital side too?
It flags where an add-back is really a balance sheet item, which is the common double count: a cost added back to earnings and then also excluded from the working capital peg. Headroom against a lender's leverage test is a separate exercise, and the covenant headroom forecast runs it on the supported figure.
EBITDA Normalization and Add-Back Review
Fill in the form and your workspace opens with the work already underway.