Reserved Instance Strategy vs EDP Negotiation: Which Saves More

The Three AWS Commitment Vehicles: How They Work

Before comparing Reserved Instances to EDP, it helps to distinguish clearly between the three commitment vehicles available to AWS enterprise customers, because they operate at different levels of the AWS cost hierarchy and serve different purposes.

Reserved Instances (RIs)

Reserved Instances provide a discount on specific EC2 instance types, RDS database instances, ElastiCache nodes, and several other service families in exchange for a 1-year or 3-year usage commitment. Standard RIs lock you to a specific instance type and region; Convertible RIs can be exchanged for different instance families but deliver slightly lower discounts. The discount versus on-demand pricing ranges from 30–40% for 1-year Standard RIs with no upfront payment to 55–62% for 3-year Standard RIs with all-upfront payment. RIs require a level of operational discipline to utilise effectively — if your RI is for an instance type that your engineering team later migrates away from, utilisation drops and you are paying for capacity you do not use.

Savings Plans

Compute Savings Plans replace most of the use cases for EC2 Standard RIs with significantly more flexibility. Instead of committing to a specific instance type, you commit to a dollar-per-hour spend level on compute services (EC2, Lambda, Fargate). The benefit applies automatically to whatever compute you run, regardless of instance family, region, or operating system. Discount rates are slightly lower than Standard RIs (approximately 66% maximum savings for 3-year all-upfront Compute Savings Plans versus 62% for equivalent Standard RIs) but the flexibility improvement is substantial. EC2 Instance Savings Plans offer deeper discounts than Compute Savings Plans but are scoped to a specific instance family within a region. In 2026, the default position for most AWS-centric teams is Compute Savings Plans as the primary vehicle, with RIs reserved for database instances and scenarios requiring capacity reservation.

Enterprise Discount Programme (EDP)

The EDP operates at a different level entirely. Rather than discounting specific services based on instance-level commitments, the EDP applies a percentage discount to your total eligible AWS spend above an annual committed minimum. EDP discounts stack on top of on-demand pricing, but importantly, they also stack on top of Savings Plans and Reserved Instance pricing in certain configurations. The EDP discount is applied to your effective cost — meaning if you are already running at 60% of on-demand cost through RI/Savings Plan coverage, the EDP discount applies to that 60% cost baseline, not the on-demand rate. This stacking behaviour is critical to understanding total effective discounts achievable.

When RIs and Savings Plans Save More Than EDP Alone

For organisations whose AWS spend is concentrated in a small number of services and is relatively stable, RIs and Savings Plans typically deliver better unit economics than a general EDP discount at equivalent commitment levels. A 3-year Standard RI on an r6i.4xlarge RDS instance delivers approximately 58% discount versus on-demand. An EDP at the $1M tier delivers approximately 10–12%. For a $500K/year RDS spend, the RI generates $290K in annual savings; the equivalent EDP discount generates $50–60K. The RI wins decisively.

The trade-off is that RIs and Savings Plans are service-specific commitments with real utilisation risk. If the RDS workload migrates to Aurora Serverless or is right-sized to a different instance family before the RI term expires, utilisation drops and the economic benefit erodes. FinOps teams that actively manage RI portfolios — monitoring utilisation monthly, exchanging Convertible RIs as architectures evolve, and applying Savings Plans to cover flexible compute — can maintain utilisation above 90% and capture most of the theoretical discount. Teams that set and forget RI purchases regularly see utilisation fall below 70%, at which point the actual savings look more like 15–20% rather than the 40–60% on paper.

This management overhead is precisely why EDP becomes relatively more attractive as AWS spend grows and diversifies across more services. An organisation spending $15M annually across 40+ AWS services faces a complex RI/Savings Plan portfolio that requires dedicated management. An EDP at that scale delivers a meaningful portfolio-level discount with no utilisation risk on the discount itself (the risk is commitment shortfall risk, which is managed differently). Our guide to enterprise FinOps governance covers the team structures and tooling that support portfolio-level commitment management at scale.

When EDP Saves More Than RIs Alone

The EDP's advantages become most pronounced in three specific scenarios. The first is high spend diversity: when AWS spend is distributed across services that do not have dedicated RI/Savings Plan vehicles — Glue, Kinesis, SQS, API Gateway, CloudFront, Secrets Manager — an EDP discount applies to all eligible services simultaneously, including those that have no RI equivalent. This breadth advantage is significant for modern microservices architectures where compute is only one component of the cost base.

The second scenario is rapid growth or architectural change. If engineering plans show major infrastructure migrations, new service adoptions, or significant headcount growth in the next 12–18 months, committing to specific instance types via RIs carries real obsolescence risk. An EDP commitment at the right level captures portfolio-level savings without betting on specific infrastructure choices that may change. The commitment is to a spend level, not a technology configuration.

The third scenario is Marketplace offset eligibility. For organisations with significant AWS Marketplace ISV software spend — security, monitoring, data platforms — the ability to offset up to 25% of EDP commitment attainment with Marketplace purchases reduces the effective risk of the EDP commitment materially. There is no RI equivalent for Marketplace software spend. For more on using FinOps data to maximise this offset, see our analysis of FinOps integration with AWS negotiations.

The Optimal Combination: A Framework

For most enterprise AWS customers above $2M annual spend, the highest-value strategy is a deliberate combination of RIs/Savings Plans for service-level commitment optimisation and EDP for portfolio-level commercial terms. The framework for finding the right combination has four steps.

Step 1: Identify the RI/Savings Plan Opportunity Set

Using FinOps tooling, identify all on-demand compute and database spend that is consistent and predictable — workloads running 24/7 or with a predictable monthly runtime profile. Size Compute Savings Plans to cover 70–80% of eligible compute spend (leaving 20–30% on-demand to absorb growth and spikes). Size Standard or Convertible RIs to cover consistent RDS, ElastiCache, and Redshift usage. Calculate the annual cost basis after applying these commitments — this is the effective spend base that the EDP will be sized against.

Step 2: Calculate Residual On-Demand and EDP-Eligible Spend

The residual on-demand spend — after RI/Savings Plan coverage — plus all service spend without RI equivalents defines the population of spend that an EDP discount will apply to. This is the number procurement should use when modelling EDP commitment levels and expected discount value. Many organisations mistakenly model EDP discount value against total spend, ignoring that a substantial portion of that spend is already discounted via commitments, producing an inflated projected EDP benefit that does not materialise.

Step 3: Size the EDP Commitment Against Forecast Residual Spend

Apply the same scenario-based commitment sizing framework used for RIs — conservative, base case, and aggressive — to the residual spend base. The EDP commitment should reflect what you expect to spend on EDP-eligible services after RI/Savings Plan coverage is applied. Shortfall risk is calculated against this residual base, not total AWS spend. Our dedicated piece on using FinOps data as EDP negotiation leverage covers the commitment sizing methodology in detail.

Step 4: Monitor Interaction Effects Monthly

Once both commitments are in place, FinOps tooling should track three metrics monthly: RI and Savings Plan utilisation rates, EDP commitment attainment run-rate, and effective overall discount rate versus baseline on-demand. The effective overall discount rate — total savings divided by what the same workloads would cost at on-demand prices — is the ultimate measure of how the strategy is performing. In well-managed organisations with mature RI portfolios and appropriately sized EDPs, effective overall discounts of 40–70% versus on-demand are achievable. Organisations that manage only one layer (RIs without EDP, or EDP without RIs) typically achieve 20–35%.

The 2026 Policy Change: Single End Customer Restriction

From June 1, 2025, AWS limited Reserved Instances and Savings Plans to a "single end customer" — restricting the ability of some organisations to pool or redistribute commitment benefits across legally distinct entities such as subsidiaries, portfolio companies, or third-party customers. For enterprises that previously relied on commitment pooling strategies across entities under a shared AWS Organisation, this change requires a review of the commitment structure to ensure compliance and that the intended discount benefits are still being realised. FinOps teams should verify their current pooling model against the new policy and, if adjustments are needed, address this before the next RI or Savings Plan purchase cycle. For the broader context of how these governance decisions connect to enterprise software licensing strategy, see our analysis of FinOps for enterprise software licensing.

Common Mistakes That Reduce Combined Savings

The most expensive mistake is purchasing RIs or Savings Plans independently of EDP sizing, then discovering that the EDP commitment was sized against a spend base that RI purchases will significantly reduce. This creates a situation where RI savings come at the cost of EDP shortfall risk. The sequencing matters: model RI/Savings Plan coverage first, then size the EDP against residual spend. Do not size the EDP first and then try to layer in RIs. Our related piece on OCI FinOps frameworks shows how equivalent sequencing applies when Oracle Cloud Infrastructure is part of the multi-cloud portfolio.

The second common mistake is treating RI utilisation as a fire-and-forget metric. RI utilisation that falls below 80% erodes savings faster than most procurement teams appreciate. A 3-year all-upfront RI running at 70% utilisation is costing you money compared to on-demand in years two and three, not saving it. Monthly FinOps review of RI utilisation, with an escalation path for utilisation below 85%, prevents this from happening silently.

The third mistake is not using the combination analysis as negotiation leverage with AWS. When procurement can demonstrate that they are managing a sophisticated, well-utilised RI portfolio — showing 90%+ utilisation across a $2M RI investment — that evidence supports a request for improved EDP terms. AWS account teams know that disciplined customers with high RI utilisation are lower churn risk, which is worth something in EDP pricing. Arrive at the EDP negotiation without that evidence and you are a number on a spreadsheet. Arrive with it and you are a preferred customer deserving of preferred pricing. Redress provides AWS contract negotiation specialists who model this analysis and lead AWS EDP negotiation discussions on your behalf.