TL
The short answer

Azure reservations come in one year and three year terms, and the three year term discounts the reserved capacity more deeply per hour. The deeper rate looks like the obvious choice, but it is a real saving only if the workload behind it actually runs for three years; if the workload shrinks, changes family, or moves before the term ends, the longer commitment can strand value that a one year term would have released. So the right term is set by how confident your forecast is, not by which discount is larger. The cost efficient pattern is to ladder terms: three years on the demand you are sure of, one year on the demand you expect but cannot guarantee, and pay as you go or a savings plan on the variable remainder.

Here is how the terms differ, how to weigh the trade, and how to ladder coverage so the portfolio stays both deep and flexible.

How do the two terms differ?

Both terms reserve a specific instance type or service capacity and discount it against pay as you go in exchange for a commitment. The one year term asks for a twelve month commitment at a moderate discount; the three year term asks for a thirty six month commitment at a deeper one. Azure reservations are relatively forgiving in that you can exchange a reservation for another of equal or greater value or cancel for a refund within published limits, which softens the risk of a longer term, but exchanges and cancellations have their own rules and are not a free pass. The headline difference is the discount depth; the difference that decides total cost is whether the committed usage survives the term.

When does each term win?

SituationBetter termWhy
Stable, long lived workload on a fixed familyThree yearDeepest rate on usage that will certainly persist
Uncertain forecast beyond twelve monthsOne yearSmaller discount buys the option to reassess sooner
First time building coverageOne yearLearn your true floor before locking three years
Rapidly evolving or migrating estateOne year or savings planFlexibility matters more than the deepest rate
Mature estate with a confident coreThree year on the coreLock the certain base, flex the rest

The pattern is to match term length to forecast confidence per slice of demand, not to apply one term across the whole estate. Discount levels are indicative and depend on service, region, and payment option; verify against current Azure pricing.

How do you ladder terms without overcommitting?

Treat coverage as layers. The bottom layer is the demand you are most confident will run for years, the genuine floor of the estate; cover it with three year reservations for the deepest rate. The middle layer is demand you expect to persist but cannot guarantee, perhaps a year out; cover it with one year reservations so you can reassess at renewal. The top layer is the variable swing; leave it on pay as you go or under a savings plan, whose flexibility suits demand that moves. Then stagger the purchase dates so reservations do not all renew at once, which avoids a single large cliff and keeps you renewing from a steady position rather than under time pressure. The result is a portfolio that captures most of the deep discount while preserving the option to adapt.

A worked example

Worked example

A scaling fintech was about to put three year reservations across its entire steady fleet to maximise the discount. Reviewing the forecast showed that only part of the fleet was genuinely fixed for three years; a sizeable slice sat on an instance family the platform team planned to migrate away from within eighteen months. Committing three years on that slice would have locked a rate against capacity due to disappear. Splitting the purchase, three years on the truly stable core, one year on the slice due to migrate, and a savings plan on the variable layer, captured almost all of the available discount while leaving the migration free to proceed without a stranded commitment. This term discipline was part of the program that left the company 41 percent lighter on cloud spend. Figures are verified against billing data and anonymised.

Frequently asked questions

Is a three year reservation always cheaper than one year?
It carries a deeper discount per hour, so on usage that genuinely persists for three years it costs less overall. But the deeper rate saves only if the workload runs the full term. If it changes or moves first, the longer commitment can strand value a one year term would have released.
When should you choose a one year reservation?
When the forecast beyond twelve months is uncertain, when the workload or family may change, or when building coverage for the first time. The smaller discount buys the option to reassess sooner, which is worth more than the extra points when the future is unclear.
How do you ladder reservation terms?
Cover the demand you are most confident in with three year reservations, layer one year on demand you expect but are less sure of, and leave the variable top on pay as you go or a savings plan. Stagger purchase dates to avoid a single renewal cliff.

Set terms to your forecast, not the headline rate

We build reservation portfolios laddered to a forecast you can defend, capturing the deep discount on the certain core while keeping the rest flexible, as an independent advisory that takes zero provider commissions and answers only to you. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on a Fixed Fee or a no risk Gainshare basis. Download the commitment negotiation playbook, read the deeper Azure cost optimization guide, and compare the instruments in Azure savings plan versus reservations. For monthly buyer side analysis, subscribe to The Cloud Spend Navigator.

Independent · buyer-side

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.

Buyer-side intelligence, monthly.

The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.