Abstract advice about support optimization only goes so far. What makes the levers concrete is watching them applied in order, and watching the difference that order makes. The following worked example is anonymized to sector and rough size, with figures verified against the signed outcome. It follows a buyer who first tried the obvious move, got nothing for it, and then reached a real saving by sequencing the same components correctly.
What was the starting position?
The starting position was a mid sized financial services estate paying a large annual Oracle support bill that had escalated steadily for years. The licenses had been bought in several tranches, so the support sat across multiple sets with different compositions. A block of licenses was genuinely unused after an application was retired, and the obvious move was to drop support on that block. On paper it promised a clean reduction. In practice the sets were the problem the buyer had not yet seen.
Why did the first attempt save nothing?
The first attempt saved nothing because the unused licenses sat in the same set as licenses still in production, and matching service levels let Oracle reprice the remainder toward list when support on part of the set was dropped. The reduction on the retired block was almost exactly offset by the higher rate on what stayed. The invoice barely moved. The lesson, learned the expensive way, was that a partial termination cannot be judged on the portion you drop. It has to be judged on the net of the drop and the repricing it triggers across the whole set.
| Approach | Sequence | Net result |
|---|---|---|
| First attempt | Drop unused block mid term | Repricing offsets the saving |
| Reworked | Review sets, time to renewal, add OCI offset | A durable, compounding saving |
What did the reworked sequence look like?
The reworked sequence began with a set composition review that mapped exactly which licenses sat in which set, so the repricing exposure was known before any action. The partial termination was then timed to the renewal window, where the remaining support could be repriced explicitly rather than punitively, and the sets were restructured so the retired licenses no longer dragged the rest. Finally an OCI based offset through Support Rewards reduced the residual support against cloud consumption the estate was going to use anyway. Three components, the same ones available at the first attempt, sequenced so each landed net positive.
What was the verified outcome?
The verified outcome was a durable annual support reduction that compounded across the remaining renewals, where the first attempt had produced no net saving at all. The headline was not a one off rebate but a lower base that escalation now worked on, so the saving grew across the five year horizon rather than shrinking. The decisive factor was sequence and timing, not any single clever move. The buyer move this example teaches is simple to state and easy to get wrong: review the sets first, act at the renewal, and model every component net of the repricing it triggers before you pull it.
What does this example teach about sequence?
This example teaches that sequence is the variable that decides whether a support reduction lands, because the identical components produced no saving in one order and a durable one in another. The components were never the problem. The unused licenses were genuinely unused, the renewal window was always coming, and the OCI offset was available throughout. What changed was the order and the timing: review before action, action at the renewal, offset layered last. A buyer holding the same pieces can reach either outcome, which is why the planning step, unglamorous as it is, carries the entire result.
For the clean mechanics of dropping support, see the support termination sequence that works. For the set composition trap in detail, see terminating support on unused licenses. The full method sits in the Oracle negotiation guide, and you can take the next step on the support cost optimization service or through contact.