The published discount is the one everyone benchmarks. The negotiated one is where the money is.
Enterprise Google Cloud pricing runs on two layers at once: committed use discounts lower the rate on resources, and a negotiated agreement applies a further percentage off total spend. The layers stack, so benchmarking one understates what is achievable. Most buyers work the visible layer hard, accept the invisible one as offered, and leave the larger number on the table.
Prepared by Redress Compliance · August 10, 2026 · Google Cloud advisory. Based on 20 to 30 commitment and agreement reviews, 2024 to 2025.
Executive summary
Discounting stacks on two layers, and 60 to 75 percent of buyers negotiated only one of them. Committed use discounts trade a one or three year commitment for a lower rate on resources, and they are published, visible, and easy to benchmark.
On top of them sits a negotiated agreement applying a percentage off list against a total spend commitment over the term, which is neither published nor fixed.
In our file the majority of buyers pushed hard on CUDs and accepted the agreement discount exactly as offered, which is the layer where the largest unbenchmarked discount actually sits.
Resource based commitments carry the deepest rate and stranded 15 to 30 percent of value when workloads moved.
The two CUD forms trade depth against flexibility: resource based locks you to specific machine families and regions for the deepest compute rate, while spend based follows workloads across eligible services at a slightly shallower rate. Three year terms consistently beat one year on both.
The common advice is to maximise the resource based rate, and in roughly two thirds of the estates we benchmarked, that advice stranded value when families or regions shifted and the deeper rate never covered the loss.
Size to evidenced steady state, because a commitment above real usage destroys the discount it earned. Cover predictable baseline consumption with commitments and let spikes run at on demand rates.
Over commitment means paying for capacity you never consume, which can wipe out the rate advantage entirely, and it is the failure mode that follows directly from a vendor built growth forecast.
Use resource based commitments only where the machine family is genuinely fixed, and hold volatile workloads on spend based commitments where the flexibility is worth more than the extra points.
Marketplace spend often counts toward the commitment, and leaving it out costs a discount tier. The agreement threshold and its discount are both negotiable, and the buyer who can credibly aggregate every eligible spend stream reaches a better tier than one who counts direct compute alone.
Eligible third party Marketplace purchases frequently qualify, and they were left out of most early agreements in our file.
At renewal the same aggregation logic applies with more force: benchmark the rate against your current scale rather than the scale you signed at, because usage grows and a rate that looked strong at signature is often below market by the second term.
The four levers and what each one costs you
| Lever | Trade | Discount depth | Flexibility |
|---|---|---|---|
| Resource based CUD | Lock to family and region | Deepest on compute | Low |
| Spend based CUD | Commit total spend | Strong across services | Medium |
| Negotiated agreement | Total spend over term | Stacks on top of CUD | Medium |
| Marketplace inclusion | Route eligible spend | Counts toward the commit | High |
The published committed use rates are the starting point, not the ceiling, and the real benchmark question is what the combined CUD and agreement discount reaches for an estate your size.
Compute commitments reach deeper rates than many managed services, and three year terms consistently beat one year, but the right commitment level is always your evidenced steady state rather than your peak.
That is the discipline the published rate card cannot teach you, because the rate card assumes the consumption is real. Size to baseline, keep spikes on demand, and treat the agreement layer as the place where preparation converts into points.
The full commitment mechanics sit in the Google Cloud FinOps and CUD playbook.
Working the agreement layer and the spend threshold
- Benchmark the agreement, not just the CUD rates. The negotiated percentage off list is where 60 to 75 percent of buyers in our file did no work at all, and it stacks on top of everything below it.
- Aggregate every eligible spend stream before you quote a tier. Third party Marketplace purchases frequently qualify toward the committed spend, and including them can lift you into a materially better band.
- Treat the threshold as negotiable. Both the spend level that unlocks a tier and the discount attached to it move for a buyer who can evidence the aggregated number.
- Cover only stable baselines with resource based CUDs, since the deeper rate never offset the lost flexibility in roughly two thirds of the estates we benchmarked.
- Re benchmark at renewal against your new scale. Usage grows over a term, so a discount that looked strong at signing is often below market by the second one. The wider commercial framework sits in the Google Cloud agreement negotiation guide.
The Google Cloud CUD negotiation brief
Both discount layers benchmarked, the commitment sizing discipline, the Marketplace aggregation lever, and the renewal moves that protect the rate as you scale.
Get the white paper →Sizing the commitment without destroying the discount
The commitment level is where most of the value is won or lost, and the rule is simple enough to state and hard enough to hold: commit to evidenced steady state, never to peak, and let spikes run at on demand rates.
A commitment sized above real usage means paying for capacity you never consume, and that unused portion can wipe out the discount the commitment earned in the first place, which turns a negotiation win into a net loss that only surfaces at the end of the term.
The form of the commitment matters as much as the level. Resource based commitments carry the deepest rate but tie you to specific machine families and regions, so they belong only where the family is genuinely fixed and the workload has a history of staying put.
Spend based commitments follow workloads across eligible services for a slightly shallower rate, and that flexibility is worth more than the rate gap wherever consumption patterns move, which in practice is most estates above a certain size.
Build the profile before the conversation: separate baseline from spike capacity with evidence rather than assertion, pull a full spend view across compute, services, and Marketplace, then benchmark both layers against your tier.
The data governance side of the same estate sits in the BigQuery cost governance guide, and the practice library in Google Cloud advisory.
- 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 Google Cloud reviews, 2024 to 2025
The standard advice, often reinforced by partners, is that maximising resource based committed use discounts is the surest path to the lowest bill because they carry the deepest published rate. We disagree, because depth bought with rigidity is only a discount while nothing moves:
Share of buyers who worked the published CUD layer hard and accepted the negotiated agreement discount exactly as offered.
Average improvement available once both layers were benchmarked and eligible spend, including Marketplace, was aggregated into the commitment view.
Three patterns recurred: single layer focus, where 60 to 75 percent of buyers negotiated CUDs but accepted the agreement discount as offered; resource lock in, where aggressive resource based commitments stranded 15 to 30 percent of value as workloads shifted machine families or regions.
And the Marketplace miss, where eligible third party spend was left out of the commitment in most early agreements.
The buyer side move is to cover only stable baselines with resource based CUDs, hold volatile workloads on spend based commitments, aggregate every eligible spend stream, and put the real negotiating effort into the agreement layer where the largest unbenchmarked discount sits.
Your first five moves
- Build a steady state usage profile separating baseline from spike capacity, with evidence, because over commitment can wipe out the discount the commitment earned.
- Pull a full spend view across compute, services, and Marketplace, then identify the eligible third party spend that can count toward the commitment and lift you into a better tier.
- Benchmark both layers for your spend tier, the CUD rates and the negotiated agreement percentage, since 60 to 75 percent of buyers benchmarked only the first.
- Cover stable baselines with resource based CUDs and keep volatile workloads on spend based, because resource lock in stranded 15 to 30 percent of value when families or regions moved.
- Re benchmark at every renewal against your current scale, not the scale you signed at, and renegotiate from the larger aggregated base. The Google Cloud practice runs the sizing and the negotiation with you.
Frequently asked questions
How do enterprise Google Cloud discounts work?
Discounting runs on two layers at once. Committed use discounts lower the rate on resources in exchange for a one or three year commitment, and a negotiated agreement applies a further percentage off list against a total spend commitment over the term.
The layers stack, so benchmarking only the published CUD rates understates what is achievable for an estate your size.
What is the difference between resource based and spend based CUDs?
Resource based committed use discounts carry the deepest rate but lock you to specific machine families and regions. Spend based commitments follow workloads across eligible services at a slightly shallower rate.
Use resource based only for stable baselines where the family is genuinely fixed, and spend based wherever consumption patterns move, because flexibility usually outweighs the rate gap.
What is the Google Cloud Enterprise Discount agreement?
It is the negotiated layer that applies a percentage off list against a total committed spend over the term. Crossing a spend threshold unlocks materially better rates, and both the threshold and the discount are negotiable rather than fixed.
This is the layer 60 to 75 percent of buyers in our file accepted exactly as offered, which is why it holds the largest unbenchmarked discount.
Can Marketplace spend count toward my commitment?
Often yes. Eligible third party Marketplace purchases frequently qualify toward the committed spend, and including them can lift you into a higher discount tier.
It was left out of the commitment view in most early agreements we reviewed, which makes it one of the most commonly missed levers in an enterprise Google Cloud negotiation.
How do I avoid over committing on Google Cloud?
Size commitments to evidenced steady state, never to peak, and let spikes run at on demand rates. A commitment above real usage means paying for capacity you never consume, which can wipe out the discount it earned.
Separate baseline from spike capacity with usage evidence before the conversation rather than accepting a growth forecast as the sizing input.
Why do buyers leave Google Cloud discount on the table?
Because they benchmark the visible layer. The published committed use rates are easy to compare, so buyers work them hard, then accept the negotiated agreement discount as presented.
In our reviews the majority under negotiated that second layer, which is exactly where the largest unbenchmarked discount sits and where preparation converts most directly into points.
How do I protect my Google Cloud discount at renewal?
Benchmark the rate against your current scale rather than the scale you signed at. Usage usually grows across a term, so a discount that looked strong at signature can be below market by the second one, and the agreement rate does not improve on its own.
Aggregate the larger spend base, including Marketplace, and renegotiate the tier from it.
Negotiating Google 5: Know What Good Looks Like
Discount benchmark bands by seat count and commit size, the priced alternative that moves the deal desk, the target term sheet, the walk away test, and the CFO memo that closes the side door.
