Why a division is not a company
A company sale starts with an entity that already has its own accounts, its own contracts and its own staff. A carve-out starts with none of those and has to manufacture them while the business keeps trading.
That manufacturing work is where these transactions succeed or stall, and almost all of it is discoverable in advance. The failures happen when it is discovered instead by the buyer's advisers in week six of diligence, at which point every finding is a reason to reduce the price.
Mapping what is actually shared
Start by listing every dependency between the division and the parent. Four areas account for most of them.
Technology. Shared identity and authentication, shared billing, shared data pipelines, shared observability, a common cloud account, and any code that sits in a repository the division does not own outright. Enterprise software agreements negotiated at group scale rarely transfer, and the divested business will be re-pricing them as a much smaller customer.
People. Employees who split their time across divisions, shared functions such as customer success or quality assurance, and any individual who holds knowledge that exists nowhere else.
Contracts. Customer agreements signed by a group entity rather than the division. Master agreements that bundle products staying behind. Supplier contracts with minimum commitments that assume group volume.
Facilities and general. Space, insurance, and the dozens of small vendor relationships that nobody has catalogued because each one is individually trivial.
The four degrees of entanglement
Technology separation is usually the long pole, and it helps to be honest about which of four situations you are in.
Fully separate, meaning its own repository, own infrastructure account and own team. Rare, and worth a great deal, because it lets financial sponsors bid rather than only strategic acquirers who can absorb complexity.
Separate product on shared platform services. The application is distinct but calls group-owned services. Workable through a transition agreement plus a defined replacement plan.
Shared repository, where code is physically intermingled. Separation is an engineering project that has to be scoped and costed before launch, not during.
Shared team, which is the hardest of the four because it is not a technical problem. It is a people problem with a technical consequence.
Building the standalone cost structure
Buyers discount heavily where they cannot see a credible standalone cost base, and they build their own version regardless. Three numbers matter.
Incremental cost, meaning what the division will spend to replace what the parent provided. Work bottom up, function by function, and cost each line at what a company of that size would actually pay. Benefits pricing is the classic trap: a two hundred person company does not buy healthcare at the rate a twelve thousand person group does, and the gap is real money.
One-off separation cost, covering data migration, system reconfiguration, rebranding and legal restructuring. Enterprise system separation alone commonly runs twelve to eighteen months and a seven figure sum on an asset of this size.
Stranded cost, meaning group cost that was serving the division and does not disappear when the division does. This one is a board conversation rather than a buyer conversation, and it needs to happen before launch. A transaction that looks accretive on the headline price can look different once stranded cost is honestly quantified, and discovering that after signing is a poor way to discover it.
A working sequence
Weeks one to four, dependency mapping and perimeter definition. Weeks five to eight, standalone financial modelling and cost allocation. Weeks nine to twelve, transition services term sheet. Weeks thirteen to sixteen, day one readiness planning and testing.
The order matters. Building a financial model before the perimeter is settled means rebuilding it, and negotiating transition services before you know the cost base means negotiating blind.
Protecting the business while you do it
The largest risk in a carve-out is that the separation process itself degrades what is being sold.
Keep the working group small and under a confidentiality agreement. Most leaks during preparation come from uncontrolled internal circulation of documents rather than from buyers.
Identify the people who have to stay for the asset to be worth what you think it is worth. Usually it is a small number: the division lead, one or two engineers holding critical knowledge, and whoever owns the largest customer relationships. Agree retention arrangements before the process reaches a stage where disclosure is unavoidable. Buyers price management continuity directly, and a departure announced mid-process is close to fatal.
Keep product development running. A roadmap that visibly stalls during a process tells every bidder that the team has mentally left, and they price accordingly.
What good preparation buys
Two things, and both show up in the price.
It removes the largest single cause of re-trading after the letter of intent, because the questions that would otherwise surface in diligence have already been answered in writing.
And it widens the buyer field. A separation plan that is already costed lets a financial sponsor underwrite the deal. Without one, only a strategic acquirer who can absorb the complexity can bid, and a field of one buyer type rarely produces a competitive price.
Editorial Team · Published for orientation, not as advice on a specific transaction. Any figure cited is orientation, not a valuation. See market notes.