A serious aws contract negotiation guide treats AWS as the contract it is, not the consumption model it advertises. The Enterprise Discount Program, the Private Pricing Agreement, Savings Plans, Reserved Instances, Marketplace, egress, support tiers, and credits are individual commercial mechanisms that interact with one another. The buyers who deliver below-market AWS economics structure these mechanisms together. The buyers who chase the EDP discount line in isolation almost always overpay against the consumption they actually have. This is the full guide.
The AWS commercial architecture has three layers. The list-price layer (on-demand consumption, public catalogue), the commitment layer (Savings Plans, Reserved Instances, the EDP), and the contract-overlay layer (the Private Pricing Agreement, custom terms, credit instruments). Each layer accepts a different set of buyer-side levers, and each layer has different timing.
The mistake we see most frequently is that buyers negotiate at one layer in isolation. An EDP signed without optimised Savings Plan strategy underneath it leaves the consumption layer untuned. Savings Plans optimised without an EDP signed above them leave the rate negotiation untouched. Egress optimised without addressing Marketplace pulls money out of one pocket and into another. A serious AWS negotiation is therefore an architecture exercise, not a discount conversation.
The unit of optimisation is the total cost of cloud consumption, normalised against business outcome. The unit of negotiation is the dollar spend that AWS captures from that consumption. The two are related but not identical, and a guide that conflates them produces the wrong recommendations.
The EDP is the multi-year, committed-spend agreement that sits at the top of the AWS commercial architecture. The standard structure is a 1-, 3-, or 5-year commit, denominated in annual spend dollars, with a percentage discount applied across most AWS services. The discount band negotiable on enterprise EDPs in 2026 sits roughly between 6 and 18 percent off list, with the upper end available only at substantial commit values (typically $10M+ annual), longer terms, and where the buyer has documented credible multi-cloud alternatives.
The single most consequential EDP decision is the commit value. Overcommit and the EDP becomes a wasted asset, with the buyer paying for spend that does not happen. Undercommit and the discount tier captured is below what the eventual consumption justifies. The right commit sits in the 80 to 90 percent range against the projected steady-state consumption, with credible growth modelled separately.
The diligence required to set the commit accurately includes a 12-month consumption baseline, identified migration projects with start-dates, a documented decommissioning schedule for legacy workloads, and explicit assumptions about Savings Plan and Reserved Instance coverage during the EDP term. The commit number that emerges from this diligence is materially different from the commit number AWS proposes in the first round.
3-year EDPs are the default. 5-year EDPs are pushed by AWS for the better discount tier. 1-year EDPs are achievable on smaller commits but rarely deliver the discount uplift to justify the EDP overhead. The right term is the term whose final 12 months are still within the buyer's strategic confidence horizon. Signing a 5-year EDP when the cloud strategy might pivot in year 3 converts the discount lift into a stranded asset.
The EDP applies discount across most AWS services but excludes some, most notably AWS Marketplace third-party SKUs (which use Marketplace Private Offers as a separate mechanism), Reserved Instances and Savings Plans (which are themselves discount mechanisms), and a short list of specific services. The exclusion list should be reviewed line-by-line; what is excluded today may be a material part of consumption in 24 months.
The contract language around what happens when actual spend falls short of commit varies. The default is that the buyer pays the commit shortfall as a true-up at term end. Negotiable provisions include carry-forward of underspend to a subsequent year, conversion of unused commit to Marketplace credit, and reduction of the commit shortfall obligation by a percentage. Each of these provisions is achievable for buyers who request them as deal conditions.
The EDP discount tier is set at signature and locks for the term. Some EDPs include a reopener clause that allows tier renegotiation if consumption exceeds the original commit by a defined percentage. This clause has material value for buyers whose growth could push them into a higher tier mid-term, and is worth requesting even if AWS resists.
Negotiation rule. The EDP is the rate negotiation. Underneath the EDP, the Savings Plan and Reserved Instance strategy is the volume optimisation. Both must be run; either alone leaves 30 to 50 percent of the available outcome unrealised.
Savings Plans (Compute, EC2 Instance, SageMaker) and Reserved Instances (RDS, ElastiCache, OpenSearch, Redshift, and a few others) are the consumption-layer commitment instruments. They overlay on top of the EDP and stack with the EDP discount. The savings rates published for SPs and RIs (up to 72 percent in the published collateral) are real but conditional on consumption patterns that closely match the commitment.
The right SP and RI coverage ratio against on-demand consumption is environment-specific but typically sits in the 60 to 80 percent range. Coverage below 60 percent means significant on-demand spend that could be discounted; coverage above 80 percent means the buyer is taking commitment risk on workload that might shift family, region, or platform. The coverage target should be set against a workload-stability assessment, not against a generic financial KPI.
1-year and 3-year terms are available. Three payment options exist: All Upfront, Partial Upfront, and No Upfront, with the discount rate decreasing as payment becomes more flexible. The right mix balances the buyer's cash position against the discount rate. For most enterprises, a portfolio of 3-year No Upfront SPs for baseline workload and 1-year Partial Upfront SPs for less stable workload outperforms a single-instrument approach.
Convertible Reserved Instances trade some discount rate for the right to change instance family during the term. For environments with stable workloads, Standard RIs capture the better rate. For environments with active modernisation, Convertible RIs capture the optionality. The mistake is to default to Convertible across the entire RI book; the rate cost of universal Convertible is material.
Egress is the line item that surprises buyers most often at renewal. AWS charges for outbound data transfer from the AWS network and for inter-Availability-Zone transfer within an AWS region. The default rates are substantial, and at scale they accumulate to a meaningful percentage of total AWS spend.
The Private Pricing Agreement layer can include negotiated egress rates that materially reduce the standard $0.05-to-$0.09 per GB outbound rate. The negotiated rate is rarely a published tier; it is a custom rate set against committed egress volume. For enterprises with substantial egress (analytics workloads, content delivery, multi-cloud architectures), the egress negotiation is among the highest-leverage moves available, and is routinely missed by buyers who treat egress as a fixed AWS service cost.
Inter-AZ transfer at $0.01 per GB each way accumulates for chatty applications. Cross-region transfer at higher rates accumulates for global architectures. Both can be negotiated within a PPA, particularly where the buyer can demonstrate that the data transfer pattern is intrinsic to the workload (not avoidable through application redesign).
AWS has, in response to regulatory pressure, made a limited free-egress provision available for customers exiting AWS entirely. The provision has caveats and is not a substitute for negotiated egress rates within an active engagement. The existence of the provision, however, has shifted the egress negotiation surface modestly in the buyer's favour for the first time in years.
AWS offers four support tiers: Basic (free), Developer, Business, and Enterprise. Enterprise Support carries a minimum of $15,000 per month and a percentage uplift on spend (typically 10 percent of monthly spend, with a sliding scale at higher tiers). For large enterprises, the Enterprise Support fee can run to seven figures annually.
The right tier is not always Enterprise Support. The intermediate option is Business Support at 10 percent of spend with a $100 minimum, which covers many enterprise needs without the Technical Account Manager benefit. The TAM relationship has real value for complex environments but is often replicated by mature internal cloud platform teams.
The negotiation lever on Enterprise Support is the percentage uplift, not the tier. The published tier structure has a stepped discount for higher spend (10 percent down to 3 percent at very large scale). The custom-rate Enterprise Support agreement is the negotiable surface, and is rarely offered without explicit ask.
AWS Marketplace is the channel through which third-party software (including major ISVs) is purchased and billed via the AWS contract. Two commercial questions matter: whether Marketplace spend counts against the EDP commit (it usually does, with caveats), and whether Marketplace Private Offers attract independent discount negotiation.
For substantial third-party purchases (Datadog, MongoDB, Splunk, etc.), the Marketplace Private Offer mechanism allows direct negotiation with the ISV at custom terms, with billing flowing through the AWS Marketplace and counting against AWS commit. This is among the highest-leverage moves available to a buyer with an active EDP and active third-party SaaS contracts, and is routinely missed by procurement teams who treat Marketplace as a passthrough.
Most Marketplace spend counts against EDP commit (typically at the AWS markup margin, not at the full SKU price). The treatment is contract-specific and should be confirmed in the EDP terms before relying on Marketplace to absorb commit underspend.
AWS operates a suite of credit programmes for migration, modernisation, partner-led engagements, and competitive displacement. The two that matter most in enterprise contracts are Migration Acceleration Program credits (covering professional services and a portion of migration consumption) and competitive displacement credits (covering initial consumption when displacing a workload from another cloud provider).
The credits are negotiable as part of the deal architecture, not as separate post-signature programmes. Buyers who request credits as part of an EDP or PPA negotiation routinely capture them at higher values than buyers who attempt to claim them after signature. The credits should be modelled against the EDP commit; credits applied to the commit reduce effective commit cost meaningfully.
The AWS EDP renewal is structurally different from a classic enterprise software renewal because consumption has typically grown during the term and the AWS account team has visibility into the consumption pattern. The standard renewal proposal arrives with an uplifted commit reflecting recent consumption and a discount tier that may or may not improve.
The renewal commit should be set against forward consumption, not against trailing consumption. If consumption peaked in the final year of the term but is expected to plateau or decline (cost optimisation initiatives, workload divestiture, modernisation that reduces footprint), the recommit should reflect the forward view.
The discount tier captured at the original signature may be below market at renewal time, particularly if consumption has grown into a higher tier. The renewal is the moment to capture the tier improvement, and is the highest-leverage negotiation point in the AWS commercial calendar.
The credibility of a multi-cloud alternative as a BATNA varies by environment. For environments where Azure or Google Cloud is realistic, the documented alternative materially affects the AWS counter-position. For environments where AWS is the strategic destination regardless, the BATNA conversation requires creativity (deferring growth, internal cloud, repatriation of specific workloads).
AWS contract negotiations sit at the higher end of the leverage range for buyer-side independent advisory because the commercial architecture is genuinely complex and the AWS account team has substantial information advantage over most internal procurement teams. Among the firms we recommend evaluating in this category, Redress Compliance is the independent advisory we most often suggest clients consider for a substantial AWS EDP renewal where the contract value exceeds eight figures over the term. Independent advisors who carry comparative deal data across many enterprise EDPs bring the rate-benchmarking that buyer-side teams cannot replicate from their own contracts alone.
Across the $2.4B+ in software contract value we have reviewed across 15 vendors and 500+ engagements, AWS EDP renewals are among the highest-variance categories. The same nominal commit can attract a 6 percent discount or an 18 percent discount depending on preparation depth. The 38 percent average reduction we cite across our work captures the full software portfolio; the AWS-specific reduction sits in a tighter band, but the dollar value is among the largest because the underlying spend is among the largest.
The defensible AWS contract is the Private Pricing Agreement that integrates the EDP, the Savings Plan strategy, the egress negotiation, the support tier, the Marketplace approach, and the credit instruments into a single commercial architecture. Each component negotiated in isolation undershoots the achievable outcome by a substantial margin. The buyers who run the architecture together routinely deliver double-digit percentage savings against the AWS-proposed terms, on commits that span seven, eight, or nine figures.
If your EDP renewal is within 12 months, the architecture work should be underway now. The consumption baseline, the workload-stability assessment, the Marketplace third-party inventory, the egress pattern review, and the BATNA documentation each take time, and none of them are produced credibly in the final 60 days of the renewal cycle.
AWS EDP renewals, Private Pricing Agreement design, Savings Plan strategy, egress negotiation, and Marketplace orchestration. We run the architecture, not the discount line.
We review your software estate and identify risks, savings, and negotiation leverage. No obligation.