$14.2M saved on a three-year Oracle ULA exit.

A ULA inherited from two CIOs ago.

The client had entered a three-year Oracle Unlimited License Agreement in 2022, originally scoped to support a global PMS replacement and a data-warehouse modernisation programme. By 2025, the PMS programme had pivoted to a SaaS platform that did not require Oracle, and the data warehouse had been migrated to a cloud-native architecture using a different database engine entirely.

Despite that, the ULA schedule still listed three Oracle product families — Advanced Compression, Active Data Guard, and Real Application Clusters — alongside the core Database Enterprise Edition and Diagnostics Pack the client did still use. With fourteen months left on the ULA, Oracle's account team had begun positioning a "ULA-to-Cloud" conversion proposal valued at $19M over five years.

Why the proposal was excessive

In our experience across 340+ engagements, Oracle's pre-exit proposals rely on a specific commercial asymmetry: the customer almost never has a complete inventory of what is actually deployed inside the ULA scope, but Oracle does. If the customer cannot prove a clean deployment number at certification, the only safe path is to renew — at terms heavily weighted toward Oracle.

In this case, the proposal assumed continued use of all listed product families at peak deployment. We knew before we started that the actual deployment for Advanced Compression and Active Data Guard was zero, and that RAC usage was confined to a single legacy cluster scheduled for decommission within the ULA term.

Eleven months. Four phases.

Phase one — independent estate scan

We ran an independent deployment scan across every Oracle host in the estate. Critically, we did this using our own scripts rather than Oracle's collection tooling, which would have left a footprint visible to Oracle LMS. The output: a verified, exportable inventory of every Oracle binary, every enabled option, and every connected feature across 312 hosts in eight data centres and three cloud regions.

Phase two — certification position design

With clean deployment data, we constructed three certification scenarios: aggressive (only what was running on the certification date), defensible (running plus reasonable buffer), and conservative (Oracle's likely interpretation). The aggressive scenario excluded all three under-used product families. The defensible scenario excluded two. We modelled the certification and exit economics under each.

Phase three — pre-certification remediation

For Advanced Compression and Active Data Guard, we worked with the database operations team to confirm — and document — that the features were disabled at the parameter level across the estate by the certification date. For RAC, we accelerated the legacy cluster's decommission timeline by three months. By the time Oracle requested the certification submission, the deployment evidence supported the aggressive scenario.

Phase four — Oracle negotiation

Oracle's initial response to the certification was predictable: a counter-proposal arguing that the historical entitlement schedule should bind future use rights and that the three excluded products should be re-added at list. We pushed back using the same ULA clause language Oracle drafts — "deployed by the end of the ULA Term" — and held the position through three negotiation rounds.

$14.2M. And a clean exit.

• Three product families excluded from ULA certification — no commercial liability for unused Oracle options.
• ULA converted to a right-sized perpetual entitlement at $4.8M, versus the $19M conversion Oracle had proposed.
• Annual Oracle support obligation reduced 31% on a permanent basis.
• Oracle LMS audit risk dramatically reduced — the certification record is now the customer's defensive baseline.
• Documented saving of $14.2M against Oracle's modelled trajectory, signed off by the client's CFO and external auditor.

Why this worked

Three things make a ULA exit go well, in order of importance: clean deployment data the customer controls; a certification position designed twelve months out, not three; and a negotiator who knows which Oracle clauses are scripted defences and which are real. We assembled the first two before Oracle was paying attention, then deployed the third when the certification submission opened. None of it would have worked in the final ninety days.

If you are inside the final eighteen months of an Oracle ULA — or you have just received the renewal proposal — the time to start designing the certification position is now. We typically do this work on a fixed-fee basis, scoped against ULA size and product complexity.