Services Work Pricing Blog About Book a free call →

S3 prefix cost allocation: what changes for SaaS chargeback

AWS can now split a shared S3 bucket bill by prefix in the Cost and Usage Report, announced November 17, 2026. What changed and how to try it this week.

On November 17, 2026, AWS announced split cost allocation data for Amazon S3. It allocates the storage and request costs of one shared general purpose bucket to the prefixes inside it, in the Cost and Usage Report (CUR). If your SaaS keeps every tenant, team or product under tenant-id/ in one bucket, you can now see an estimated bill per prefix without building your own pipeline. If you already use one bucket per tenant, or your S3 bill is small, you can mostly ignore it.

What was announced, and what changed?

Before this, AWS says teams had to build custom pipelines on S3 Inventory or Storage Lens reports to split a shared bucket's cost. Now the CUR carries the split itself:

  • Costs go to the deepest matching prefix within the prefix levels you choose.
  • Each line item carries its full prefix path as AWS-generated cost allocation tags.
  • AWS lists chargeback, showback and finding which prefixes drive S3 spend as the use cases.

The S3 documentation adds that the figures are estimates, and that S3 Storage Lens does not calculate costs: it publishes prefix-level metrics, and AWS Billing and Cost Management computes the estimated cost from them.

What exactly is allocated?

As of November 2026, the CUR documentation says the feature allocates storage charges for all storage classes and Tier 1 and Tier 2 request charges. The S3 documentation lists what is not included:

  • Storage and request costs for annotations.
  • Request costs for cross-Region replication.

Costs that cannot be attributed, and root-level objects (those not under any prefix), are not split by prefix. They land in an "unattributed" line at bucket level.

For Requester Pays buckets, request estimates are less accurate when many requests come from other AWS accounts.

Who should care, and who can ignore it?

Care if:

  • Several tenants or products share a bucket and finance asks what each one costs.
  • You want showback for data-heavy features, such as exports, uploads or RAG document stores, that live under their own prefix.
  • You currently estimate per-tenant storage with a script nobody wants to maintain.

Ignore if:

  • You use one bucket per tenant. You likely do not need this feature.
  • Your keys are not organized by a stable prefix. The feature can only split what your key design already separates.
  • You use directory, table or vector buckets: the documentation says only general purpose buckets are supported.

What does it cost, and what are the limits?

AWS says the feature is available in all commercial Regions where S3 Storage Lens advanced metrics are available, with no additional charge beyond S3 Storage Lens advanced tier pricing. The advanced tier itself is billed, so check the Amazon S3 pricing page before you turn it on for many buckets. The documented limits, as of November 2026:

  • The management account or a standalone account must enable the feature in Billing and Cost Management.
  • The Storage Lens prefix delimiter must be /.
  • You choose a range of prefix levels, from 1 to 15, and report at most 5 adjacent levels. The default is levels 1 to 3.
  • S3 Storage Lens supports up to 10,000 prefixes per bucket at the configured depth. If a level goes over, it falls back to a shallower one. If your minimum level is over the limit, that bucket gets no prefix cost data.
  • The new tags (aws:s3:PrefixLevel1 to aws:s3:PrefixLevel15, and aws:s3:BucketName) are inactive by default. Activate them on the Cost allocation tags page. AWS says it can take up to 24 hours for them to appear.

The report also grows. AWS gives this formula: (prefixes with usage + 1) × usage types (up to 2) × days in the month.

If you load the CUR into a query engine, plan for that volume before you enable many buckets at once.

How can a team try it safely this week?

  1. Pick one shared bucket with a clear prefix scheme, ideally with tenant or team at level 1.
  2. In S3 Storage Lens, upgrade or create an advanced tier dashboard that covers that bucket only, and enable Billing and Cost Management publishing with the prefix levels you need.
  3. From the management account, enable split cost allocation data for Amazon S3 in Billing and Cost Management, and make sure your CUR export includes split cost allocation data and resource IDs.
  4. Activate the aws:s3: cost allocation tags.
  5. After the next daily Storage Lens publication, AWS says prefix data starts to appear. Query a few days of it.

Keep the first trial to one bucket and one export so you can measure the extra line items and the Storage Lens charge.

What would we check first before adopting it?

We would start with these questions:

  • Key design. Is the tenant or team always the first prefix segment? Mixed naming makes the split meaningless.
  • Share of unattributed cost. If root-level objects or excluded costs are a large part of the bill, the per-prefix view will understate reality.
  • Cost versus value. Compare the Storage Lens advanced fee for your buckets with what the S3 line is worth. For a small bill, a monthly estimate may be enough.
  • Estimate, not invoice. Treat the numbers as showback. Do not invoice customers from them without checking how the estimate is calculated.
  • Who reads it. Decide who reviews the report and how often, or it becomes another unread dashboard.

If you need help with FinOps design across accounts, see our services.

This article offers general information based on public AWS documentation as of November 2026. It is not financial advice: check current prices and terms, and confirm with a qualified professional before making cost decisions.