Eliminating waste before a commitment is the cleanest dollar you will ever save, because waste removed from the baseline is waste you never agree to buy. A discount applied to idle infrastructure is still a bill for idle infrastructure. Before you sign an AWS EDP, an Azure MACC, or a GCP committed use deal, the job is to shrink the real number first. This sits inside our broader FinOps optimization for cloud commitments practice and feeds directly into commitment structuring and sizing advisory.
Why eliminating waste before a commitment changes the whole deal
A committed use deal converts your spend into a contractual floor. Whatever waste is in your bill on the day you sign becomes waste you are obligated to keep paying for, at a discounted rate that feels like a win but is not. Eliminating waste before a commitment lowers that floor permanently. It is the difference between negotiating a discount on what you need and negotiating a discount on what you forgot to turn off.
As of June 2026, none of the major programs refund or roll over unused commitment. Azure MACC unused commitment is generally lost. AWS EDP overcommitment creates a shortfall you must pay. So the waste you carry in is waste you fund whether or not anyone uses it.
Where the waste hides
Idle and underused compute is the largest category in most enterprises. Instances sized for a peak that never comes, dev and test environments left running overnight and on weekends, and clusters provisioned for headroom nobody measured. Storage is next. Unattached volumes, orphaned snapshots, old backups past their retention need, and hot tier data that has gone cold months ago.
Then there is the architectural waste. Duplicate tooling, redundant data transfer, chatty cross region traffic, and services left enabled after a project ended. None of this belongs in a baseline you are about to commit to for years.
Sequence the cleanup so it lands in the baseline
Order matters. Quick reversible wins first, like scheduling non production environments to stop outside working hours and deleting clearly orphaned resources. Then the structural work, like downsizing and consolidating, which needs validation. Finish before the deal closes. If the cleanup lands after signature, the savings free committed capacity you have already bought, which is the worst outcome.
Rebaseline after each wave so your forecast tracks the cleaned figure. The number you carry to negotiation should be the post cleanup demand, supported by utilisation evidence.
Keep the waste out for good
Eliminating waste once is a project. Keeping it out is governance. Put guardrails in place so the cleaned baseline does not refill before signature and does not drift after it. Tagging, budget alerts, idle resource policies, and an owner for every environment. We go deeper on this in our work on FinOps governance and on continuous optimization, because a commitment sized on a clean baseline only stays efficient if the discipline holds.
Quantify the waste in dollars, not percentages
Percentages persuade nobody in a negotiation. Dollars do. Translate every waste category into an annual figure so the value of removing it is undeniable. Idle compute at a given run rate, unattached storage at its monthly cost, oversized instances at the delta between current and right sized.
The total is the number you are about to avoid committing to. Framed that way, eliminating waste before a commitment stops being a cleanup chore and becomes the highest return work in the whole deal, because every dollar removed is a dollar you never pay at a contracted rate.
The waste categories vendors never flag
Providers will happily recommend a reservation. They are far less eager to point out that you do not need the capacity at all. Cross region data transfer, redundant logging and monitoring pipelines, premium support tiers on workloads that do not need them, and licensing bundled into the bill all hide in plain sight.
These categories rarely appear in vendor optimization tooling because removing them reduces the vendor's revenue. An independent buyer side review looks at them precisely because nobody on the sell side will.
Reversible waste versus structural waste
Sort the waste by how hard it is to remove. Reversible waste, like environments left running overnight, can be cut immediately with no risk and clawed back if needed. Structural waste, like an oversized architecture or a redundant service, takes engineering effort and validation.
Attack reversible waste first to bank quick savings into the baseline, then schedule the structural work so it lands before the deal closes. The goal is that the committed number reflects the post cleanup estate, not the one you started with.
Carry the cleaned number into negotiation
Cleanup that finishes after signature is cleanup that helps the provider, not you, because it frees capacity you already bought. So the timeline matters as much as the work. Finish the elimination, rebaseline, and only then open the commitment conversation.
Bring the documented savings to the table. When a seller argues your number is too low, the list of waste you removed and the utilisation behind the rest is the evidence that holds the line.