Cost Optimization Fees: Why Percentage of Savings Aligns Incentives

Every commercial model carries an incentive, and the incentive shows up in the work whether anyone intends it to or not. For most OCI consulting the misalignments are tolerable: an hourly architect has some interest in the engagement running long, but the work itself is visible enough that padding gets noticed. Cost optimization is different. The deliverable is the absence of spend, the analysis happens inside billing data the buyer rarely reads closely, and the provider controls both the diagnosis and the prescription. That combination makes the choice of fee structure the single most important decision in the engagement, more important than the provider's tooling, methodology, or slide deck. This article is part of our complete guide to hiring an OCI partner, and it takes the commercial question head on: who should bear the risk that the savings never materialise, and what should they earn when they do.

Our answer, and the basis on which our own optimization practice runs, is that the provider should bear that risk entirely. A fee charged as a percentage of verified savings, with no savings meaning no fee, is the cleanest contract in the industry. But clean does not mean simple, and a savings share deal signed without understanding the baseline, the measurement window, and the exclusions can sour just as badly as any hourly engagement. The detail below is what keeps it honest.

The incentive problem with hourly cost work

Run a cost optimization project on time and materials and you have built a machine that pays its operator to move slowly. The provider is rewarded for every additional week of analysis, every workshop, every revision of the findings deck, and is rewarded nothing extra for the savings themselves. Nobody needs to act in bad faith for this to go wrong. Hourly billing simply removes the deadline pressure that forces prioritisation, so the work expands, the easy wins wait politely behind the comprehensive review, and three months in you have a beautiful report and an unchanged bill. The general mechanics of hourly pricing, and where it does make sense, are covered in our piece on OCI consulting rates in 2026. For cost work specifically, the verdict is blunt: the meter and the mission point in opposite directions.

The quieter problem with reseller led optimization

The second misalignment is structural rather than behavioural. A reseller or Oracle aligned firm earns margin or rebate on your OCI consumption, which means every dollar it helps you save is a dollar removed from its own revenue line. That does not make reseller optimization reviews worthless, but it predicts their shape with remarkable accuracy: the recommendations cluster around commitment restructuring and product substitution, the kind that reorganise spend, and thin out around decommissioning and rightsizing, the kind that genuinely reduce it. We are not Oracle and we are not a reseller, and that independence is not a slogan, it is the precondition for the fee model this article is about. A firm cannot credibly charge a percentage of savings while earning a percentage of spend. The broader question of when Oracle's own consulting arm is and is not the right choice gets a full treatment in OCI partner vs Oracle consulting.

How a percentage of verified savings actually works

The model has three load bearing parts: a baseline, a measurement window, and a definition of what counts. Get those three right and everything else is arithmetic.

The baseline

The baseline is your normalised OCI run rate before the work starts, typically built from three to six months of billing data, adjusted for one off spikes and signed by both parties before any change is made. The signature matters. A baseline agreed after the fact is an invitation to argue, and a provider who resists writing the baseline down before starting is telling you how the measurement conversation will go later.

The measurement window

Savings are measured as the gap between the baseline and actual spend over an agreed window, usually three to twelve months after implementation. The window needs to be long enough to prove the savings are real and persistent, and short enough that the attribution stays plausible. A provider claiming credit for your bill two years out is claiming credit for weather.

What counts as a saving

The honest categories are the ones caused by the provider's work: rightsizing of compute shapes against observed utilisation, storage tiering and lifecycle policies, commitment and Universal Credits optimization, license aware shape choices that exploit BYOL positions properly, and decommissioning of resources nobody could justify. What does not count is just as important. Usage that fell because the business shrank, a project ended, or a workload moved elsewhere is not a saving anyone should pay a fee on, and a well drafted contract excludes it explicitly. The verification mechanics are tedious and they are the entire point: a savings number both sides trust is what lets the rest of the relationship stay simple.

The four commercial models side by side

Here is the comparison that matters when you are deciding how to buy cost work, with retainers included because steady state cost governance is often bundled into managed operations, a structure we price out in OCI managed services pricing.

No row is wrong in all cases, and the last section of this article covers when the middle rows beat the bottom one. But for a first serious optimization pass on an estate of any size, the bottom row is the default for a reason: it is the only model in which a provider who finds nothing earns nothing.

What the market charges

Savings share fees in the OCI market in 2026 mostly land between 20 and 35 percent of first year verified savings. Below 20 percent, be curious about how the provider makes money, because deep optimization work is expensive to deliver and a fee that cannot cover senior people usually means junior people or shallow analysis. Above 35 percent, the alignment story starts to wobble in the other direction, and you should expect either an exceptional scope or a hard cap in exchange.

The first year framing matters more than the percentage. A rightsized compute fleet keeps saving money in year two and year three, but paying a fee on those years overcompensates the provider for work done once, so the standard market treatment is that recurring savings pay a fee on the first twelve months only. Genuinely one time savings, such as recovering a credit or eliminating a duplicated environment, pay once on the amount recovered. If a proposal charges on multi year savings, treat the effective fee as the multiple it really is and compare accordingly.

For calibration on outcomes rather than fees: across our optimization engagements the average result is a 40 percent reduction in OCI spend. The arithmetic that follows from a number like that is why the model survives contact with procurement. On a meaningful estate the fee pays for itself out of money that was already leaving the building, the buyer keeps the majority of every dollar saved, and the provider's incentive runs in one direction only: find more.

A worked example

Take an estate running at $2 million a year on OCI. An optimization pass finds and implements changes worth $700,000 a year in verified recurring savings, a 35 percent reduction and slightly below our own average. At a 25 percent savings share, the fee is $175,000, paid once against the first year. The buyer banks $525,000 in year one and the full $700,000 every year after, against a fee of zero if the work had found nothing. Now run the same engagement hourly: a team of two specialists for four months lands somewhere near $250,000 in fees whether the savings arrive or not, and the invoice is identical in the scenario where the bill never moves. The savings share is not always the cheaper number when the work succeeds. It is the only number that is guaranteed to be smaller than the value it created.

Seven contract checks before signing a savings share deal

The model is only as honest as its drafting. Before you sign, walk the contract through these seven checks in order, and treat any resistance to the first two as disqualifying.

When a fixed fee or retainer is the better model

A savings share is the wrong tool in three situations. First, when the estate is small or recently optimized, the discoverable savings may not justify the verification machinery, and a fixed project fee for a focused assessment is cheaper and faster for both sides. Second, when what you really need is governance rather than a rescue, meaning tagging discipline, budget alerts, monthly reviews, and an owner for cost as a standing concern, a managed monthly retainer fits the shape of the work, because the value is continuous rather than episodic. Third, when your organisation cannot grant a provider the access or the change throughput to implement, the honest options are a fixed fee for the analysis or waiting until implementation is possible, not a savings share that will end in an argument about unactioned recommendations.

In practice the models chain together. A typical sequence in our own OCI cost optimization practice starts with a savings share engagement to take the estate down hard, then transitions to a light retainer that keeps the gains from eroding as new workloads land. The first phase carries no fee risk for the buyer, and the second phase costs a fraction of what the first one saved. Teams with 20+ years of combined Oracle experience have watched enough optimization gains evaporate over eighteen months to insist on the second phase, because the cloud regrows spend the way a garden regrows weeds.

One last practical note on evidence. Everything in this article assumes the savings can actually be measured, and that assumption depends on instrumentation that most estates do not have on day one. Utilisation data deep enough to defend a rightsizing decision, tagging clean enough to attribute spend to owners, and a billing history long enough to build an honest baseline are prerequisites, not nice to haves. Estates already under continuous observation, the kind that 24/7/365 monitoring provides, walk into a savings share engagement with the verification problem largely solved, because the before picture already exists in detail. Estates without it usually spend the first weeks of the engagement building that picture, which is worth doing properly even though it delays the savings, since every later conversation about the fee stands on it.

The conclusion is the one this article opened with, now with the machinery visible. Commercial structure is not paperwork that follows the engagement; it is the engagement's operating system. Hourly cost work pays for motion, reseller cost work pays the wrong party, and a verified savings share pays for the outcome and nothing else. Get the baseline signed, the exclusions written, and the verification sequence enforced, and the cleanest fee model in OCI consulting will do exactly what it says on the contract.