redactedEvery dollar of adjusted EBITDA carries the deal multiple on its back. Accept a $100,000 add-back on a deal priced at five times earnings, and the price moves half a million dollars; strike it, and it moves the same distance the other way. That leverage is why the add-back schedule draws so much of the negotiation, and why the word itself causes most of the trouble. "Add-back" sounds like a list of bonuses the seller collects for having run expenses through the company. It's really one piece of a bigger exercise called normalizing EBITDA, and the point of this issue is to lay out how that exercise actually works: what holds, what falls, and the logic that decides it. redactedLet’s start with what a buyer is actually paying for. It’s not profit from the seller's last tax return, and not the growth from the buyer's plans for the business either. A buyer pays for an earnings stream: what the business should produce over the next twelve months, run the way the seller runs it today, under the seller's model and cost structure, cleaned up so the numbers tell the truth. That's what adjusted EBITDA, normalized EBITDA, whatever the deal happens to call it, is trying to be. Nobody expects the next twelve months to land exactly on that number, and every buyer makes changes the seller wouldn't have made: new investment, different priorities, strategic moves. That's the buyer's upside (or risk), and the seller isn't paid for it. The seller is paid for the economic stream they created as they run it. Every add-back, and every adjustment running the other way, is an argument about the size and validity of that stream. Adjustments make the business model honest. They don't make it different. The classic add-backs work because they're truth corrections to the current model. The portion of the owner's pay above what a hired manager would earn comes back to earnings. Personal spending run through the business comes back. Rent the owner pays their own LLC above market comes back. But the same logic giveth and taketh. Remove the owner's compensation and benefits, and a replacement manager goes into the model at a market salary with a fair benefits package, because nobody runs the business for free. A family member doing real work isn't an add-back at all; the role stays, benchmarked at what it would cost to fill. Even the vehicle resolves this way: either the role needs it or the model carries mileage instead. redactedThe cleanest illustration I know is gifts, because the same line splits two directions. The diamond tennis bracelet for a spouse is an “add back” to earnings; it leaves with the owner. The gold watch every twenty-five-year retiree receives stays in the model, and I've struck schedules that tried it the other way, because cutting what a team has come to expect isn't normalizing the business, it's modeling a worse one. Employee morale isn't free, and a buyer knows it. Recurring means recurring, even on a five-year cycle. The everyday costs of producing the revenue stay: equipment that keeps being replaced, marketing, commissions, the accountant used every year. The harder calls are the big expenses that land every four or five years without being capitalized, a machine rebuild, a roof, that kind of cycle. Infrequent is not the same as one-time. If the business will face that cost again on the buyer's watch, it belongs in the stream, usually spread across the years it covers. Costs and the revenue they produce travel together. An expense that helped generate revenue can't leave the model while the revenue stays. Either both remain or both come out. And pro forma adjustments to revenue, earnings the current model hasn't produced yet, rarely survive; I struck them as a rule. That's the seller asking to be paid today for work the buyer will do tomorrow. Classify first, then prove. Every adjustment that survives passes two gates. First it has to fit the model: a cost that truly won't recur under the seller's business as it runs today. Then it has to be supportable: a document that proves the expense, a plain reason it's personal or one-time, and a benchmark behind any number, the salary and the rent especially. An adjustment that fits the logic but has no file behind it is a claim, and buyers don't pay for claims. Reasonable people can disagree. Adjustments to EBITDA aren't always black and white. What one buyer accepts, another strikes; a strategic buyer, a search fund, and a family office can read the same schedule three ways, and the same item can land differently deal to deal. The schedule is a starting position, not a settlement. Which is why the most useful work happens on the sell side before the negotiation: setting the seller's expectations against what actually tends to hold, so the price built on the schedule is a price that survives it. The practical takeaway is the unglamorous one. Above-market pay, above-market rent: fine, visible, easy to normalize, buyers see it weekly. The noise is what hurts. I've seen months of grocery bills, family Christmas gifts, a kid's college tuition, all run through a business heading to market. Every one of those is a conversation, a document request, and a little less benefit of the doubt. The most useful nudge an advisor can give an owner a year or two out is often the simplest: reduce the noise. Coming up next: the same normalizing exercise, applied to the balance sheet. Working capital, the cash and inventory a buyer expects to come with the business, and the true-up that lands months after everyone thought the deal was done. Ryan Anoskey, CPA, CFO Partner, LIMESTONE.