The S3 storage classes are tiers that trade lower storage price for higher access cost and, in the archive tiers, retrieval delay. Picking the cheapest tier that works means matching the class to how often you actually read each object: S3 Standard for frequently accessed data, the Infrequent Access classes for data you read occasionally, Intelligent-Tiering when the pattern is unknown or changing, and the Glacier classes for archives you rarely or never touch. The cheapest tier is not always the lowest storage price, because retrieval fees and per-object minimums can erase the saving if you read more than you expected.
This explainer sits under our complete guide to AWS cost optimization, the pillar for this cluster, and supports the Cut step of our See, Cut, Lock, Run method. Storage and access cost are joined at the hip, so read it alongside AWS data transfer and egress charges, where the cost of moving data out is often the larger bill.
Every tier below Standard charges more to retrieve data and most impose a minimum storage duration and a minimum billable object size. For cold data those costs are trivial. For data you read more than expected, they can cost more than staying in Standard.
The tiers, from hot to cold
S3 Standard is the default: highest storage price, no retrieval fee, no minimum duration, built for data accessed frequently. S3 Standard-Infrequent Access and S3 One Zone-Infrequent Access lower the storage price for data read occasionally, in exchange for a per-GB retrieval charge and a minimum storage duration; One Zone trades cross-AZ redundancy for a further discount, suitable only for reproducible data. S3 Intelligent-Tiering moves objects between access tiers automatically based on usage, for a small per-object monitoring fee, and is the right default when you do not know the access pattern. The Glacier classes are for archives: Glacier Instant Retrieval for rarely accessed data that still needs millisecond access, Glacier Flexible Retrieval and Glacier Deep Archive for true cold storage where minutes to hours of retrieval time is acceptable in exchange for the lowest storage prices AWS offers.
How to pick the cheapest tier that works
Start from the access pattern, not the price sheet. If an object is read regularly, Standard is cheapest once you account for retrieval. If it is read a few times a month, an Infrequent Access tier wins. If it is read once or twice a year, a Glacier tier wins by a wide margin. When the pattern is unknown, mixed across a bucket, or likely to change over the object's life, Intelligent-Tiering removes the guesswork by tiering automatically, which is usually the safe default for large, heterogeneous buckets. The transition between tiers over an object's life is best automated rather than done by hand, which is the job of S3 lifecycle policies.
Want your storage footprint tiered correctly?
Our AWS cost audit profiles access patterns across your buckets, models the cheapest tier per prefix including retrieval, and sets the lifecycle rules to get there. On the performance model you pay only from realized savings. No savings, no fee.
Book an AWS cost audit →The traps that erase the saving
Three mistakes turn a storage discount into a loss. The first is tiering hot data: moving frequently read objects to an Infrequent Access or Glacier class so the retrieval fees outweigh the storage saving. The second is the small-object penalty: the IA and Glacier classes bill a minimum object size and a minimum duration, so a bucket of millions of tiny files can cost more in a discounted tier than in Standard. The third is forgetting retrieval time: Glacier Flexible Retrieval and Deep Archive are not instant, so they are wrong for anything an application or user might need on demand. Profile object size and access frequency before you move anything.
| Access pattern | Best class | Watch out for |
|---|---|---|
| Read daily | S3 Standard | Nothing, this is the baseline |
| Read a few times a month | Standard-IA | Retrieval fee, minimum duration |
| Unknown or changing | Intelligent-Tiering | Per-object monitoring fee |
| Archive, rare reads | Glacier Instant or Deep Archive | Retrieval time and cost |
S3 class names, minimums and retrieval behavior reflect AWS offerings as of May 2026. Verify current per-GB prices, minimum object sizes and retrieval options in the S3 pricing pages before setting lifecycle rules, as AWS adjusts these over time.
The AWS Cost Optimization Field Guide includes the bucket-profiling queries and the tier-decision model we use to place objects in the cheapest workable class. It is the downloadable companion to this article.
The short version
S3 storage classes let you pay far less for data you rarely read, but the cheapest tier is the one that fits the access pattern after retrieval fees and minimums, not the one with the lowest sticker storage price. Use Standard for hot data, Infrequent Access for occasional reads, Intelligent-Tiering when the pattern is unclear, and Glacier for archives, and automate the transitions with lifecycle policies. Getting this right across large buckets is a standard part of an AWS cost optimization engagement, as in our SaaS on AWS case study.