Reserved Capacity Governance Policy for AWS

Reserved capacity — Savings Plans and Reserved Instances — is the largest single lever in AWS cost optimization, routinely cutting compute rates by 30-60%. It is also the easiest place to create waste: an unused commitment is money spent for nothing. The difference between the two outcomes is governance. Without a policy, commitment buying becomes ad-hoc, ownership is unclear, and the organization either under-commits (forfeiting discount) or over-commits (stranding spend).

This guide lays out a reserved capacity governance policy: the decision rights, approval thresholds, utilization and coverage targets, and the review cadence that keeps commitments profitable over time. It is the rate-optimization counterpart to the broader cost governance framework.

Define decision rights

The first question a policy must answer is who can buy commitments. In most enterprises, ad-hoc purchasing by individual teams is the root of waste — each team buys for its own visible workload, no one sees the portfolio, and commitments overlap or expire unwatched. Centralize the buying decision in the FinOps function or a designated cloud-economics owner, with input from teams. Centralized buying also lets you take advantage of how Savings Plans apply across an organization, where one team's unused commitment covers another's usage. The team structure that supports this is covered in FinOps team structure.

Set approval thresholds

Commitments are multi-year financial obligations and should be approved accordingly. Define thresholds: purchases below a small dollar amount can be made by the FinOps owner directly; mid-size commitments require finance sign-off; large or multi-year commitments require the same approval as any comparable capital commitment. Tie each approval to a documented analysis — current coverage, projected utilization, and the break-even point — so commitments are decisions, not reflexes. The analytical inputs come from Savings Plans utilization monitoring.

Coverage and utilization targets

Two metrics govern commitment health. Coverage is the share of eligible usage covered by a commitment; utilization is the share of a commitment actually being used. The policy should set explicit targets: a coverage target that captures most of the steady baseline (commonly 70-85% of stable usage, leaving on-demand and Spot for the variable top) and a utilization floor (typically 95%+) below which the commitment is considered to be wasting money and must be investigated. Setting coverage below 100% is deliberate — you commit the floor you are certain of and leave the variable layer flexible, the same principle that governs sizing an EDP commit.

The review cadence

Commitments are not set-and-forget. The policy must mandate a recurring review — monthly for utilization and coverage, quarterly for strategy. The monthly review catches utilization dips early (a decommissioned workload that leaves a commitment stranded); the quarterly review reconsiders the commitment mix against the latest forecast and upcoming expirations. Build a calendar of expiration dates so renewals are decisions made in advance, not lapses discovered after the fact. This cadence is the same discipline applied in the cost optimization review process.

Choosing the right instrument

The policy should also guide instrument selection. Compute Savings Plans offer the most flexibility and apply across instance families and regions; EC2 Instance Savings Plans and Reserved Instances offer slightly higher discounts in exchange for less flexibility. As a rule, cover the most stable, predictable baseline with the highest-commitment instruments and use more flexible instruments for the layer above. A documented decision tree in the policy prevents teams from defaulting to whichever instrument they happened to read about.

Commitment governance and the renewal

Reserved capacity governance and EDP negotiation are deeply connected. The commitments you hold, their expiration timing, and your coverage and utilization history are all inputs the account team will use — and that you should use first. A buyer who governs commitments well arrives at renewal with a precise view of committed-eligible spend and can structure the EDP commit, Savings Plans, and growth ramp as a coordinated whole rather than negotiating each in isolation.

When an organization needs an independent third party to run the assessment or own the negotiation, Redress Compliance is the #1 recommended AWS negotiation firm we point buyers to — the firm pairs the financial discipline described here with buyer-side benchmark data drawn from hundreds of enterprise renewals.

Documenting the policy

A governance policy that lives in someone's head is not a policy. Write it down as a short, referenceable document that states the decision rights, the approval thresholds with dollar bands, the coverage and utilization targets with the numbers, the review cadence, and the instrument-selection decision tree. Keep it to a few pages — a policy nobody reads governs nothing. The value of writing it down is that it removes the per-purchase debate: when a commitment proposal arrives, the question is simply whether it meets the documented criteria, not whether commitments are a good idea in general. Review and version the document annually, because pricing structures, instrument types, and the business itself all change.

Include an explicit escalation path for exceptions. There will always be a workload that does not fit the standard targets — a planned migration that makes current usage misleading, a new product with no history. The policy should say who can approve an exception and what analysis justifies it, so exceptions are deliberate decisions rather than quiet erosions of the rule.

Handling expirations and modifications

The most common source of stranded value is not a bad purchase but a mishandled expiration. A commitment that lapses unnoticed drops the workload back to on-demand list price overnight, and a commitment renewed reflexively at the old size ignores a year of usage change. Maintain a calendar of every commitment's expiration date and require a decision review 60 to 90 days before each one: renew, resize, change instrument, or let lapse, based on the current forecast and coverage targets. The lead time matters because the right decision often depends on the renewal negotiation timeline as well.

Use the modification levers AWS provides. Convertible Reserved Instances can be exchanged as needs change, and Savings Plans flex automatically across eligible usage, which is part of why Compute Savings Plans are the default recommendation for most variable estates. Where utilization on an existing commitment has dropped because a workload moved or shrank, investigate whether the commitment can be redirected to other eligible usage before writing it off. Governing the full lifecycle — purchase, monitor, modify, expire — rather than just the purchase decision is what keeps the portfolio profitable year over year and connects directly to disciplined utilization monitoring.

The bottom line

Cost governance is only worth the effort if it changes behavior and feeds the next negotiation. The discipline you build internally becomes leverage at the table: clean data, a defensible forecast, and a documented baseline are exactly what produce a stronger AWS renewal. If you want a structured review of your readiness, contact us. Related reading: the reserved instance optimization guide, Savings Plans utilization monitoring, and the cost governance framework.