There is a number on the cover of every CIM that the management team believes is real. Adjusted EBITDA. Pro-forma for the bolt-ons, normalized for the one-time costs, run-rate for the synergies that are coming. It is the number the model multiplies. It is the number the whole exit hangs on.
Then a buyer’s quality of earnings team gets the data room, and the number starts to shrink.
A large and growing share of portfolio companies are stalling at exit for exactly this reason. The adjusted number does not survive contact with diligence. The bid comes in below the floor the firm modeled, or it does not come in at all. The company is not unsellable because the business is bad. It is unsellable because the EBITDA it is being sold on cannot be defended.
The macro backdrop makes this worse. Bain’s read of the market is that multiples are expected to stay broadly flat, which means firms cannot count on multiple expansion to bail out a soft number. GF Data and others have shown buyer scrutiny intensifying, with more deals repriced or pulled in diligence than in the easy-money years. When the multiple will not move in your favor, every dollar of EBITDA you cannot prove is a dollar of enterprise value you give back.
The four addbacks buyers routinely reject
Adjusted EBITDA is a legitimate concept. Buyers are not hostile to it. They reject specific addbacks for specific reasons, and the same four categories come up again and again.
One-time costs that turn out to recur
The classic addback is the genuinely one-time cost. A litigation settlement. A facility move. A restructuring charge. Add it back, because it will not happen again.
The problem is the costs that get labeled one-time and then show up the next year under a different name. The “one-time” rebrand that becomes an annual marketing refresh. The “one-time” consulting project that turns out to be three years of the same firm on the books. The severance that recurs because the business restructures every year.
A QoE team will pull the general ledger and look across periods. When a one-time cost appears in year one, year two, and year three, it is not one-time. It is operating expense wearing a costume. The addback gets rejected, and the credibility of every other addback drops with it.
The second-order damage is worse than the rejected dollars. Once a buyer catches one fake one-time cost, they stop trusting the label. Every other addback now gets the same forensic treatment, the diligence timeline stretches, and the management team spends the next three weeks defending a number instead of selling a business. The cost is the adjustment plus the confidence.
Run-rate synergies that were never realized
This is the most aggressive category and the first one buyers attack. The company acquired three businesses, projected cost synergies, and added the full run-rate benefit back to EBITDA as if it were already banked.
Sometimes the synergies are real and partially captured. Often they are a slide. The headcount was never reduced. The systems were never consolidated. The procurement savings live in a model, not in the actuals.
A buyer will ask one question that ends the conversation. Show me the realized savings in the financial statements. If the synergy is real, it has already moved a cost line down. If it has not moved a cost line, it is a forecast, and buyers do not pay today’s multiple on your forecast of your own performance.
Owner and management adjustments
Owner compensation normalization is standard and usually accepted. Above-market salary, a family member on payroll, personal expenses run through the business. Strip them out, normalize to a market rate, defend it with documentation.
Where it goes wrong is scope creep. The personal car becomes the personal car plus the country club plus travel that was half business and half not plus a “consulting fee” to a related party. Each one might be defensible alone. Stacked together with thin documentation, they read as a management team reaching for every dollar, and the buyer starts discounting the entire adjusted number on principle.
The test a buyer applies is simple. Can you produce the underlying transactions and a clear rationale for each adjustment, fast, without the CFO building a new spreadsheet to explain it? If the answer is no, the adjustment is exposed.
Pro-forma acquisitions that were never integrated
A company makes acquisitions during the hold and presents EBITDA pro-forma, as if the acquired businesses had been owned for the full period and fully integrated.
The pro-forma is legitimate when the integration actually happened and the combined entity can produce one consolidated, reconciled set of numbers. It collapses when the acquired companies still run on their own systems, their own charts of accounts, and their own definitions of revenue and margin.
When a buyer cannot get a single consolidated view across the acquired entities, the pro-forma EBITDA is a stack of numbers that were never tied together. This is the exact failure pattern I have written about in why QoE gaps come from data failures, not accounting failures. The accounting can be clean inside each entity and still produce a number that does not hold at the group level because the data was never integrated.
This is the addback PE firms should worry about most, because the buy-and-build strategy is the one most likely to produce it. Growth through acquisition is the story PE firms love to tell. The same story creates the diligence problem, because every acquisition that was bought but not integrated adds another set of definitions, another chart of accounts, and another reconciliation the buyer will ask you to perform live. The deal thesis that justified the multiple becomes the reason the multiple cannot be defended.
The pattern underneath all four
Look at what the buyer is actually doing in each case. They are not arguing about accounting policy. They are asking you to trace a number back to its source and prove it.
Show me the recurring cost across periods. Show me the synergy in the actuals. Show me the transactions behind the owner adjustment. Show me the consolidated view across the acquisitions.
Every one of those is a data lineage question. Can you take a number on the cover of the CIM, follow it back through your systems to the underlying transactions, and reconstruct it on demand in front of a skeptical team that is being paid to find the holes.
A defensible EBITDA is a data lineage problem before it is an accounting problem. The adjustment can be perfectly reasonable and still fail diligence because the company cannot produce the trail that proves it. I have seen this cost real money on real deals, where reasonable, fully justifiable addbacks got rejected purely because the documentation was a private spreadsheet nobody else could follow. That story is in how data problems cost one company half a turn on their multiple.
What a defensible number actually requires
The difference between an adjusted EBITDA that holds and one that gets cut is not the size of the adjustments. It is the trail behind them.
For each material addback, you need three things. The source transactions, accessible without a rebuild. A one-page rationale that someone other than the person who created it can follow. And consistency across periods, so the same definitions and the same logic apply in year one, year two, and year three.
For the business as a whole, you need numbers that reconcile across systems. Revenue in the CRM that ties to the GL. Retention calculated one way, not three. Acquired entities rolled up into a single consolidated view with one chart of accounts and one set of definitions. When two leaders are asked the same question, they give the same number.
This is the work that turns a contested adjusted number into a number a buyer can underwrite. It is also the work that QoE readiness depends on, which is why I treat QoE preparation as a data problem rather than a finance exercise. The finance team can produce a beautiful adjusted EBITDA bridge and still lose half of it in diligence if the data underneath cannot carry the weight.
Find out before the buyer does
The cruel part of the pro-forma trap is the timing. The number gets tested for the first time in live diligence, in front of the one audience you cannot afford to lose confidence in. By then the adjustments are public, the bid is anchored, and every rejection is a visible markdown.
The fix is to run diligence on yourself first. Take the adjusted EBITDA bridge and, line by line, ask whether you can defend each addback with source data, on demand, the way a QoE team will. The ones you cannot trace are the ones that will not survive. Either build the lineage now, while you have time, or take them out of the number before a buyer takes them out for you.
If you want a fast way to pressure-test how your data room would actually hold up under that scrutiny, work through the Data Room Survivor and see where the gaps are while you can still close them.
A defensible EBITDA is not the number you can calculate. It is the number you can prove. The companies that cannot sell are the ones that confused the two, and found out which one mattered only after the buyer’s team opened the data room.