The committed spend ramp grew 15 to 30 percent faster than the workload it was supposed to track, so the discount everyone negotiated was repaid out of the back half of the term
The headline discount is what the field team wants to discuss. The ramp is what decides the effective rate, and it is agreed in a schedule nobody reads twice.
Prepared by Redress Compliance · August 18, 2026 · Google Cloud advisory. 25 to 35 Private Pricing Agreements reviewed, 2024 to 2025.
Executive summary
The committed spend ramp grew 15 to 30 percent faster than actual workload growth. Later years therefore carried commitment nobody consumed, which raises the effective rate exactly where the discount was supposed to be paying off.
Shortfall exposure ran 10 to 20 percent of the total commitment wherever the ramp was set on optimistic forecasts. The floor is owed whether or not the workload arrives.
Marketplace and partner spend was assumed to count in four deals out of five. It was confirmed in writing in fewer than half of them, which is a large assumption to leave undocumented.
The discount percentage is the part the field team will talk about. It sounds generous and it distracts from the floor being signed, which is the number that actually decides the cost.
How is a Google Cloud PPA actually priced?
As a negotiated discount against a total spend commitment over a fixed term. The commitment is the number that matters. You can model the baseline yourself from the Google Cloud pricing pages, because the PPA is the layer that sits on top of published rates.
The commitment is a dollar floor. You agree to spend at least that amount across the term, and if you spend less you still owe it.
| Element | What it sets | Why it decides the cost |
|---|---|---|
| Term | One, two or three years | Longer locks buy deeper discounts and less optionality |
| Ramp | The year by year spend schedule | A steep ramp outruns the workload and strands commitment |
| Scope | Which products and marketplace purchases qualify | An assumption here moves the effective rate more than the discount does |
| Shortfall | What is owed if consumption falls below the floor | 10 to 20 percent of the commitment in the deals reviewed |
The discount is the distraction
A headline percentage sounds generous and is easy to compare. The ramp is a schedule in an annex, and it is where the effective rate is decided.
Why does the ramp cost more than the discount saves?
Because it compounds against a workload that grows more slowly than it does. A ramp set 15 to 30 percent ahead of real growth means the final years carry commitment the estate never uses.
The discount is applied to qualifying spend. The floor is owed regardless. Once those two diverge, the effective rate in the back half rises past the rate you would have paid with no agreement at all.
Flatten the ramp before you negotiate the percentage. A flat commitment at a slightly worse headline discount beats a steep one at a better headline in every case the reviews modeled, because the steep version is repaid out of unused floor.
The forecast behind the ramp is not neutral
The schedule is usually built from a growth model the vendor helped write. That is not bad faith, it is whose forecast it is, and it is why the ramp needs testing against trailing consumption before it is signed.
The Google Cloud PPA brief
The six levers that decide what the aggregate discount is worth, the ramp arithmetic, the marketplace draw down rules, and the clause set to hold.
Get the brief →What 25 to 35 Google Cloud PPAs showed
Across roughly 25 to 35 Google Cloud Private Pricing Agreements Fredrik Filipsson reviewed in 2024 and 2025, the headline discount was almost never the part that cost buyers money. The commitment structure was.
The committed spend ramp grew 15 to 30 percent faster than actual workload growth. The ramp is usually built from a forecast written with help from the party selling the commitment, which is the part nobody records in the file.
Shortfall exposure ran 10 to 20 percent of the total commitment wherever the ramp was set on optimistic numbers. In roughly 20 of the 30 commitments modeled, the steep ramp pushed the commitment past what the estate consumed.
Marketplace and partner spend was assumed to count toward the commitment in four deals out of five, and confirmed in writing in fewer than half. That is a draw down rule worth more than most discount points, left as an understanding.
The median effective rate gap on the first proposal was 18 percent, and the average ramp oversize against usage was 24 percent. Both close with the same work, which is modeling the floor against trailing consumption rather than against the growth story.
- Percentile standing for your exact deal size and industry, from real closed transactions
- The committed ramp tested against trailing consumption, with shortfall priced per year
- Draw down and marketplace language flagged with the exact quote, the page, and the replacement text
How do committed use discounts stack underneath a PPA?
They sit below it and apply to specific resources or to flexible spend. Committed use discounts are a published program, documented in the committed use discount documentation, so they are predictable in a way the negotiated PPA layer is not.
Resource based commitments lock a machine family in a region. Flexible commitments apply across services. Most enterprises blend them, with the flexible layer covering the volatile part of the estate.
- Read the structure before you accept it, because a PPA that assumes deep resource commitments can trap you on a machine family you plan to retire.
- Get marketplace draw down in writing, since the Google Cloud Marketplace rules change what qualifies and four deals in five simply assumed it.
- Keep the flexible layer large enough to absorb a forecast miss, which is the practical hedge against the shortfall clause.
The same commitment question runs across the other two hyperscalers. Our briefs on AWS EDP commit levels and the Azure MACC price the equivalent floors.
Where the two layers interact badly
Stranding is the risk that repeats. Our comparison of resource based against flexible commitments and our brief on how CUDs behave below the floor both work that interaction.
What the agreements measured, 2024 to 2025
Two cuts of the engagement file frame the gap between the schedule and the estate.
How much faster the committed schedule rose than the consumption it was built to track, across the agreements reviewed.
Owed whether or not the workload arrived, wherever the ramp was set on an optimistic forecast.
Neither number appears in the discount conversation. Both are settled in the schedule that sits behind it, which is why the schedule is the document to open first.
Watch the briefing · 6:33Google Cloud: Is There Leverage? Five TacticsWhy a credible alternative is the only lever that improves the rate without committing you to more volume, and why it evaporates after signing.
Your first five moves
- Model the ramp against trailing consumption before discussing any percentage, since the schedule ran 15 to 30 percent ahead of workload growth.
- Price the shortfall clause year by year, because exposure ran 10 to 20 percent of the commitment wherever the forecast was optimistic.
- Get marketplace and partner draw down confirmed in writing, which four deals in five assumed and fewer than half documented.
- Flatten the ramp before you trade for discount, as a flat floor at a slightly lower headline beat a steep floor at a better one in the deals modeled.
- Open the renewal at 120 days, not 90. The Google Cloud practice builds the consumption model before the commitment conversation starts.
Frequently asked questions
What is a Google Cloud Private Pricing Agreement?
A negotiated discount layer applied against a total spend commitment over a fixed term. It sits on top of published rates and committed use discounts rather than replacing them.
What actually decides the cost of a PPA?
The commitment, not the discount. Across the agreements reviewed the ramp grew 15 to 30 percent faster than workload growth, so the back half of the term carried commitment nobody consumed.
What is shortfall exposure worth?
Between 10 and 20 percent of the total commitment wherever the ramp was set on optimistic forecasts. The floor is owed whether or not the workload arrives.
Does marketplace spend count toward the commitment?
Sometimes, and it needs confirming in writing. Four deals in five assumed it counted and fewer than half had it documented, which is a large assumption to carry.
Should the ramp or the discount be negotiated first?
The ramp. A flat commitment at a slightly worse headline discount beat a steep ramp at a better headline in every case modeled, because the steep version is repaid out of unused floor.
How do committed use discounts interact with a PPA?
They sit underneath it. CUDs are a published program applying to specific resources or flexible spend, while the PPA is the negotiated and confidential layer above them.
What is the risk with resource based commitments?
Stranding. A resource based commitment locks a machine family in a region, so a PPA built on deep resource commitments can trap the estate on hardware it planned to retire.
How big is the gap on a first proposal?
The median effective rate gap was 18 percent and the average ramp oversize against usage was 24 percent, both of which close by modeling the floor against trailing consumption.
When should a Google Cloud renewal open?
At 120 days rather than 90. The consumption model that flattens the ramp cannot be built inside the last quarter of a term.
What is the strongest external lever on a Google deal?
A credible multi cloud alternative. It improves the committed use rate without committing you to more volume, and it only exists before signing.
