To size Azure reservation coverage, find the floor of your usage, the capacity that runs continuously and that you can defend for one or three years, and cover that. Do not cover your peak, and do not chase the headline discount by over committing. Azure reservations discount a specific VM family in a region by roughly 20 to over 60 percent against pay as you go in exchange for a one or three year commitment, and the Azure Savings Plan trades some of that discount for flexibility. The buyer's job is to put deep commitments on the certain core and flexible cover on the band above it, so utilization stays high and you never pay for hours you do not use.
This sits in the Azure commitments cluster with Azure reservations explained for buyers and the risk adjusted approach to Azure commitments. Read those for the instruments and the philosophy this article turns into a number.
What coverage percentage should you target?
There is no universal number, and any tool that suggests 100 percent coverage is optimising the wrong thing. The right target is the floor of your usage over the commitment term. Plot your compute usage over the last several months and find the level it never drops below. That floor is what you can commit to with confidence. For most estates it lands somewhere in the region of 60 to 80 percent of current compute, with the rest being genuine variability, growth uncertainty, or workloads that may move.
Covering above the floor means buying reservations for hours that sometimes do not run, and an unused reservation hour is pure waste, the exact downside you accepted in exchange for the discount. The asymmetry matters: under cover and you pay a little more on demand, which is recoverable next month; over cover and you have locked in spend you cannot use for the term. Lean toward the defensible floor.
Reservations or the Azure Savings Plan?
They solve different problems and the best estates use both.
Reservations for the certain core
An Azure reservation commits to a specific VM family in a region and returns the deepest discount. It suits the part of your estate whose shape you are sure of: the database tier that will run the same instances for three years, the steady production fleet. Where you have certainty, take the bigger discount.
Azure Savings Plan for the flexible band
The Azure Savings Plan commits to an hourly spend amount rather than a specific resource, so it flexes across instance types and regions. It returns a smaller discount than a matched reservation but removes the risk of being locked to a family you might move off. Use it for the stable spend level whose composition you expect to change.
Buyer takeaway: layer them. Reservations on the certain core capture the deepest discount, a savings plan on the flexible band above captures the rest without the lock in, and on demand absorbs the genuinely uncertain top.
A worked coverage method
A European SaaS company ran an indicative steady compute base with seasonal peaks roughly 30 percent above the floor. Their previous approach covered close to peak with three year reservations, so utilization sagged whenever load dipped and they were paying for idle commitment. We rebuilt coverage in three bands. The bottom band, the capacity that never went idle, took three year reservations on the specific families for the deepest discount. The middle band, stable spend with a shifting instance mix, took an Azure Savings Plan for flexibility. The top band, the seasonal and uncertain peak, stayed on demand. Reservation utilization rose back toward full, the blended discount improved, and no commitment sat idle. They carried the risk they could defend and no more. Figures are verified against billing data and anonymised.
The method generalises: separate usage into certain, stable, and variable bands, then match the instrument to the certainty of each band. Coverage is a layering decision, not a single percentage.
What if utilization drops anyway?
Forecasts are wrong sometimes, so build in the recovery path. Azure allows exchanging and refunding reservations within limits, which softens a mis sized commitment, and the Azure Savings Plan's flexibility absorbs mix changes without action. The standing discipline is to monitor reservation utilization continuously, treat a sustained drop as a signal to exchange or re plan, and resize at renewal against the current floor rather than rolling over the old number. We cover the ongoing watch in the risk adjusted approach to Azure commitments.
Frequently asked questions
What coverage percentage should Azure reservations target?
Reservations or the Azure Savings Plan?
What happens if Azure reservation utilization drops?
Where coverage sits in the program
Commitment coverage is the biggest single lever on an Azure bill and the biggest risk if sized by discount rather than forecast. We size and manage coverage as an independent buyer side advisory across AWS, Azure, GCP, and OCI, taking zero provider commissions and standing behind a guarantee: we reduce your cloud spend or we reimburse our service fee.
For monthly buyer side commitment analysis, subscribe to The Cloud Spend Navigator.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.