Oracle estates are complicated. The TSA exit is engineering, not licensing.

The Oracle estate inventory and scope boundary.

Every Oracle separation starts with an estate inventory. Oracle EBS or Oracle Cloud ERP at the application layer. Oracle Database on premises or in OCI for transactional data. Oracle middleware including SOA Suite, WebLogic, Identity Manager, and Integration Cloud. Oracle Analytics or BI Publisher for reporting. Each component sits in a configuration that mixes Newco usage with seller usage in ways that are rarely cleanly bounded.

The inventory documents the production environment, the non production environments, the integration points, the customizations, and the licensed entitlements per component. Where the seller maintains a global Oracle instance, Newco data sits as one or more legal entities, business units, or operating units inside that instance. The separation work has to extract that footprint cleanly.

The scope boundary defines what Newco takes and what stays with the seller. Pure Newco transactions move with Newco. Shared master data is cloned or split depending on the use case. Customizations that benefit Newco move with Newco. Customizations that only serve the seller stay. The boundary is finalized in the first 60 days of the runway and reflected in the data migration plan.

A clean inventory and boundary decision unlocks every downstream Oracle separation decision. The licensing strategy. The target architecture. The migration approach. The integration redesign. Where the inventory is incomplete, the downstream decisions are made in the dark and reopened later under cost pressure. The pattern is consistent with the TSA exit IT separation framework.

Licensing strategy and the Oracle commercial negotiation.

Oracle licensing is the most consequential decision in the separation. Newco needs licensed Oracle entitlements to run production. The options vary by product. New direct contract with Oracle. Assignment of a portion of the seller’s entitlements where the contract permits. Time bound TSA usage under the seller’s license while Newco completes its own arrangement. Each option has cost, timing, and audit implications.

Oracle’s standard contract terms restrict assignment in most cases. The result is that Newco typically signs a new agreement with Oracle covering the relevant licenses, support, and cloud subscriptions. The negotiation is a one shot opportunity. Oracle reads the situation as a captive buyer with deal pressure. Without preparation, the resulting agreement reflects that read.

The preparation covers usage metrics, the migration plan, the alternative architecture options, and the documented BATNA. Where Newco can credibly migrate workloads to alternative databases or applications, the negotiation has leverage. Where Newco is locked into the current footprint with no alternative, the leverage is thin. The work runs in parallel with the broader vendor commercial strategy.

Cloud migration is the lever that has produced the most savings over the last several years. Newco signs an Oracle Cloud Infrastructure agreement that consolidates compute, database, and applications. The unified architecture reduces operational complexity and creates a defensible commercial position for future renewals. The decision is informed by the long term operating model not just the carve out exit.

Target architecture and the migration path.

The target architecture decision drives the rest of the work. Three patterns dominate. A clone and prune approach where Newco lifts a copy of the seller’s Oracle estate and removes seller data. A reimplementation approach where Newco rebuilds on a new Oracle instance with selective data migration. A replatform approach where Newco moves away from Oracle in part of the stack while retaining it where appropriate.

Clone and prune is the fastest path and the most common for short TSA windows. The seller’s production database is cloned. Newco runs a series of data purge scripts to remove non Newco transactions, master data, and customizations. The Newco instance gets new endpoints, new identity setup, and new integrations. The cutover is a defined window of read only seller access followed by Newco production stand up.

Reimplementation is the cleanest end state but the longest and most expensive path. Newco implements a new Oracle Cloud ERP instance with a target chart of accounts, target organizational structure, and target business processes. The data migration is selective and the cutover is preceded by parallel testing. The path is typically 12 to 24 months end to end.

Replatform decisions are made selectively. Oracle Database may stay on premises with new licensing. Oracle EBS may move to Oracle Cloud ERP. Oracle BI may migrate to a different analytics platform. The decisions are informed by the workload economics, the integration complexity, and the operating team’s capability. The pattern aligns with TSA exit ERP separation planning.

Data migration, identity, and integration redesign.

Data migration is the most technically intricate workstream. Master data first. Customers, suppliers, items, employees, chart of accounts, organizational hierarchies. Each master data domain has a defined extract, transform, and load process with data quality checks and reconciliation reports. The master data migrates before transaction data because transaction data references it.

Transactional data follows. Open transactions move first because they have to continue processing in the new environment. Historical transactions migrate in waves based on business need and storage economics. Some historical data stays in a read only archive that Newco accesses through a reporting interface. The archive approach reduces migration scope while preserving the data.

Identity migration moves from the seller’s Oracle Identity Manager or other directory to Newco’s identity provider. The user accounts, the role assignments, and the access entitlements are mapped and migrated. Where the seller’s identity continues under TSA for a transitional period, the federation arrangement is documented and operational.

Integrations are the workstream that most carve outs underestimate. The Oracle estate has tens or hundreds of integration points to other systems, file shares, and external counterparties. Each integration is documented, tested, and rebuilt in the Newco environment. Where integrations cross the seller boundary, the boundary contract defines who maintains the connection and at what cost.

Testing, cutover, and stabilization.

Testing is sequenced through unit testing, integration testing, user acceptance testing, parallel testing, and dress rehearsals. Each test phase has a defined entry criteria, exit criteria, and remediation cycle. The most consequential test is the parallel run where Newco processes a defined transaction set in both the legacy and the target environments and reconciles the outputs.

Cutover is the window where production processing moves from the seller’s Oracle estate to Newco’s. The window is sized to the migration complexity and the business calendar. Most cutovers happen over a weekend with a defined freeze on transactions on Friday afternoon and a Monday morning go live. The runbook is rehearsed twice before the actual cutover.

Stabilization runs for 30 to 90 days after cutover. Hypercare support is on standby. Production issues get triaged within defined SLAs. Data reconciliation runs against the legacy environment to confirm completeness. Where issues persist into the stabilization period, the carve out PMO and the technology lead decide whether the issue blocks TSA exit certification.

The TSA exit certification confirms that Newco is operationally independent and the seller can shut down the Oracle services that supported Newco. The certification is granular per service in the TSA service catalog. Each service has a defined exit date that triggers the financial and operational decoupling. The discipline runs through the TSA exit milestones framework.

Cost discipline and the integrator commercial.

Oracle separation programs run between $3M and $30M depending on scope, complexity, and the chosen architecture path. The cost dispersion is wide because the underlying decisions vary so much. The economic discipline is to scope the program tightly, choose the lowest cost path that meets the business requirement, and hold the integrator to the agreed scope and timeline.

The integrator commercial is the second biggest line item after the Oracle license commitment. Most Newco buyers select a single system integrator to lead the Oracle separation. The selection criteria include Oracle Cloud ERP experience, carve out experience, fixed fee willingness, and senior team continuity through the project. The contract is fixed fee for defined deliverables with change control discipline.

The most common cost overruns trace to scope creep, integration discovery, and customization recreation. The fix is the disciplined scope definition before contract signing, the early discovery phase that documents integrations and customizations, and the change control process that prices every scope addition in writing. Where these controls are in place, the program lands within the original budget plus or minus 10 percent.

A clean Oracle separation produces a Newco that runs its own Oracle estate with licensed entitlements, supported architecture, and the optionality to evolve the stack on Newco’s timeline. The work is engineering. The economics depend on commercial discipline. The discipline runs through the TSA exit acceleration program.

More on platform separation.