The number on the sell-side teaser is never the number a PE buyer underwrites to. By the time a sponsor closes a deal, "EBITDA" has been pulled apart, scrubbed, and reassembled into something a banker has to defend line by line. The gap between reported and adjusted is where leverage capacity is set, where covenants live, and where deals get won or lost on diligence.
This post walks through what adjusted EBITDA actually means in PE diligence — the categories of adjustments, how sponsors triage real from aspirational, and what Hilton's $1.2B of LTM EBITDA looked like under that lens when Blackstone closed at the peak of the 2007 cycle.
What adjusted EBITDA actually is
Reported EBITDA is what the company files: operating income before interest, taxes, depreciation, and amortization, per the company's accounting policies. It's GAAP-adjacent. It is what it is.
Adjusted EBITDA is what the diligence team thinks the business would have earned in a normal year, after stripping out non-recurring items, normalizing for timing, and pro-forma-ing in changes management says are already locked in. It's a forward-looking estimate of run-rate earnings power, dressed up as a historical number.
PE buyers care about adjusted EBITDA for three concrete reasons:
- Leverage sizing. Lenders set debt capacity off a defined "Bank EBITDA" — a credit-agreement-level definition that controls headroom under covenants.
- Valuation math. Entry multiple equals enterprise value divided by adjusted EBITDA. The cleaner the addback case, the lower the effective multiple at the same purchase price.
- Underwriting model. Year 1 forecast usually starts from LTM adjusted, plus management's case. If LTM is overstated, every downstream year is too.
The three categories of addbacks
Most addback schedules split into three buckets, in roughly descending order of how much a lender will credit:
Non-recurring items
One-time costs (or revenues) that won't repeat. Litigation settlements, restructuring charges, severance from a discrete workforce action, environmental remediation, M&A transaction expenses. Lenders generally accept these when they're documentable as discrete events and modest in dollar terms.
Run-rate adjustments
Costs or revenues that LTM doesn't fully reflect because of timing. A facility closure mid-year — annualize the savings. A new customer signed in Q3 — annualize the revenue. A headcount reduction that's only partway through the LTM period — pull forward the run-rate.
These are the gray area. They're forward-looking even though they're presented as historical adjustments. Aggressive buy-side teams stretch them; conservative lenders haircut them.
Management adjustments
The most aspirational category. Pro forma synergies from a planned acquisition. Cost savings from an operational program that hasn't started. "Public company costs" addback for going private, or "private company costs" addback for going public. Owner compensation normalization for founder-led businesses.
This is where the diligence fight happens. A clean Quality of Earnings (QofE) report from the buy-side accountant categorizes and dollar-quantifies every addback so each side can argue them line by line.
Why the category split matters for leverage
Lenders care about the split because they're underwriting to a debt service coverage ratio. If buy-side Adjusted EBITDA is $100 but Bank EBITDA — excluding the most aspirational addbacks — is $80, debt sized off $100 is the wrong number. Actual coverage will be 80% of what the buy-side model says.
When you size debt off optimistic EBITDA, you set covenant levels you can't sustain. That's the same dynamic that broke a generation of retail and consumer LBOs — sized to a number that didn't materialize. The free debt capacity calculator walks through how the binding constraint shifts from the leverage cap to the coverage covenant as you stress EBITDA down. Run a deal against it before you sign the LOI, not after.
Hilton 2007: when the headline multiple was already rich
Blackstone closed Hilton on October 24, 2007 at a $26.0B enterprise value on $1.2B of LTM EBITDA — a 21.7× multiple, financed with 8.4× leverage, $5.6B of sponsor equity, and $0.3B of cash on the balance sheet at close.
That headline multiple was peak cycle. Hotel REIT comparables were trading materially below it. The sell-side justification leaned hard on brand value, franchise-fee growth, and the planned operational program under Blackstone ownership.
The interesting diligence question wasn't whether 21.7× was high — everyone knew it was. It was whether $1.2B of LTM EBITDA was a defensible base to multiply against. A reasonable buy-side team would have worked through questions like:
| Underwriting question | What it does to LTM EBITDA |
|---|---|
| Are hotel demand levels at a cyclical peak? | Normalizes LTM toward through-cycle if yes |
| Is the franchise-fee revenue mix being underweighted in the multiple? | Lifts the effective base if franchise economics deserve a premium |
| What's the run-rate of recent property additions and openings? | Annualizes for full-year contribution |
| Which planned cost programs has the lender already agreed to credit? | Splits Bank EBITDA from buy-side adjusted |
The 21.7× headline was on reported LTM. Strip out cycle elevation and the effective multiple was higher; credit the planned operational program and the effective multiple was lower. Both stories appeared in the marketing book. The eventual outcome settled which story the diligence team should have believed.
The 2008–09 hotel-demand collapse cratered EBITDA. By 2010, Blackstone had restructured roughly $4B of the debt stack — the move that kept the deal alive long enough to be worth restructuring. It paid off: the December 2013 IPO at a $32B enterprise value started a multi-stage sell-down that totaled $14B in proceeds by 2018, producing a 2.5× MOIC and an 18.7% IRR over a 6.2-year weighted hold.
The lesson on adjusted EBITDA isn't that Blackstone's diligence got it wrong — it's that when you're paying a peak multiple, the quality of the EBITDA base matters more than the multiple itself. A 21.7× multiple on through-cycle earnings is a different deal than 21.7× on peak earnings. The first survives a downturn; the second needs a restructuring. (For how the resulting MOIC and IRR break down across leverage, multiple expansion, and operating improvement, see our post on MOIC vs IRR.)
Three diligence traps in addback math
Trap 1: Double-counting cost programs against run-rate adjustments
A sponsor models a planned cost-out program worth $20 in year 1. They also annualize $8 of mid-year headcount reductions that already happened. Often those overlap — the headcount reductions ARE part of the cost program. Counting both inflates LTM adjusted and year 1 forecast.
The diligence test: any planned cost-out that's already started gets credited to LTM adjustments OR forecast, never both. If you can't draw the bridge cleanly from one to the other, you're double-counting.
Trap 2: Treating one-off revenue gains as recurring
A discrete government contract that closed in Q4 of LTM, a one-time license fee, an insurance settlement booked above the line. These are non-recurring on the cost side; they're frequently non-recurring on the revenue side too, and the addback should reduce EBITDA. The reflex is to scrub costs and ignore revenue — that's wrong.
A senior IC will sometimes flag this and ask why the EBITDA bridge doesn't show a downward adjustment. "Conservatively presented" is the right answer, and the only one that survives the lender's QofE review.
Trap 3: Asymmetric public/private cost addbacks
Take-privates routinely add back public-company costs — board fees, SOX compliance, investor relations, public-filing audit, D&O insurance. Take-publics rarely subtract the same number on the way out, even though the cost is real both directions. If you address it in one transaction but not its mirror, your numbers don't compare apples to apples across the hold period.
This trap also shows up in capital structure work — the S&U at close looks different if you're adding $10–15 of recurring cost vs. removing it. (For the full mechanic on how transaction expenses and run-rate cost changes flow into the close, our sources & uses walkthrough is the companion piece.)
What the IC asks before signing off
When an investment committee reviews the EBITDA bridge, the questions are predictable:
- Are addbacks categorized — non-recurring, run-rate, management — and what's the dollar in each bucket?
- Of the management adjustments, which has the lender agreed to credit in the Bank EBITDA definition?
- What's the LTM EBITDA at the 25th-percentile addback case vs. the 75th-percentile?
- How does year 1 forecast compare to LTM adjusted? Is there a step-up that needs explaining?
- Does the returns model clear hurdles under the conservative EBITDA case, or only under the aggressive one?
If the deal only clears the IRR threshold under the optimistic adjusted EBITDA, you don't have a deal — you have wishful thinking.
What adjusted EBITDA should look like in your model
Most LBO templates plug a single Adjusted EBITDA cell at the top of the LBO tab and run forward. That's how junior associates get caught: the number gets baked into purchase price, debt sizing, and covenant headroom — but there's no second view that strips the management adjustments to see whether the conservative case still works.
A real diligence-grade model has:
- Three EBITDA columns side-by-side — reported, lender (Bank EBITDA), and buy-side adjusted
- A visible bridge from reported to adjusted with every addback line item separately broken out
- Covenant headroom calculated off the lender column, not the buy-side one
- Returns sensitivity tables that use the conservative column under a downside case
The All-in-One PE Model builds the LBO sheet around exactly that structure — reported, lender, and buy-side EBITDA columns with addback line items, plus the covenant headroom and returns sensitivities flowing from the correct definition. See the sheet preview before buying.
Reported EBITDA gets you to a teaser. Adjusted EBITDA gets you to a credible offer. Lender-credited EBITDA gets you to a financing commitment. If your model only knows one of those numbers, you only know part of the deal.
Sources:
- Hilton S-1 Filing (Dec 2013), SEC EDGAR — entry multiple, LTM EBITDA, leverage and equity at close
- Bloomberg, "How Blackstone Made One of the Greatest Deals in PE History," May 2018 — total proceeds, IPO enterprise value, and the 2010 debt restructuring
- Blackstone Q4 2018 Earnings Release, IR.Blackstone.com — final exit MOIC and IRR