A quality of earnings report runs sixty pages, and most of it exists to support one exhibit: the bridge from reported EBITDA to adjusted EBITDA. That bridge works as a stack of arguments, each one asserting that some expense will not exist under your ownership, and each dollar it adds survives into the purchase price at the full multiple. Buying at 6x means a $200,000 add-back you accepted carelessly cost $1.2 million. Reading a bridge well means grading the arguments; the arithmetic is the easy part.
The short version
Start from the bottom of the bridge and work up, because the largest adjustments deserve the most time. Sort every add-back into its class: owner compensation, true one-timers, pro-forma adjustments, related-party repricing, personal expenses, and repairs that are really capex. Each class has a known failure mode and a known standard of proof. Grade each add-back against that standard, tie it to the general ledger line it came from, and rebuild the bridge with only the adjustments you accepted. Then read the rest of the report for the three sections that check the bridge from outside: proof of cash, revenue recognition, and customer concentration.
What the bridge actually is
The exhibit walks from reported net income up through interest, taxes, depreciation, and amortization to EBITDA, then through two layers of adjustment: management's proposed add-backs, and the QoE provider's position on each. Those are different things. Management's column is advocacy. The provider's column is a professional opinion with hedged language, and the hedges are load-bearing. An adjustment the provider "notes as presented by management" without concurrence was transcribed, and nobody verified it.
Read the bridge in both directions before grading anything. Top down tells you the story the seller wants told; bottom up, starting with the largest adjustments, tells you where the money actually is. In most bridges two or three line items carry the majority of the value, and those deserve the first hour and the hardest evidence, while a long tail of small adjustments can be graded in a batch.
Note the period the bridge covers and whether it matches the period the price is based on. Bridges built on a trailing twelve months that ended four months ago are describing a business you are not buying; ask for the stub period update, and expect the adjustments to be reapplied to it, not assumed forward.
Grading the add-backs, class by class
Owner compensation. The most common and usually the most legitimate: replace what the owner actually paid themselves with market compensation for the role. Check it in both directions. An owner taking $600,000 for a job that costs $250,000 to fill is a real add-back. An owner taking $40,000 and working eighty-hour weeks is a negative adjustment the bridge conveniently omits, because replacing them costs more than their salary.
True one-timers. A litigation settlement, a flood, a one-time regulatory penalty. The standard of proof is non-recurrence, and the test is the trailing general ledger: pull the same account across three or four years. A "one-time" legal expense that appears every year is a cost of doing business with a hopeful label.
Pro-forma adjustments. The most argumentative class: a signed price increase annualized forward, a lost customer removed from history, a new contract counted before it has shipped. These are not adjustments to what happened; they are forecasts wearing the bridge's credibility. Accept them only with their own evidence: the executed amendment, the renewal that survived at the new price, the purchase order. Run-rate is a story until a full cycle has confirmed it.
Related-party repricing. Rent paid to the owner's real estate entity, services bought from a family company, inventory from an affiliated supplier. The adjustment to market can cut either way, and the proof is a market quote, not an assertion. This class also tells you something the bridge does not: how much of the company's cost structure changes character the day the seller leaves.
Personal expenses. Vehicles, travel, family payroll, the boat. Individually small, collectively meaningful, and graded entirely on documentation. A schedule tied to specific GL entries is acceptable; a round-number estimate of "personal expenses" is a negotiation position.
Repairs that are really capex. The reverse direction: expenses pushed below the line to flatter EBITDA, or genuine recurring maintenance labeled one-time. A roof patched every year is not a one-time repair, and a fleet that needs replacement is a cost the bridge will not show you. This is where adjusted EBITDA and actual owner earnings part company.
The red flags that grade the whole bridge
Some findings are not about a single add-back; they characterize the document. Round numbers where the ledger would produce ragged ones. A miscellaneous or other adjustments line of any size. Adjustments that do not tie to a general ledger account. Revenue-side adjustments, which deserve roughly triple the scrutiny of expense-side ones. Provider language that presents without concurring. And a total adjustment stack that rebuilds a third or more of adjusted EBITDA, which does not make the deal wrong, but does mean the bridge is the deal, and the diligence budget should be spent accordingly.
Two structural patterns matter as much as any single flag. Adjustments that grow as the deal progresses, where a bridge presented at the letter of intent has expanded by the time the report is final, tell you the number is being managed toward a price. And adjustments that only appear in the most recent year, when the same conditions existed in prior years, are usually a normalization applied selectively to the period that matters most to valuation.
Check the direction of every adjustment too. A bridge with no downward adjustments at all is a statistical curiosity; real businesses have understated costs somewhere, whether it is deferred maintenance, an owner working below market, or a lease that reprices at renewal. A one-directional bridge is advocacy, and saying so plainly to the provider often produces the missing half.
Read the three sections that check the bridge
The bridge can be internally tidy and externally false, which is why the report's narrative sections matter. Proof of cash is the anchor: the provider ties reported revenue to actual bank deposits, and any material unreconciled gap outranks everything else in the report. Revenue recognition tells you whether the earnings are real but mistimed. Customer concentration tells you whether the earnings are real but fragile. A strong bridge over weak proof of cash is a well-argued description of income that may not exist.
Proof of cash deserves a paragraph of its own attention because of what it actually tests. The provider ties reported revenue and expenses to bank activity for the period, which catches revenue that was recorded but never collected, expenses paid outside the books, and periods where the ledger and the account diverge. An unreconciled gap raises a question about whether the financial statements describe the business at all.
Working capital gets analyzed in the same report and drives a separate negotiation, so read it while the bridge is in front of you. The monthly working capital trend, the seasonality inside it, and the provider's proposed target are the inputs to the peg, and every add-back you accept should be checked for whether it also moves that target.
A graded bridge that holds up in committee
The working product is the provider's bridge with two columns added: a grade for each adjustment and your accepted amount, which is often partial. Sum the accepted column and you have your adjusted EBITDA, which is the number your model should use, not theirs. Under the table, a sentence per rejected or haircut adjustment explaining why. Every line cites the exhibit, schedule, or GL account it came from, because the negotiation that follows will be conducted add-back by add-back, and the side that can cite sources faster usually wins the point.
Keep the grade sheet in a form that survives the deal. A tab per adjustment class, a row per adjustment with the provider's amount, your accepted amount, the grade, the evidence reviewed, and the exhibit or account it ties to. It becomes the working paper behind your model, the briefing document for your investment committee, and the first thing your lender's field examiner will want to see.
Share a version of it with the seller's side when the negotiation reaches specifics. Handing over a reasoned schedule with grades and citations converts a price argument into a document exchange, and sellers who can produce the missing evidence usually do so within days, which is faster than any amount of back and forth about the total.
Keep the report next to the grade sheet
Six weeks later the question will not be what your adjusted number was; it will be why you haircut the pro-forma pricing adjustment, and the answer lives in a renewal exhibit somewhere in the data room. Keep the QoE, its exhibits, and your graded bridge in the same folder, and settle every challenge from the source document itself. The grading is only as durable as its citations.
The same file earns its keep after close. Post-acquisition, the accepted adjustments become the assumptions in your operating plan: the owner compensation you actually have to pay, the one-time costs that should not recur, the pro-forma pricing you now have to deliver. Reading the graded bridge against actual results at six and twelve months is the cleanest test of whether diligence was right, and it is where the next deal's grading gets better.
The scope letter decides what the report can prove
Before grading a single add-back, read the engagement scope, which is usually summarized in the report's opening pages. Quality of earnings work is a consulting engagement, not an audit, and the provider performed exactly the procedures the scope letter bought: perhaps a full proof of cash, perhaps only a high-level tie-out; perhaps site visits and management interviews, perhaps a desktop review of management-prepared schedules. An adjustment supported by a procedure the provider actually performed carries different weight than one resting on a schedule management handed over. The scope also tells you what was excluded, and exclusions cluster where problems live: tax exposure, inventory existence, cutoff testing. When the deal turns on one of the excluded areas, buy the missing procedure before close instead of arguing with the report.
Ask for the databook alongside the report. Most providers deliver an Excel companion with the underlying monthly detail, the GL extracts, and the tie-outs behind each exhibit. The narrative report is written to be read; the databook is where the bridge can actually be verified, and a seller or provider reluctant to share it is telling you something about the bridge's foundations.
The bridge meets the working capital peg
Add-backs do not live alone; each one interacts with the net working capital target that the same deal is negotiating on a parallel track. When the bridge removes an expense from EBITDA, the history used to set the working capital peg may still contain the payable, the accrual, or the inventory behavior that expense produced. The classic double-count runs in the buyer's favor going one way and against the buyer the other: owner compensation added back to EBITDA while the owner's unpaid accrued bonus still sits in the historical working capital average, or a one-time inventory build excluded from earnings while it quietly raises the peg. Every accepted add-back should get a second look with one question attached: does this adjustment have a working capital shadow, and is it being counted consistently on both tracks? Deals leak six figures through that seam without anyone acting in bad faith.
Ask the provider directly how the two analyses were coordinated, since the same firm usually produces both. The useful question is whether any adjustment accepted in the bridge has a corresponding effect on the working capital target, and a good provider will have a schedule showing exactly that. When the answer is vague, build the check yourself for the largest few adjustments; it is arithmetic, not judgment.
The same seam runs through debt-like items. Deferred revenue, accrued bonuses, unpaid taxes, and customer deposits all sit near the boundary between working capital and debt, and where a seller's bridge normalizes the expense while the item stays out of the debt-like list, the buyer pays twice for the same thing.
Negotiate with the grades, not the total
The graded bridge is a negotiating instrument, and its power comes from disaggregation. Wholesale positions, such as accepting or rejecting the adjustment stack as a block, produce stalemates; grades produce structure. Adjustments you graded A flow into price without argument, and conceding them early buys credibility for the fights that matter. B-grade adjustments, the ones that are plausible but unproven, convert into deal structure: an earnout that pays if the pro-forma price increase survives its first renewal cycle, an escrow sized to the unproven portion, a purchase price adjustment tied to the metric the add-back asserts. C-grade adjustments come out, and the sentence that removes them names the missing evidence, because a citation is a request the seller can satisfy while an opinion is an insult they will argue with. Sellers meet this structure better than buyers expect: the seller's own advisors built the bridge knowing which arguments were reaches, and a buyer who prices the strong claims fairly while structuring around the weak ones is making the deal easier to say yes to, not harder.
One more discipline: keep your graded bridge dated and versioned. Diligence produces new evidence weekly, and an add-back graded C in week two becomes a B when the renewal letter arrives. The negotiation is a living reconciliation between two documents, theirs and yours, and the side whose document is current tends to set the agenda.
Adjusted EBITDA is not owner earnings
For buyers at the smaller end of the market, especially anyone underwriting with debt service that must be paid from the first month, the bridge's destination number needs one more translation. Adjusted EBITDA is a capital-structure-neutral convention; it is not the cash the business throws off to its owner. Between the two sit maintenance capital expenditure, working capital growth that scales with revenue, and the real cost of replacing whatever the seller personally did that the compensation adjustment normalized. Build a second, private bridge from adjusted EBITDA to owner earnings: subtract a normalized annual capex figure derived from the fixed asset ledger and the age of the equipment. Skip the seller's recent capex as a guide; owners suppress it before a sale as reliably as homeowners paint before listing. A business showing $1.5 million of adjusted EBITDA with $400,000 of true annual equipment replacement is a $1.1 million business for debt service purposes, and the lender's field examiner will eventually do this arithmetic whether or not the buyer did it first. The QoE's fixed asset and capex exhibits usually contain everything this second bridge needs; the report simply is not obligated to build it for you.
Debt service is the practical test. Take the accepted adjusted EBITDA, subtract normalized maintenance capital expenditure and the working capital that revenue growth consumes, then compare what remains to the annual debt service the deal structure implies. A deal that covers service on paper but not after those two subtractions is a deal that will feel tight from the first quarter, whatever the model said at signing.
This is also where seller financing and earnouts get evaluated honestly. Both are paid from the same cash the business produces, and stacking an earnout on top of full leverage against an adjusted number that has never been earned in cash is the structure that fails most often in small-company acquisitions.
Test the revenue before you trust the bridge
The adjustment stack works on expenses, but the number every add-back multiplies against is revenue, and revenue deserves its own quality pass before the bridge earns any trust. Sort the top line into its honest classes: contractually recurring revenue, reoccurring revenue that repeats out of habit with no contract behind it, and episodic project work, because a dollar of each carries a different multiple and sellers present all three as one number. Read the cutoff behavior around period ends, where a business being dressed for sale pulls shipments and invoices forward; a revenue spike in the final trailing months that outruns the cash collections beneath it is the classic tell, and the proof of cash section is where it shows. Where customers prepay, deferred revenue is work you will owe after close for cash the seller already spent, and it belongs in the price discussion as the liability it is. And connect concentration back to the bridge directly: a pro-forma adjustment that removes a lost customer's costs while the revenue section shows two more customers at similar size is not an adjustment, it is a preview.
Two databook requests make this section fast: revenue by customer by month for the trailing period, which turns concentration and cutoff from narrative into a sortable tab, and for any recurring revenue claim, the beginning-and-ending customer count with adds and losses, because retention arithmetic done on real cohorts either supports the recurring label or quietly retires it. A seller who cannot produce those two tabs is not necessarily hiding anything, but the buyer is now underwriting a story instead of a schedule, and price should know the difference.
Frequently asked questions
What is a quality of earnings report?
A diligence analysis, usually commissioned by a buyer or lender, that examines how reported earnings were produced: revenue recognition, one-time items, related parties, proof of cash, and working capital. Its centerpiece is the EBITDA bridge.
How is a QoE different from an audit?
An audit opines on compliance with accounting standards. A QoE asks how durable the earnings are for a buyer. Clean audits and poor earnings quality coexist all the time, because audits do not grade add-backs or concentration.
What are add-backs?
Adjustments that remove expenses the seller argues will not continue under new ownership: owner compensation above market, personal expenses, true one-timers, related-party repricing. Each dollar of add-back becomes several dollars of price at the purchase multiple.
How much in add-backs is too much?
There is no fixed line, but when adjustments rebuild a third or more of adjusted EBITDA, the bridge is the deal and deserves proportionate scrutiny. Composition matters more than the total.
Who pays for the QoE?
Usually the buyer, and lenders often require one. Sell-side QoEs are increasingly common; read them with the same grading discipline and close attention to what the scope letter excludes.