Carve-out valuation: the allocated cost trap CFOs miss

September 2, 2026 · 5 min · Free, with full transcript

Can you value a divested division from the financial statements the seller hands you? The position here is no. A carve-out is a business that never operated alone, and its statements allocate shared costs such as legal, IT and the CEO's salary by formula rather than showing what the unit costs to run. Examples include GE's healthcare unit and Dell's 2021 sale of Boomi. You come away able to test a carve-out price properly: strip every allocated cost line, build a standalone operating model function by function, check transition services agreement pricing and lock-in terms, budget 3 to 5 percent of revenue for one-time separation spend, and price dis-synergies before bidding.

Listen to the podcast

4 min

Chapters

Key takeaways

  • Throw out every allocated cost line in the carve-out statements and rebuild each function from zero, as if founding the company.
  • Compare seller transition services agreement fees against your own standalone cost, and check the lock-in term before signing.
  • Budget 3 to 5 percent of revenue for one-time separation costs such as new systems, rebranding and legal entity setup.
  • Price the dis-synergies that appear on day one rather than assuming synergy, the way Boomi's buyers funded full standalone infrastructure.
  • If the standalone cost number is more than 10 percent worse than the allocated one, reprice before you bid, not after close.
Read the full transcript

Host:Welcome to the Leaders Insights Podcast. Today's episode, carve out valuation and separation accounting. What CFOs consistently get wrong? Here's my provocation to start. Most CFOs think a carve out is just a smaller version of the parent company. You're telling me that assumption alone destroys deal value?

Expert:It destroys it before anyone signs. A carve out, where you sell off a division or a subsidiary rather than the whole company, isn't a shrunk down parent. It's a business that never actually existed on its own, and the financial statements you're handed are a fiction.

Host:A fiction? That's a strong word for audited numbers.

Expert:They're not lying. They're allocating. The parent spreads shared costs, or hunt. The legal team, the IT systems, the CEO's salary, across every division using some formula. When GE sold its healthcare unit, the carve out statements showed corporate overhead allocated by revenue. But allocation isn't reality. The real question is what that business actually costs to run a loan. And nobody in the room knows on day one.

Host:So the buyer looks at a clean EBITDA figure. Earnings before interest, taxes, depreciation and amortization, the standard profit proxy, and it's basically made up?

Expert:It's directionally made up, which is worse, because it looks precise. Here's the trap. The allocated overhead in the statements is almost never what standalone costs will be. I've seen carve outs where the parent allocated $18 million of shared services, and the true standalone cost to replicate those functions was $31 million. That $13 million gap is pure earnings you thought you were buying and aren't.

Host:Wait. Replicate. Why replicate? Can't the buyer just keep using the parent's systems?

Expert:For a while, yes. Through a transition services agreement, CEE, a temporary contract where the seller keeps running your payroll and IT for a fee. And that fee is the second trap. Sellers price these things to hurt. I watched a mid-market industrials deal, where the transition agreement ran 40% above what the buyer's own standalone cost would have been, and it locked them in for 24 months. The seller made more on the exit ramp than on the sale.

Host:So the consensus advice, just normalize the EBIT data, is what everyone gets wrong?

Expert:Normalize is the most dangerous word in carve out work. People add back the allocated overhead and drop in an estimate for standalone costs, and they estimate low because low makes the model work. The discipline nobody does is building a genuine standalone operating model from zero. Every function. What does HR cost when it's 80 people, not a shared department? What does your cyber insurance premium look like when you're no longer under the parent's policy? Those are real dollars.

Host:Give me the number that should scare a listener.

Expert:One-time separation costs. Standing up new systems, rebranding legal entity setup. A company that never existed needs a birth certificate. On a deal with roughly $500 million in revenue, expect 3 to 5 percent of revenue in one-time separation spend. That's $15 to $25 million that hits after close and never showed up in your valuation, multiple. Buyers routinely forget it entirely.

Host:Because the model was built on the fiction.

Expert:Right. And here's the part that's almost funny. CFOs will spend six weeks arguing over the discount rate to the second decimal place, and zero minutes asking whether the payroll system even transfers. Your precision engineering the wrong number.

Host:Let me push you. Isn't some of this just conservatism dressed up as insight? Deal still close and make money.

Expert:They do when the buyer prices the DIS synergies. The costs that appear the moment you cut the cord. The winners treat the carve-out as a build, not a purchase. When Dell sold Boomi in 2021, the buyers went in knowing they were funding a full stand-alone infrastructure, and they priced it. The losers assume synergy on day one and find DIS synergy for three years.

Host:So the one thing a listener should do tomorrow?

Expert:Take the carve-out statements you were handed and throw out every allocated cost line. Then rebuild each function from zero as if you're founding the company. Because you are. If the stand-alone number is more than 10 percent worse than the allocated one, and it usually is, you've been mispricing the deal. Reprise before you bid, not after you close.

Host:Rebuild from zero. That's the discipline. That's a wrap. Fresh CFO briefings drop daily at MBA-training.com.

Read and go deeper