Certification is free, but the estate it leaves behind is not. The running cost is a support base that does not fall on its own, a growth line, and the offsets that decide your real five year number.
By Daniel Voss · Ex Oracle LMS · 4 June 2026
The running cost of a certified estate has three parts. The support base continues at the ULA level and does not fall because you certified, typically rising a few percent a year. The growth line is the cost of new licenses bought when deployment exceeds the certified count. The offsets are the deliberate moves that reduce the net, such as terminating support on cleanly separable unused licenses or moving workloads to OCI. The base is the largest and stickiest line, so the model is decided by what you do to it, not by the one time certification.
Organisations often model the ULA as a single decision at exit and then stop thinking about cost. The decision to certify is indeed a one time event, and it carries no fee. But the estate it produces has a running cost that compounds for years, and that cost is where the real money sits. To plan it, separate it into three lines that behave very differently, because lumping them together hides exactly the lines you can influence.
The largest line is the annual support you already pay, continued at the ULA level after you certify. This is the single most important fact in the model: certification does not lower it, and certifying a larger count does not raise it. The base is fixed by your prior support spend, not by the number of licenses you convert. It then tends to rise by a small annual uplift, commonly a few percent, which compounds. Left untouched, the base is the dominant cost of holding Oracle for the next five years.
The second line is the cost of buying new licenses when deployment grows past the certified count. During the term, growth was free. After certification, growth beyond the ceiling has to be licensed deliberately. If you monitor deployment against the count, this line is planned and modest. If you do not, it arrives as an audit finding with penalty pressure attached, which is the same licenses at a far worse price. The growth line is small or large depending almost entirely on whether anyone is watching.
The third line is negative, the deliberate reductions you can make. Terminating support on a cleanly separable set of unused licenses removes its fee, where the matching rule allows it. Moving workloads to OCI can convert some Oracle spend into a form that carries different economics. These offsets do not happen on their own, and several are constrained by the same support pricing rules that govern the base, so each has to be modelled rather than assumed.
The base is where the money is, so the base is where the work belongs. Certification gets the attention because it is the dramatic moment, but it is free, and the support base that follows it is the figure you will pay every year for as long as you hold Oracle. An organisation that certifies well and then ignores the base has solved the cheap problem and left the expensive one running. Model the base first, plan the growth line, and pursue the offsets that survive the pricing rules.
Build it as a five year projection of the three lines, then test the levers. The illustrative figures below are indicative and use round numbers to show the shape, not real prices. They assume a support base of one hundred units in year one and a three percent annual uplift.
| Year | Support base | Growth line | Offsets | Net (indicative) |
|---|---|---|---|---|
| 1 | 100 | 0 | 0 | 100 |
| 2 | 103 | 4 | 0 | 107 |
| 3 | 106 | 8 | minus 6 | 108 |
| 4 | 109 | 12 | minus 6 | 115 |
| 5 | 113 | 14 | minus 10 | 117 |
The shape is the lesson. The base drifts upward on its own, the growth line climbs if deployment is unmanaged, and the offsets bend the net back down only where the pricing rules permit. The figures are indicative, but the model shows where attention pays: a few percent shaved from the base each year outweighs almost anything that happens at the moment of certification, because the base recurs and the certification does not.
Consider an anonymized financial services firm that certified a strong count and then modelled its five year estate cost before deciding anything. The model showed the support base, untouched, dominating every other line. It identified a separable set of unused licenses it could drop without repricing the rest, planned its growth purchases against monitored deployment, and tested an OCI move for one workload group. None of these was dramatic, but together they bent the five year net down meaningfully against the do nothing base. A second anonymized firm certified equally well, then treated the estate as settled and let the base compound untouched while unmonitored growth pushed up the growth line. The figures are indicative, but the first firm paid materially less over five years for the same certified position, purely because it modelled the running cost instead of the one time event.
The cost model is only as good as the levers you can actually pull, so understand the constraints before you build it. See where support can and cannot be reduced in terminating support on unused licenses, explore the reductions that bend the net in support rewards and OCI offsets, and ground your post certification cost work in our post certification audit defense guide.
Book a ULA assessment and we will build the cost model of your certified estate, find the offsets the pricing rules allow, and bend the running cost down where it actually compounds.