Marketplace spend ran past the 25 percent contribution cap on half the estates, and the cap itself is negotiable
Commit a dollar volume, receive a percentage off, accept the flexibility limits, and live with the renewal. Most of the buyer side attention goes to the first two. The line that moved most money in our file was the one routing third party software through the same commitment.
Prepared by Redress Compliance · August 16, 2026 · AWS advisory. 45 to 60 EDP negotiations and renewals benchmarked, 2024 to 2025.
Executive summary
Marketplace pass through was the largest under audited line on most estates. It ran above 25 percent of total commit on roughly half the estates benchmarked, and renegotiating the cap explicitly returned 5 to 12 percent on routed SaaS spend.
The 25 percent contribution cap is a mechanism, not a discount, and it should never be blended into the discount ladder when you model the deal. It governs how much third party spend counts toward commitment.
Carry forward clauses were available on roughly six in ten enterprise agreements when requested and were never offered by default. Rolling unused commit into the next contract year is the cheapest protection in the document.
AWS proposed commit sizing exceeded trailing twelve month draw by 22 to 38 percent. Sizing on trailing run rate plus a defensible 8 to 14 percent growth band landed materially better, and over commit on the forecast averaged 18 to 28 percent of total contract value.
How the commitment model works
An EDP commits a minimum dollar spend across the AWS organization over three or five years, applying to most services and most regions, with the discount tier scaling by commitment size. Cross account aggregation expands eligibility across member accounts.
| Element | How it behaves | Negotiable |
|---|---|---|
| Commitment figure | Minimum dollar spend across the term | Yes, on both size and ramp shape |
| Discount tier | Steps by commitment threshold, 5 to 20 percent | By moving a threshold, not within a band |
| Marketplace contribution | Third party spend counting toward commit, capped at 25 percent | Yes, the cap itself is negotiable |
| Carry forward | Unused commit rolling into the next contract year | Yes, granted about six times in ten when asked |
| Savings Plans and Reserved Instances | EDP layers on top of existing commit mechanisms | Structure, not the layering itself |
Keep the two mechanisms separate when you model. The discount ladder steps by commitment threshold and runs 5 to 20 percent: roughly 5 to 10 percent between $1m and $5m, 10 to 15 percent between $5m and $25m, and 15 to 20 percent above $25m. The marketplace contribution cap is a different thing entirely, governing how much third party software spend counts toward the commitment. Blending the two produces a model that overstates the discount and understates how much of the commitment is being satisfied by software you were buying anyway.
The marketplace line, and why it goes unaudited
- Marketplace pass through ran above 25 percent of total commit on roughly half the estates benchmarked, which means the routed SaaS spend exceeded what the cap allowed to count toward the commitment.
- Renegotiating the cap explicitly returned 5 to 12 percent on the routed SaaS spend, and the conversation is available because the cap is a negotiated term rather than a published constant.
- It is under audited because it sits between two owners. The software is procured by application teams and the routing is a cloud finance decision, so neither side carries the full picture into the negotiation.
- Routing software through marketplace is not automatically wrong. It satisfies commitment with spend you had already committed to elsewhere, which is genuinely useful when the cap accommodates it.
- Audit which SaaS is routed and at what margin before the renewal, since the reseller economics differ by vendor and the pass through is rarely priced against a direct alternative.
- Model the cap against your actual third party estate, not against a percentage in a template, because the right cap depends on how much software you genuinely intend to route.
The AWS EDP negotiation guide
Commitment sizing, discount benchmarks by threshold, flexibility provisions, and buyer side moves across the full AWS spend estate.
Get the brief →A commitment is a floor, and the forecast that sets it is not yours
The mechanics of an Enterprise Discount Program are simple enough to state in a sentence, which is part of why the negotiation goes wrong. Commit a volume, receive a percentage. What that sentence hides is that the commitment is a floor rather than a budget: you pay it whether or not you consume it, and the discount is applied to consumption rather than to the commitment. So an over sized commitment does not buy a bigger discount in any useful sense. It buys an obligation.
That is where the sizing finding matters. AWS proposed commit sizing routinely exceeded trailing twelve month draw by 22 to 38 percent, built from forward forecasts rather than from realised consumption. Commits sized on the trailing run rate plus a defensible 8 to 14 percent growth band landed materially better, and over commit on the forecast averaged 18 to 28 percent of total contract value. On a large agreement that is a very substantial sum paid for nothing, and it is paid in exchange for a discount step that is worth a few points at most. The arithmetic almost never favours buying the next threshold with capacity you do not expect to use.
The marketplace line is the counterweight, and it is the most interesting mechanism in the document because it works in the buyer's favour when it is understood. Third party software routed through AWS Marketplace counts toward the commitment, up to a cap, which means spend you were already committed to elsewhere can help satisfy an obligation you have taken on. Half the estates benchmarked were running past that cap, meaning they had more eligible spend than the agreement permitted to count. Renegotiating the cap explicitly returned 5 to 12 percent on routed SaaS. It goes unaudited because it sits between two owners: application teams buy the software and cloud finance owns the routing, so neither carries the whole picture into a negotiation.
Flexibility provisions complete the picture and follow the same pattern as governance clauses generally. Carry forward, rolling unused commit into the next contract year, was available on roughly six in ten enterprise agreements when requested explicitly and was never offered by default. True up treatment and over commit protections are individually negotiable and individually absent from the template. None of these change the discount, and all of them change what happens when the forecast turns out to be wrong, which it usually does. The private pricing instrument itself is covered in the PPA pillar, the sizing arithmetic in EDP negotiation, and the wider library in the AWS practice.
- AWS, Azure, GCP, and OCI exports analysed: savings by workstream with a phased plan
- Commit sizing from real usage data, and the same workload priced across all four clouds
- Invoice lines matched daily against what the contract says you owe
The flexibility provisions that are absent by default
| Provision | What it does | Availability when requested |
|---|---|---|
| Carry forward | Rolls unused commit into the next contract year | About six in ten enterprise agreements |
| Over commit protection | Limits exposure when consumption falls short | Individually negotiable, never default |
| True up treatment | Sets how and when a shortfall is assessed | Individually negotiable |
| Marketplace cap | Governs third party contribution toward commit | Negotiable, worth 5 to 12 percent on routed spend |
| Cross account aggregation | Expands eligibility across member accounts | Usually available, rarely modelled |
Every row is a term that changes what happens when the forecast is wrong, and none of them alters the headline percentage. That is why they are conceded relatively readily and why they are missing until somebody asks.
What the EDP negotiations showed, 2024 to 2025
Across roughly 45 to 60 AWS EDP negotiations and renewals benchmarked, three patterns recur:
How far AWS proposed commit sizing exceeded trailing twelve month draw, built on forward forecasts rather than realised consumption.
Share of total contract value lost where the commitment was sized on an AWS forecast rather than on trailing run rate plus a defensible growth band.
Marketplace pass through ran above 25 percent of total commit on roughly half the estates, and renegotiating the cap explicitly returned 5 to 12 percent on routed SaaS spend. Carry forward clauses were available on roughly six out of ten enterprise agreements when requested and never offered by default.
Commits sized on the trailing twelve month run rate plus a defensible 8 to 14 percent growth band landed materially better than commits sized on AWS forward forecasts. The discipline is to make the growth band something you can evidence rather than something you hope for.
Watch the briefing · 4:15Negotiating AWS 8: Size and Structure the CommitFixing the measurement basis before the number, back weighting the ramp, and naming Available Balance and Rollover Amount.
Your first five moves
- Size the commitment on trailing twelve month draw plus a growth band you can evidence, at 8 to 14 percent, rather than on the AWS forward forecast.
- Audit routed marketplace spend against the cap and establish how much eligible third party software exceeds what the agreement permits to count.
- Negotiate the marketplace cap as a named term, since doing so returned 5 to 12 percent on routed SaaS in the estates benchmarked.
- Ask for carry forward, over commit protection, and true up treatment by name, because none appear in the default template and about six in ten agreements granted carry forward when requested.
- Model the threshold rather than the band, and refuse to buy a step with capacity you do not expect to consume. The AWS practice sizes the commit with you.
Frequently asked questions
What is an AWS Enterprise Discount Program?
A private pricing agreement committing a minimum dollar spend across the AWS organization over three or five years in exchange for a percentage discount across most services and regions. The discount tier scales with commitment size, and cross account aggregation expands eligibility across member accounts.
What discount should we expect?
Between 5 and 20 percent, stepped by commitment threshold: roughly 5 to 10 percent between $1m and $5m, 10 to 15 percent between $5m and $25m, and 15 to 20 percent above $25m. Moving a threshold is worth more than arguing within a band.
Does marketplace spend count toward the commitment?
Yes, up to a contribution cap of 25 percent, and that cap is a negotiable term rather than a published constant. It is a separate mechanism from the discount ladder and blending the two produces a model that overstates what you are actually getting.
Why is marketplace the most under audited line?
Because it sits between two owners. Application teams procure the software and cloud finance owns the routing decision, so neither side carries the full picture into the negotiation. Half the estates benchmarked were running past the cap without having modelled it.
What is renegotiating the cap worth?
Between 5 and 12 percent on the routed SaaS spend in the estates we benchmarked. The right cap depends on how much third party software you genuinely intend to route, which is why it should be modelled against your actual estate rather than accepted from a template.
How should the commitment be sized?
On the trailing twelve month run rate plus a defensible 8 to 14 percent growth band. AWS proposed sizing exceeded trailing draw by 22 to 38 percent, built on forward forecasts, and over commit on those forecasts averaged 18 to 28 percent of total contract value.
Is a bigger commitment worth the deeper discount?
Usually not, if the extra capacity will go unused. The commitment is a floor you pay whether or not you consume it, and the discount applies to consumption. Buying the next threshold with capacity you do not expect to use trades a certain obligation for a few points.
What is a carry forward clause?
A provision rolling unused commit into the next contract year. It was available on roughly six out of ten enterprise agreements when requested explicitly, and it was never offered by default. It is the cheapest protection in the document against a forecast that turns out wrong.
Does EDP stack with Savings Plans and Reserved Instances?
Yes, EDP layers on top of those existing commitment mechanisms rather than replacing them. What is worth negotiating is the structure and interaction rather than the layering itself, which is standard.
Which flexibility provisions are missing by default?
Carry forward, over commit protection, and true up treatment. All three are individually negotiable, none appear in the default template, and none change the headline percentage. They change what happens when the forecast is wrong, which is why they matter more than they look.
Negotiating AWS 5: Know What Good Looks Like
Published benchmark tables disagree by a factor of three. The one verifiable AWS discount is 9 percent, filed with regulators. Plus the bands, the breakpoints, competition worth 3 to 8 points, and effective against headline rate.
