Exit is one of the few moments you hold real leverage on Oracle support, because Oracle would rather you renew than certify. Used well, that leverage shapes years of future support cost. Used blind, the same conversation can trigger repricing that costs more than it saves.
The support negotiation at exit is about the future, not the certification fee, which is zero. Your leverage is the choice between certifying and renewing, and Oracle's preference for the latter. Trade that choice deliberately, and always model the repricing matrix before you agree to drop anything.
A ULA ends with a decision that only you can make: certify the deployment into perpetual licenses, or renew for another term. Oracle has a strong commercial preference for renewal, because a renewal is new revenue and a certification is not. That preference is your leverage. While the decision is open, Oracle has reason to be accommodating on the terms that surround it, including the shape of your future support arrangement.
The leverage is real but narrow. It exists while the certify or renew decision is genuinely live, and it fades the moment you signal which way you are going. The negotiation has to be set up before that signal is sent.
Never reveal whether you intend to certify or renew until the surrounding terms are agreed. The open decision is the leverage. The moment it closes, the leverage closes with it.
Oracle moves most readily on items that can be folded into a wider deal. It moves rarely or never on items that would set an awkward precedent. The indicative split below reflects typical behaviour and is not a guarantee, because every negotiation turns on your specific position and contract.
| Usually negotiable in a deal | Rarely moves alone |
|---|---|
| A capped or held uplift for a period | Removing the uplift permanently |
| Consolidating fragmented support sets | Repricing policy on dropped products |
| Terms on a future purchase or renewal | A standalone support discount with nothing in return |
| Cloud credits or migration terms | Refunds on support already paid |
The most common mistake is to ask Oracle to drop support on licenses you no longer use, expecting a clean saving. Oracle prices support as a set, and dropping part of it can reprice the remainder to standalone rates under its policy. The net saving can be a fraction of the headline, and in some matrices it disappears. This is contract and policy specific, so the matrix has to be built on your actual support set, with the repricing impact calculated, before any reduction is tabled. Walk into the conversation knowing your own numbers and you negotiate from strength. Walk in without them and Oracle's model is the only one in the room.
Set up the support negotiation while the certify or renew decision is still open, model the repricing matrix first, and trade your leverage for terms that matter for years. Two companion notes complete the picture: support escalation after the ULA explains the uplift you are negotiating against, and shelving versus using certified licenses covers what to do with entitlement you are not deploying. The wider context sits in our post certification guide.
Yes, but only on the terms Oracle is willing to move, and only with leverage. Exit is one of the few moments you hold leverage, because Oracle wants you to renew rather than certify. The opportunity is to shape the future support arrangement, not to win an arbitrary discount.
Sometimes, as part of a wider deal such as a renewal or a new purchase, Oracle may cap or hold the uplift for a period. It rarely moves the uplift in isolation. Any concession on the uplift should be weighed against what Oracle asks in return.
Only after modelling the repricing matrix. Dropping support on shelved licenses can trigger repricing of the licenses you keep, shrinking the saving. The decision is contract and policy specific and must be modelled on your actual support set first.
We set up the support negotiation, model the repricing matrix, and represent your position while the certify or renew decision is live. Book a confidential assessment.