Third party support, the 42 percent that survives the arithmetic
Leaving Oracle support saves roughly half the maintenance line in year one, and what matters is the number left after the contract mechanics take their cut: on the estates we model that lands at 30 to 45 percent over five years. The binding constraint is not legality and not the provider; it is that patches and upgrade rights stop on the termination date and do not resume until you pay to come back.
Prepared by Redress Compliance · August 7, 2026 · Oracle advisory. Based on 40 to 45 support exit decisions advised 2024 to 2025.
Executive summary
The worked model lands at 42 percent, eight points below the quote. On a $10 million net license estate, Oracle support at 22 percent is $2.2 million a year and the third party quote lands near $1.1 million: across five years with Oracle's 4 percent uplift compounding, the gross difference is $6.42 million, and after internal engineering at $250,000 a year and one time transition costs, $5.02 million remains, 42 percent. Every point of the gap between the quote and the result is a line item you can identify in advance.
The uplift you never pay is a fifth of the prize. Five years of 4 percent compounding adds about $0.92 million to a $2.2 million support line, roughly one fifth of the whole Oracle cash comparison, and buyers routinely leave it out. Where no cap was negotiated, model the uplift at the rate your last three invoices actually moved, not the rate you hope for.
The partial move can cost you money. Matching service levels plus repricing of the retained licenses can leave a 70 percent move roughly $150,000 a year worse off than doing nothing: the repricing of survivors is the single largest destroyer of partial move business cases, and it killed them with arithmetic, not policy debates. The exits that worked were clean or carefully carved along contract lines Oracle's policies actually permit.
The door swings one way, at 150 percent. Oracle prices reinstatement at 150 percent, and after twelve months the multiplier runs against the fees for the whole lapsed period, so the return gets more expensive the longer you stay away. The release date decides suitability: a release already in Sustaining Support gives up patches Oracle stopped writing, which changes the comparison entirely, and Support Rewards die with the support contract, removing the bill the OCI credits were reducing.
The five year model, on a $2.2 million support line
| Year | Stay with Oracle, 4 percent uplift | Third party at half, flat | The difference |
|---|---|---|---|
| Year 1 | $2.20M | $1.10M | $1.10M |
| Year 3 | $2.38M | $1.10M | $1.28M |
| Year 5 | $2.57M | $1.10M | $1.47M |
| Five year total | $11.92M | $5.50M | $6.42M gross |
| Less internal engineering | $1.25M | $5.17M | |
| Less transition, one time | $0.15M | $5.02M, which is 42 percent |
The six items that quietly eat the saving
- Repricing of retained licenses: the matching service levels machinery reprices the survivors of a partial move, the single largest destroyer of those business cases.
- Forfeited Support Rewards: OCI consumption credits offset the technical support bill, and terminating support removes the bill they were reducing.
- Internal engineering: compensating controls, backports, and platform testing now have an owner, one to two full time equivalents on a large estate.
- Re certification work: every OS patch, storage refresh, and browser change needs validation the vendor used to imply.
- The transition project: knowledge capture, archive downloads, and the provider onboarding, one time but real.
- The return premium: reinstatement at 150 percent, compounding against the lapsed period, the exit's true reversal cost.
The Oracle third party support brief
The economics and the decision end to end: the five year model, the partial move arithmetic, the release gate, and the provider selection sequence.
Get the white paper →Which estates fit, and the release gate that decides
The buyers who switched successfully could name the release they were running and the year they would retire it; the ones who could not, should not have. The fit profile is stability: mature releases with retirement dates, no upgrade dependency inside the horizon, and the internal or provider capability to own compensating controls, with the release status as the gate, Premier releases give up patches you would have applied, Sustaining releases give up patches Oracle stopped writing. The decision pages divide the labor: the provider selection framework for coverage, indemnity, and security models, the legal position for what the courts actually held, and the Rimini Street guide for the largest provider's specifics. The Support Rewards interaction models before anything terminates, because the credits die with the contract.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across support exit decisions, 2024 to 2025
Across roughly 40 to 45 Oracle support decisions Fredrik Filipsson advised in 2024 and 2025, about two thirds of the buyers who opened the question did not switch:
Mostly because the corrected model, repricing and rewards included, no longer cleared their threshold.
Every serious third party quote made Oracle cheaper after Oracle heard about it, a result in itself.
The first model was almost always too optimistic, comparing the provider quote to the Oracle fee and stopping there, and the correction discipline is the page's whole method: price the repricing, the rewards, the engineering, and the transition before deciding, because 42 percent is worth chasing and 50 percent was never real. The last finding stands alone: the credible, costed exit moved Oracle's number in every case, which makes the evaluation worth running even for estates that stay, inside the wider support cost strategy and the reduction playbook.
Your first five moves
- Compute your net fees after every discount, because 22 percent of net, not list, is the input the whole model runs on.
- Run the five year model with all six deductions: repricing, rewards, engineering, re certification, transition, and the return premium.
- Test the partial move against matching service levels first, because the surviving licenses reprice and can eat the entire saving.
- Gate on the release: name what you run and when it retires, because the buyers who could not should not have switched.
- Table the costed quote either way, since every serious one made Oracle cheaper. The Oracle practice runs the decision with you.
Frequently asked questions
How much does Oracle third party support save?
Gross, about half the support line: a $2.2 million Oracle fee typically quotes near $1.1 million. Net over five years, 30 to 45 percent once repricing of retained licenses, forfeited Support Rewards, internal engineering, and transition are priced, 42 percent in our worked model, and the gap between the quote and the result is identifiable line by line in advance.
What is the biggest risk in leaving Oracle support?
The one way door: patches and upgrade rights stop on the termination date, reinstatement prices at 150 percent, and after twelve months the multiplier runs against the whole lapsed period. The release gate manages it, estates on named releases with retirement dates fit, and estates that cannot name their upgrade horizon do not.
Can we move only part of the estate to third party support?
Carefully or not at all: Oracle's matching service levels rules plus repricing of the retained licenses destroyed most partial move cases, leaving a 70 percent move roughly $150,000 a year worse than doing nothing in our modeling. Partial moves work only along contract lines the policies actually permit, priced before, not after, the termination notice.
Does the Oracle support uplift matter in the comparison?
It is a fifth of the prize: 4 percent compounding adds about $0.92 million to a $2.2 million line over five years, pure avoided cost the third party's flat fee never pays, and routinely omitted from first models. Without a negotiated cap, model the uplift at the rate your last three invoices moved, not the rate you hope for.
What happens to Oracle Support Rewards if we leave?
They die with the support contract: the OCI consumption credits offset the technical support bill, and terminating support removes the bill they were reducing, which converts an apparent saving into a transfer for OCI heavy estates. The rewards interaction models before anything terminates, because it alone can close the case.
Is opening the third party question worth it if we stay?
Yes, measurably: every serious third party quote in our file made Oracle cheaper after Oracle heard about it, and two thirds of buyers who opened the question stayed, many on improved terms. The costed evaluation is renewal leverage whether or not the estate moves, which makes it worth running on its own economics.
