Transitional Services Agreements: The IT Trap

A transitional services agreement is negotiated by lawyers and lived with by the CIO, and the two are rarely in the same room. The document is usually a schedule at the back of the sale and purchase agreement, agreed late, priced roughly, and drafted by people who will never have to operate it. Twelve months later, the carved-out business is still on the seller's ERP, still paying for it, and the exit date in the schedule has become a negotiation rather than a deadline.

Every month of TSA overrun is a cost the deal model did not carry. This guide covers what a carve-out actually is, why IT makes it hard, what a transitional services agreement covers and quietly does not, what separation costs, and what to establish before anyone signs.

What is a carve-out, and how it differs from a spin-off

A carve-out is the sale of part of a business - a division, a product line, a geography - to a buyer, where the sold entity has never existed as a standalone company. It is separated from a parent that still exists and still operates around the space it left.

The distinction that matters is against a spin-off. A spin-off distributes a business to existing shareholders as a new independent company; there is no third-party buyer and no change of control. A carve-out sells the business to someone else, usually a corporate acquirer or a private equity house. Both require separation. Only the carve-out requires separation against a deadline set by a buyer who has paid for the asset and would like to run it.

There is a third case worth naming because it is often mislabelled: an equity carve-out, where a parent floats a minority stake in a subsidiary. The operational separation problem is similar. The urgency is not.

Why the definition affects the IT problem

The reason this matters operationally rather than semantically is entanglement. A business that has never been standalone has never needed its own general ledger, its own identity provider, its own licences or its own security perimeter. It has been running on the parent's, and nobody has been keeping a list. The carve-out is the moment that bill arrives.

Why IT is the thing that makes a carve-out hard

Nothing in the systems knows the business is being sold. That is not a joke; it is the whole problem.

A typical mid-market carve-out finds a shared ERP instance where the divested entity is a set of company codes rather than a separate installation, so it cannot simply be lifted. It finds a single Active Directory or Entra tenant containing everyone, staying and going, with group memberships built up over a decade. It finds software licences that are non-transferable by contract, meaning the carved-out business must buy its own and the seller cannot lawfully lend them. It finds a data estate with no entity boundary at all, where customer, contract and employee records for both businesses sit in the same tables with no reliable flag to separate them.

Each of those is solvable. What makes a carve-out in M&A hard is that they are solvable slowly, and the deal has already closed.

What a transitional services agreement actually covers, and what it quietly does not

The TSA exists to bridge the gap between completion and standalone operation. The seller continues to provide services - IT, finance, HR, payroll, sometimes facilities - to the carved-out business for a defined period, at a defined price, while the buyer builds its own.

What a good TSA covers is specific: the services in scope, the service levels attached to each, the charges and how they escalate, the notice required to terminate a service early, the overall term, and the exit arrangements. What most TSAs actually contain is a list of service names with a monthly fee and a date.

The gap between those two is where the money goes. "The seller will provide IT support" is not a service definition. It does not say what happens at 3am on a Tuesday when the ERP is down and the seller's on-call engineer, who now works for a company with no commercial interest in the buyer's month-end, has to decide whose incident to take first. It does not say whether change requests are in scope, which is the thing the carved-out business will need most and the thing the seller has least incentive to provide. It does not say who pays when the seller's own upgrade breaks the buyer's integration.

Why TSAs get priced by lawyers and paid for by the CIO

TSA charges are usually set as an allocation of the seller's cost base, sometimes with a modest uplift. That looks reasonable in the SPA and is frequently wrong in both directions. It understates the seller's true cost of running a service for a business it no longer owns, which breeds resentment and slow service. And it understates the buyer's true cost, because the charge covers the service and not the parallel cost of building the replacement, recruiting the team to run it, and dual-running both for the handover.

The deal team sees one number. The CIO carries three.

The exit clause nobody reads until month fourteen

The provision that decides how a TSA ends is the one that gets least attention when it is drafted. Two questions are worth more than the rest of the schedule combined. Can the buyer exit individual services early, without penalty, as each replacement lands? And what happens if the term expires and the buyer is not ready?

If the answer to the first is no, the buyer pays for a full bundle until the slowest workstream finishes. If the answer to the second is an automatic extension at a punitive rate, the seller has every incentive to be unhelpful and none to be quick. Extension terms should be agreed at the outset, when both parties still want the deal to complete, rather than at month fourteen when one of them has all the leverage.

How long a TSA should run, and why yours will overrun

The honest answer for a mid-market carve-out with a shared ERP is twelve to eighteen months. Deal teams routinely agree six to nine, because a shorter TSA looks like a cleaner deal and nobody in the room owns the delivery.

It overruns for three reasons, and they are the same three every time. The ERP separation is discovered to be a migration rather than a copy, because the divested entity's data was never bounded. The carved-out business does not have the people to run its own IT, because it never had its own IT function, and hiring one takes months the plan did not allocate. And the seller's team, whose bonus depends on the parent's performance and not the buyer's, deprioritises the work in a way that is entirely rational and completely unfixable by escalation.

Plan for the honest number. A TSA that ends early is a good outcome. A TSA that has to be renegotiated is a bad one, and it is negotiated from a position of weakness.

What an IT carve-out costs

Four costs, and most deal models carry one and a half of them.

• Stand-up cost. New tenant, new ERP or a genuine instance split, new licences bought at list rather than inherited at the parent's enterprise rate, new security tooling, new network. This is the number everyone budgets, and it is usually the smallest of the four.
• TSA fees. Monthly, for longer than planned, with an escalation clause.
• Dual-running. Both estates live simultaneously through cutover, which is not a fortnight. This is routinely omitted entirely.
• Capability cost. The carved-out business needs an IT function it has never had. That is recruitment, salary, and a period of being run by people who do not yet know the systems. The last of these has no line in any model I have seen.

The pattern is consistent: the visible cost is the small one, and the invisible costs compound with time. Which means every month of TSA overrun is expensive in three places at once.

Carve-out due diligence: what to establish before signing

Standard technology diligence asks whether the target's systems are fit for purpose. Carve-out due diligence has to ask something different and harder: can this thing actually be separated, how long does that take, and what does it cost. Separability is a distinct question from quality, and it is rarely on the list.

Four things are worth establishing before signing rather than discovering afterwards. Whether the divested entity's data has a clean boundary in the ERP, or whether separation means migration. Whether the licences transfer, which is a contract question with a large price tag attached. Which people who operate the systems are transferring with the business and which are staying with the seller. And whether the seller has any operational interest in the TSA succeeding, or is simply contractually obliged to tolerate it.

Our guidance on technology due diligence in private equity M&A covers the wider diligence discipline, and the separability questions above sit on top of it rather than instead of it. For a carve-out in private equity specifically, the answers feed straight into the value creation plan, because a business spending eighteen months on separation is not spending them on growth. Our guide to private equity value creation covers what that trade-off does to a hold period.

Separating a business with the Embedded Change Model™

Day-one separation is a people problem wearing a migration plan's clothes. The people who run the systems on day one work for the seller. They know where everything is, why it was built that way, and which undocumented job runs at 2am. Their employer has just sold the business they support, and their own future sits with the parent.

No governance forum fixes that, and no escalation route makes someone care about an outcome they are not measured on. What works is embedding the carve-out team inside both organisations from the start, so knowledge transfers as a working relationship rather than as a documentation exercise, and so the seller's people have a stake in the separation landing rather than simply an obligation to permit it. That is the principle behind the Embedded Change Model™, and a carve-out is the sharpest test of it we encounter.

The same logic applies on the other side of the transaction. If you are the seller, the questions a buyer will ask about separability are the questions you should have already answered. Our guide to technology exit readiness covers running that review on yourself, early. And where the carved-out business is being bolted onto an existing platform, the post-merger integration problem starts before the separation has finished, which is a sequencing question worth resolving before completion rather than after.

Intology advises acquirers, sellers and private equity backed businesses on separation and integration. We are independent of the platform vendors and the systems integrators, which matters more than usual here: the TSA exit is the moment a business is most exposed to advice that suits somebody else's practice mix.

Frequently asked questions

What is a carve-out in M&A?

A carve-out is the sale of part of a business - a division, product line or geography - to a third-party buyer, where the divested entity has never operated as a standalone company. It has to be separated from a parent that continues to exist, which makes it operationally harder than acquiring a whole company.

What is the difference between a carve-out and a spin-off?

A spin-off distributes a business to the parent's existing shareholders as a new independent company, with no third-party buyer and no change of control. A carve-out sells the business to someone else. Both require operational separation, but only the carve-out does it against a deadline set by a buyer who has paid for the asset.

What does a transitional services agreement cover?

A TSA sets out the services the seller will continue to provide to the carved-out business after completion, typically IT, finance, HR and payroll, along with the service levels, charges, term, early-termination rights and exit arrangements. The clauses that matter most in practice are service definitions specific enough to be enforceable and the right to exit individual services early without penalty.

How long should a TSA run for?

For a mid-market carve-out involving a shared ERP, twelve to eighteen months is realistic. Six to nine is commonly agreed and commonly overruns, usually because the ERP separation turns out to be a migration rather than a copy, and because the carved-out business needs to build an IT function it has never had.

What does an IT carve-out actually cost?

Four costs: standing up the new estate, the TSA fees themselves, dual-running both estates through cutover, and building the IT capability the business has never needed. Deal models reliably carry the first and reliably omit the third and fourth, which are the ones that grow with every month of delay.

What should carve-out due diligence look at?

Separability, which is a different question from quality. Whether the entity's data has a clean boundary in the ERP or requires migration; whether software licences transfer under their contracts; which operating staff transfer and which stay with the seller; and whether the seller has any real incentive for the TSA to succeed.

What are carve-out financial statements?

Carve-out financial statements present the divested business as though it had operated standalone, allocating shared costs and assets from the parent. Under IFRS there is no single prescribed standard for them, so the allocation basis is a judgement that buyers scrutinise closely, and one that depends on the same entity-boundary data problem that makes IT separation hard.