Separating shared software licenses in a carve out.

Separating shared software licenses in a carve out is the buyer side work of splitting the entitlements that two businesses used as one, so the carved out unit ends up properly licensed in its own name and the buyer does not pay for the same software twice. Shared licenses are the quiet centre of carve out risk. A single enterprise agreement, a pooled set of named users or a bundled suite was bought once for an organisation that is about to become two, and there is rarely a clean line down the middle. Where the split is wrong, one side ends up paying for software it cannot use and the other ends up using software it does not license, and both outcomes turn into cost or an audit.

Why separating shared software licenses in a carve out is hard

The difficulty is that a shared license was never designed to be divided. An enterprise wide agreement gives the whole organisation the right to use a product, but it usually gives no standalone right to any single unit, so when the unit leaves it leaves with nothing it can keep. A volume discount was priced on combined consumption, so splitting it raises the unit price for both sides. A shared user pool counts named users across both businesses, so the moment they separate the count has to be reconstructed or both sides risk double counting. None of this appears on a contract list. It only appears when someone reads the agreement against the actual usage on each side of the line.

The parent has every incentive to keep its own position clean and to let the carved out unit absorb the awkward edges. Left unmanaged, the unit inherits the shortfall, the lost discount and the unlicensed use, and the buyer pays for all three. Separating the licenses properly is how the buyer stops that transfer of cost. It is part of the wider carve out software licensing playbook and depends on the entitlement map built earlier in that sequence.

The five ways a shared agreement splits

Most shared arrangements fall into one of five patterns, and each splits differently. An enterprise wide agreement usually cannot be divided at all, because the unit has no separable right, so the new entity has to license from scratch. A volume or tier discount can be split, but both sides lose the price break that combined volume bought. A shared user pool has to be reconstructed user by user, which is where double counting and unlicensed use hide. A bundled suite, where several products are licensed as one, often cannot transfer a partial entitlement, so the unit loses access to components it relied on. And a shared support contract typically covers many products on one line, so coverage lapses for the unit at separation unless a new contract is in place. The table sets these out with what gates each split and the exposure it creates.

Knowing the pattern matters because it decides the remedy. An enterprise agreement that cannot be split becomes a re licensing cost, the subject of re licensing the carved out business from scratch. A lost discount becomes a negotiation, covered in negotiating software terms for the new entity. A reconstructed user pool becomes a reconciliation. Each pattern has a defined path, and naming the pattern is the first step to costing the fix.

Where stranded cost and audit risk appear

Two exposures sit on either side of a badly split license. Stranded cost is the software the parent keeps paying for after the unit leaves, because the contract was sized for a larger organisation and cannot be reduced mid term. That cost is the parent problem on paper, but parents routinely try to push a share of it onto the carved out unit, so the buyer has to watch for it. The detail is in stranded software costs in a carve out. On the other side sits licensing shortfall, where the unit uses software it no longer has the right to, which is unlicensed use and a direct audit risk once the new entity is operating alone.

A dependency on a parent entitlement that does not transfer is therefore both a cost gap and a future audit. Publishers know that carve outs reshuffle entitlements and they treat the event as a reason to count. Surfacing every shared license before close, and re contracting the ones that do not transfer, is what keeps the new entity off the audit list. This is commercial and licensing work, and the buyer own counsel should interpret any clause that governs how a shared agreement may be divided or assigned.

A worked example of a shared split

Consider a composite case. A parent holds an enterprise database agreement covering 400 processor licenses across the whole group, bought at a deep discount sized to that volume. The unit being carved out runs on 90 of those processors. There is no line in the agreement that gives the unit a right to 90 processors on its own, so the agreement cannot simply be assigned in part. The new entity has to license 90 processors from scratch, at a unit price far above the group discount, because it no longer brings 400 processors of volume to the table. Meanwhile the parent is left with an agreement still sized for 400 processors while only 310 are now in use, which is stranded cost it will try to share.

The buyer side response is to model both sides before close. The 90 processor re licensing cost is a known number that belongs in the deal, ideally pushed back to the seller as a price adjustment. The lost volume discount is a negotiation lever, because the publisher would rather keep the new entity as a customer than lose it, and an independent advisor can run that negotiation without the conflict a reseller carries. Handled this way, a split that would have cost the new entity an unbudgeted seven figure license bill becomes a priced, negotiated item resolved before the deal closes. Handled badly, it surfaces as an audit finding once the new entity is alone and the publisher counts 90 processors with no contract behind them.

How to separate shared licenses cleanly

The method is methodical, not clever. Start from the entitlement map and flag every license that more than one unit uses. For each shared license, read the agreement to find whether the unit has any separable right, then measure the unit actual consumption against that right. The difference is the shortfall to re contract or the surplus the parent retains. Resolve the gated ones first, because agreements with assignment or change of control clauses need publisher involvement and a long lead time. Carry every figure into the deal model so the cost of separation is priced, not absorbed. The whole exercise sits inside our carve out and TSA separation service.

The test of a clean separation is simple. On the day the TSA ends, every application the unit runs should sit on an entitlement the new entity owns in its own name, counted on its own metric, with its own support line. Nothing should depend on a parent agreement, and nothing should be running without a license behind it. Getting to that state before the exit date is the whole point of separating shared licenses, and it is the foundation of standing up the new entity software estate and the wider carve out playbook.