Quality of Earnings: The Number You Are Buying Is Not the Number Reported

Reported, Adjusted and Sustainable Are Three Different Numbers

A multiple is applied to a number. Everything about the transaction — the headline, the debt quantum a lender will support, the equity cheque, the return the buyer has underwritten — is a function of that single figure, and the figure itself is presented as though it were an observation. It is not. Reported EBITDA is the output of accounting policy choices. Adjusted EBITDA is the output of an argument about which of the reported costs should be treated as not belonging to the business a buyer is acquiring. Only the third number — the earnings the business can be expected to generate on a repeatable basis under the ownership that follows completion — is a valuation input, and it is the one nobody circulates.

The gap between the three is not marginal. A vendor pack that presents reported EBITDA and then walks the reader through a schedule of add-backs to arrive at an adjusted figure is doing something entirely legitimate; the exercise exists because reported earnings genuinely do contain costs that will not recur. But the schedule is advocacy, and it is prepared by the party whose consideration rises with it. Each line on it converts a cost into value at the transaction multiple. At eight times, an add-back of one is worth eight of price. That arithmetic is the reason the schedule exists, and the reason it is worth testing line by line rather than in aggregate.

This article sets out how the earnings basis is actually examined: the taxonomy of add-backs and what distinguishes a defensible one from a contested one, why items described as one-off are the most fought-over category on the schedule, how revenue recognition and cut-off can move earnings without any add-back appearing at all, why cash conversion is the most reliable single test available, and what changes when the target is a carve-out rather than a standalone business. It is the companion to the enterprise-to-equity bridge: the bridge decides what happens to an agreed value, and this decides what the value was calculated from. Where our transaction advisory practice acts for a buyer, this work precedes the bridge, because a definitions schedule negotiated against an unsustainable earnings figure is a careful allocation of the wrong number.

Every Add-Back Belongs to a Category, and the Category Decides the Argument

Add-backs are not a single class of item to be accepted or resisted as a block. They divide into categories that are argued on entirely different grounds, and the fastest way to lose a defensible adjustment is to bury it in a schedule alongside an indefensible one. In the mandates our advisers run, the first thing done to a vendor's adjustment schedule is to sort it — genuine one-offs, owner compensation normalisation, pro-forma synergies, run-rate adjustments, capitalisation policy items, and related-party pricing — because each of those sits at a different point on the scale from arithmetic to assertion.

The distinction that matters most is between an adjustment that removes something which happened and an adjustment that adds something which has not. Removing a settled legal claim that will not recur is a statement about the past, and it can be evidenced. Adding the earnings the business would have made had a cost reduction been implemented twelve months earlier is a statement about a counterfactual. Both may be reasonable. They are not equally reliable, and a schedule which presents them in the same column, in the same typeface, with the same tone of inevitability, has flattened a distinction the buyer is entitled to see.

A One-Off That Happens Every Year Is a Cost of Doing Business

The most contested items on any adjustment schedule are the ones described as exceptional, non-recurring or one-time. The description is doing the work: an item accepted as one-off is removed from earnings and paid for at the multiple, and the label alone is frequently the only support offered. The test is not whether the specific event will happen again — no particular legal claim recurs — but whether the category of cost is a feature of how the business operates.

This is why the analysis has to run across a series rather than a single year. A business that settled an employment claim in one year, a supplier dispute in the next and a customer claim in the third has not had three one-offs. It has a litigation cost line that it has chosen not to describe as one. A business that rebranded once has had a one-off; a business that rebranded, then refreshed the brand two years later, then relaunched the website has a periodic marketing investment cycle. Our financial due diligence team runs proposed one-offs back across every period for which records exist, not only the periods presented in the vendor pack, because the schedule is drawn from a window the seller chose.

Restructuring costs are the sharpest case. A single restructuring in a decade is genuinely exceptional. A restructuring charge in four consecutive years is not exceptional in any meaningful sense — it is what the business spends to stay at its current level of profitability, and removing it presents earnings the business has never actually delivered. The same logic applies to items that are individually small enough to escape scrutiny. A schedule of fifteen adjustments, none of which is large, can move the earnings basis further than any single line a buyer would have challenged, which is why the aggregate matters as much as the composition.

There is a second, quieter version of the problem. Some costs are genuinely non-recurring in the form they took but will recur in another form: a systems implementation completed this year avoids a systems implementation next year, but a business operating aging infrastructure will implement something again within the investment cycle. Treating the whole charge as an add-back assumes a cadence of capital investment that never returns. Where the underlying need is periodic, the more defensible treatment is to normalise the cost across its cycle rather than to remove it entirely — and our advisers put that position in writing, because it is far easier to argue for a partial adjustment before the seller has anchored on the full amount than after.

Revenue Can Move Earnings Without Any Add-Back Appearing at All

Everything above concerns costs. The larger risk to an earnings basis frequently sits above them, in when revenue was recognised and in which period it landed, because a revenue timing question moves EBITDA without appearing anywhere on an adjustment schedule. Nothing has been added back. The number is simply not describing the period it claims to describe.

Cut-off is the first test and the least glamorous. Revenue recognised days before a period end, on shipments that left the warehouse days after it, moves earnings between periods with no entry that looks unusual. The examination is mechanical: trace a sample of transactions around each period end back to delivery evidence, and check whether the pattern of sales in the final week of a period differs from the pattern in the equivalent week of the preceding quarter. Our financial due diligence team runs the cut-off test on the periods that flank the measurement date, in both directions, because a period that has been pulled forward is followed by one that has been drained.

Channel loading — shipping product to distributors ahead of end-market demand — produces the same effect over a longer window and is visible in the relationship between shipments and sell-through. Where distributor inventory is rising while end-market demand is flat, revenue has been recognised that the market has not yet absorbed, and the subsequent period carries the correction. Bill-and-hold arrangements, where goods are invoiced but remain with the seller, raise the same question in a form the accounting standards address directly: the conditions under which control has transferred are specific, and whether they have been met is a matter of evidence rather than intention.

Percentage-of-completion accounting concentrates the judgement further. Where revenue is recognised as a contract progresses, the measure of progress is an estimate, and a modest change in the estimated cost to complete moves both revenue and margin in the current period. The estimate is not wrong because it is an estimate — long contracts cannot be accounted for any other way — but it is a judgement whose sensitivity should be understood by anyone applying a multiple to the resulting earnings. Where our valuation team is asked to value a contracting business, the sensitivity of reported margin to the cost-to-complete assumption is tested before the multiple is discussed, because the two are not independent.

Reseller and agency arrangements pose a different question with the same consequence: not when revenue was recognised but whether it should have been recognised gross or net. A business acting as agent records the commission; a business acting as principal records the gross amount and the corresponding cost. The distinction turns on who controls the good or service before transfer, and it does not change EBITDA in absolute terms — but it changes revenue, it changes margin percentage, and where a valuation has been argued on a revenue multiple or benchmarked against peers on margin, it changes the conclusion.

Cash Conversion Is the Test That Does Not Depend on Anyone's Judgement

Every question above involves a judgement that can be argued. Whether a cost recurs, whether a benchmark is right, whether progress on a contract has been measured correctly — each is contestable, and each is contested. Cash conversion is different in kind. Cash either arrived or it did not, and the relationship between reported earnings and the cash the business actually generated is the most reliable single indicator of earnings quality available to a buyer.

The comparison is straightforward: operating cash flow before interest and tax, set against EBITDA, across every period presented. A business converting a high and stable proportion of its earnings into cash is producing earnings that behave like earnings. A business whose conversion is falling, or is persistently low, is producing earnings that are being absorbed somewhere — into receivables that are not being collected, into inventory that is not moving, into capitalised costs that reduce the expense line without reducing the cash outflow. None of those is necessarily improper. All of them mean the earnings figure and the cash generation of the business are telling different stories, and only one of them pays down acquisition debt.

Working capital is where the divergence usually lives, and it connects this article directly to the bridge. A business that has improved reported profitability while lengthening receivable days has converted a collection problem into an earnings presentation. A business that has extended payment terms with suppliers to release cash before a sale has borrowed from the buyer, who will have to normalise the position afterwards — and the working capital peg is where that argument is eventually settled, at completion, on a balance sheet nobody has yet seen. Our financial due diligence team reads the earnings analysis and the working capital analysis as one exercise for that reason: an add-back schedule and a peg derived from the same distorted months will agree with each other and both be wrong.

The other reason cash conversion earns its place is that it is difficult to present selectively. An adjustment schedule can be composed. A trend in cash conversion across three or four years is a property of the accounts as a whole, and where it diverges from the earnings trend, the divergence is the finding. In the mandates our advisers run, a deteriorating conversion trend alongside a rising adjusted EBITDA is the single pattern most likely to change the shape of an engagement — not because it proves anything on its own, but because it identifies precisely where the remaining diligence hours should be spent.

A Divisional P&L Describes a Business That Has Never Existed

Where the target is a division rather than a company, the earnings question changes shape entirely. The financial information presented is not the historical performance of a business; it is an allocation, prepared by the seller, of a larger entity's results. Nothing in it is necessarily wrong, and all of it is a set of choices about what to attribute. The buyer is not acquiring the division as presented. They are acquiring a standalone business that has never operated on its own, and the earnings basis has to be normalised to what that business will cost to run.

The costs at issue are the ones the division never bore. Group functions — finance, legal, human resources, information technology, insurance, treasury, procurement — are typically recharged to a division on an allocation basis that reflects internal policy rather than the cost of buying the service. A finance function delivered by a shared service centre carries a per-unit cost no standalone business of that size can replicate. Insurance purchased on group terms prices differently from a standalone policy. Procurement conducted at group volumes achieves terms the division alone will not. Each of those is a real cost the buyer inherits and the presented earnings do not show.

Transitional service arrangements make the timing visible without solving the problem. Under a transitional services agreement the seller continues to provide certain functions for a period after completion, at an agreed charge — which means the buyer sees a cost, but not the cost of the steady state. The charge is negotiated, has a defined end, and is frequently set below the standalone cost as part of the commercial package. An earnings basis built on the transitional charge describes a temporary arrangement. The number that should carry the multiple is the cost of the function once the arrangement has expired, plus the one-off cost of standing it up — which sits outside earnings but firmly inside the buyer's funding requirement. On carve-out mandates our transaction advisory practice builds the standalone cost base from the function list outward rather than by adjusting the seller's allocation, because an allocation reviewed line by line tends to inherit the omissions in it.

There is a symmetrical error in the other direction, and buyers make it as often as sellers exploit it. Divisions frequently carry allocated group overhead they would not incur standing alone — a share of head office costs that will not follow the business, corporate charges attached to activities the division does not use. Those are proper add-backs, and a buyer who resists the whole standalone conversation on principle will have overpaid the seller nothing and understated the asset. The exercise is a genuine reconstruction in both directions: what will this business actually cost to run, on its own, once the seller has gone. Our valuation team prices the function list against external reference points rather than against the allocation, since the allocation is the artefact being tested.

Grade the Add-Backs and Watch the Price Follow

The instrument below is seeded with the illustrative schedule above. Each proposed add-back carries a defensibility grade — accepted, contested or rejected — and only accepted items are included in the earnings the multiple is applied to. Change a grade, change a value, or change the multiple, and the implied purchase price moves with it.

The readout worth watching is not the price but the price at risk: what the contested lines are worth at the current multiple. That figure is the value still in play on the earnings basis alone, before any commercial negotiation has taken place — and it is usually larger than the parties expect, because a single unit of earnings is never worth a single unit of price.

The Earnings That Survive Testing Are the Ones Worth Paying For

None of the work described here requires an adversarial posture, and very little of it requires access a buyer does not already have. It requires reading the adjustment schedule as a set of propositions rather than as a calculation, sorting those propositions by the kind of evidence that would support each one, and then asking for that evidence. An item that can be evidenced is settled quickly and stays on the schedule. An item that cannot is usually withdrawn rather than defended, and the withdrawal happens sooner where the test was put in writing early. Our transaction advisory practice therefore scopes the earnings analysis before the price discussion rather than alongside it, since the sequence determines which party is responding to whom.

For a seller the same discipline pays in the opposite direction. A schedule where every line has been tested internally before it is circulated — where the one-offs have been run back across every available period, the run-rate items carry their evidence, and the synergy lines were never in the earnings column to begin with — is a schedule that survives diligence largely intact. A schedule assembled to maximise the presented figure invites a review that questions all of it, including the adjustments that were entirely defensible. The defensible items are the casualties of the indefensible ones, and that is a self-inflicted cost. These positions are drawn from work delivered by a network of 80+ specialist consultants across 22 countries, which is the only reason they are offered as observations about the mechanism rather than as one adviser's preference.

What survives the review is the number the multiple should have been applied to. Everything after that — the definitions schedule, the debt-like items, the working capital peg examined in the companion article on the enterprise-to-equity bridge — allocates a value that this number produced. And several years later, if the acquisition was priced on earnings that did not hold, the impairment test will re-mark the same assumptions against what actually happened, which is the subject of the third article in this series. The instrument above will show, in a few minutes, what a given set of grades does to price. It is worth running on the seller's own schedule before it is circulated, and on the buyer's reading of it before a position is taken — which is available to anyone willing to do it.