Learn how to automate AWS Savings Plans and Reserved Instances in six steps, with guardrails, native AWS options, and Harness Commitment Orchestrator.
TL;DR
- Shrink first, then commit. A commitment sized on idle or oversized resources locks that waste in for one to three years.
- Buy in tranches. AWS states that Savings Plans terms cannot be changed after purchase, and returns are allowed only in a narrow window. One large purchase turns a forecasting error into a long-term cost.
- Set guardrails first: an approval step, a cap on commitment size, and an owner for each commitment.
- AWS tells you what to buy and gives you the APIs to buy it. You still need a process that decides, approves, tracks, and rebalances. Commitment Orchestrator runs that process for AWS commitments. Approval is the default, and full automation is an option you choose.
Short Answer
To automate AWS Savings Plans and Reserved Instances (RIs), remove waste first, measure your steady usage, set a coverage target, buy commitments in smaller tranches, track utilization and expiry, rebalance as usage changes, and report results by owner. Add an approval step, a cap on commitment size, and a named owner for every commitment so automation cannot over-commit. You can script the loop with AWS APIs, or use Harness Commitment Orchestrator, part of the Harness Cost Management Agent.
Savings Plans vs Reserved Instances
Both give you a lower rate in exchange for a one or three year commitment. Savings Plans commit you to an hourly spend. RIs commit you to a specific instance configuration. AWS now offers four Savings Plans types, and each has its own maximum discount.
AWS's discount figures are maximums. Actual savings depend on the plan type, term, payment option, and workload.
Two coverage details trip up Kubernetes and database teams. Compute and EC2 Instance Savings Plans cover the EC2 instances under EKS, ECS, and EMR clusters, but not EKS charges themselves. Database Savings Plans and RDS RIs cannot be combined on the same workload, though you can use one for one workload and the other for another.
Sources: Savings Plans types, Returning a purchased Savings Plan, Reserved Instance offering classes, EC2 Reserved Instances, Database Savings Plans announcement, Database Savings Plans pricing. New to the basics? Start with What is an AWS Savings Plan?.
Before You Commit: Remove Waste First
A commitment is a promise to keep spending. If your baseline includes idle or oversized resources, you lock them in.
AWS's own data supports the order. In The AWS State of Cost Efficiency Report (June 2026, more than 71,000 opted-in customers), larger customers who combined Savings Plans with rightsizing improved their median Cost Efficiency score 4x faster than customers using Savings Plans alone. Only 47.1% of customers with Savings Plans coverage were actively rightsizing. AWS also notes that buying commitments first means rightsizing later frees up commitment instead of lowering your bill.
Shut down idle non-production resources and right-size workloads first, then size commitments on what remains. The Harness Cost Management Agent (CMA) covers both. AutoStopping shuts down idle non-production resources and restarts them on demand. Compute and Kubernetes recommendations flag oversized workloads. Kubernetes cost allocation and Cost Categories show spend by cluster, workload, and team. For more tactics, see AWS Cost Optimization: 15 Tactics to Cut Your Bill This Quarter.
Six Steps to Automate Commitment Management
1. Inventory Usage and Coverage
Pull hourly or daily usage by account, service, and instance family. Record what existing Savings Plans and RIs already cover, and what runs at On-Demand rates. AWS exposes coverage and utilization through Cost Explorer, the Cost and Usage Report, and APIs such as GetSavingsPlansCoverage and GetReservationCoverage.
In Harness: Commitment Orchestrator shows your existing RIs and Savings Plans in an inventory view with status filters. Compute coverage can be viewed by cost or by hours, broken out by purchase option: On-Demand, Savings Plans, RIs, and Spot. You can report on a Net Amortized or Unblended basis, and filter by account, Region, or instance family. Savings and utilization for the commitments Harness manages are reported alongside.
2. Set a Coverage Target
Pick a target for how much of your steady usage commitments should cover. Leave headroom for usage that can drop. Set the target by workload type, not as one number for the whole bill. Do not copy a percentage from another company. Your usage pattern decides it.
In Harness: Target Coverage is a setup setting. It is the maximum percentage of compute spend you want covered by Savings Plans or RIs, and any remainder stays On-Demand. You can also choose whether the orchestrator prefers RIs or Savings Plans when both can cover the same usage, and exclude specific accounts, Regions, or instances from orchestration.
3. Buy in Tranches
Split purchases over time instead of making one large buy. Tranches spread term end dates, so a single expiry does not leave you exposed, and they let you correct course as usage changes. This matters most for Savings Plans. Outside the seven day return window, plan on living with what you bought. AWS's own guidance is to buy in smaller, more frequent increments and raise coverage as you rightsize. Its Savings Plans Purchase Analyzer has a target coverage feature for layering purchases over time.
AWS gives you the inputs. Cost Explorer and the GetSavingsPlansPurchaseRecommendation API generate Savings Plans recommendations from a 7, 30, or 60 day look-back, and RI recommendations use the same windows (reservation recommendations). The Savings Plans API lets you create a plan with a future purchase time, which queues the purchase, and deletes a queued purchase before it runs.
In Harness: For RIs, Commitment Orchestrator uses Atomization. It buys a small "atom" RI each month in each Region, so a slice of your coverage expires every month instead of everything landing on one date. The default atom term is one year, and you can set three. Where Harness SP Staggering is enabled, new Savings Plan coverage is built through a series of smaller purchases, with a balanced default pace that covers about 50% of coverable usage in the first month. A maximum hourly commitment setting caps the size of any single new Savings Plan.
4. Track Utilization and Expiry
Watch utilization of each commitment. Low utilization means you are paying for capacity you do not use. Track expiry dates and decide ahead of time whether to renew, resize, or let a commitment lapse. AWS can alert you ahead of Savings Plans expirations and queued purchases (Savings Plans alerts). Assign each alert to an owner.
In Harness: The inventory view shows status for each commitment, and utilization trends are reported across Savings Plans and RIs. Notifications cover successful purchases, recommendations waiting for approval, and Savings Plans nearing expiry (up to seven days ahead), plus an optional daily summary email.
5. Rebalance
Usage shifts as teams migrate, scale, or retire services. Review coverage on a fixed schedule and after major changes. Add commitments where coverage is thin. Know what each commitment type allows. You can modify Standard and Convertible RIs, and exchange Convertible RIs for another Convertible RI of equal or higher value. Savings Plans terms cannot be changed, so rebalancing means buying new coverage or letting plans expire.
In Harness: Commitment Orchestrator analyzes coverage and recommends purchases, exchanges, and renewals as usage patterns change. Savings Plan purchase recommendations arrive monthly, and RI recommendations arrive several times a month. Each recommendation expires if it is not approved within 24 hours, and new ones are generated as spend changes. AWS recommendation windows top out at 60 days. Harness forecasts compute spend with machine learning and an extended look-back, so a single unusual week matters less. For expiring Savings Plans, the optional Roll-Down Policy sets what percentage renews as a Savings Plan, with the remainder converted to RIs. At 80%, a $10 per hour plan renews $8 per hour as a Savings Plan and moves $2 per hour to RIs.
6. Report
Report coverage, utilization, and realized savings by team or business unit. Finance and engineering leaders need the same numbers. Tie each commitment to a cost owner.
In Harness: Commitment Orchestrator reports savings, spend breakdown, coverage, and utilization in the dashboard and through its API. Cost Categories group spend by team and business unit across the rest of Cost Management, so commitment results can be read in the same terms Finance and engineering already use.
Guardrails: Keep Automation From Over-Committing
Automation fails when nobody owns the result. Set these before you turn anything on:
- Approval step. Decide which purchases a person must approve, and who that person is.
- Commitment size cap. Set a maximum hourly commitment for any single new Savings Plan.
- Gradual buying. Spread purchases over time so you can adjust if usage drops.
- Named owners. Assign one owner to each commitment. That person approves renewals and answers for low utilization.
- Exclusions. Keep accounts, Regions, or workloads you plan to migrate or retire out of scope.
- Utilization review. Set a utilization level that triggers a pause on new purchases and a review.
- Audit trail. Log every recommendation, approval, and purchase.
In Harness: The Cost Management Agent lets you choose the autonomy level for each kind of action: Recommend, Approve, or Autonomous. In Commitment Orchestrator, that choice shows up as the orchestration mode. In Manual mode, every purchase needs explicit approval, and recommendations appear in the Actions tab. In Fully Automated mode, purchases run without manual approval, based on the preferences you set. Automation is a setting you choose, not the default. Around that, you can set a maximum hourly commitment for any single new Savings Plan recommendation, set a target coverage level, and exclude accounts, Regions, and instances. Role-based access control (RBAC), Open Policy Agent (OPA) policies, and audit trails support compliance, and notifications flag purchases and pending approvals. Purchases happen at the management account level and apply to child accounts.
Native AWS Tools vs Harness Commitment Orchestrator
AWS gives you strong building blocks. Cost Explorer and Cost Optimization Hub show purchase recommendations. The Hub covers Savings Plans (Compute, EC2 Instance, and SageMaker AI) and Reserved Instances across services such as EC2, RDS, ElastiCache, and Redshift (Cost Optimization Hub). APIs and the CLI let you script purchases (programmatic Savings Plans management). Queued purchases and expiration alerts help with timing. AWS Compute Optimizer handles rightsizing and idle resources, not commitments.
What AWS does not provide is an automated decision engine that sets coverage, sizes purchases, and rebalances over time. Your team builds the schedule, approvals, tracking, and rebalancing around AWS's recommendations, and that is where most manual processes break down. Commitment Orchestrator runs that workflow for AWS commitments, next to the cost data and governance controls teams already use to remove waste.
For setup steps and settings, see the Commitment Orchestrator documentation and the EC2 setup guide. For the dashboard features, see Harness Commitment Orchestrator: A Modernized FinOps Experience.
What a Team Does in Practice
- See the baseline. Open the inventory view to see existing RIs and Savings Plans, and coverage by cost or by hours.
- Set the rules once. Choose target coverage, preferred commitment type, a cap on new Savings Plan size, and any accounts, Regions, or instances to exclude.
- Start with approval. Review each recommendation within the 24 hour window, and get notified when approval is needed.
- Build coverage gradually. Atomization spreads RI expirations monthly, and SP Staggering builds Savings Plan coverage in smaller purchases where enabled.
- Automate when you are ready. Fully Automated mode is set during setup. Move to it for the commitment types where recommendations match what your team would have chosen.
- Review on a schedule. Check utilization trends and upcoming expirations, and use the Roll-Down Policy to shift your mix toward flexibility over time.
Next Step
Want to see commitment management next to the workloads that drive your spend? Contact Us.
.png)