AWS Cost and Usage Report: Setup, CUR 2.0 and Athena Queries
The AWS Cost and Usage Report is the most detailed billing dataset AWS offers. Learn CUR 2.0 setup, Athena queries and the pitfalls to avoid.
AWS CUR Athena FinOps

Quick answer: The AWS Cost and Usage Report (CUR) is the most detailed billing dataset AWS provides, with line items down to the resource and hour. It is now created through AWS Data Exports as CUR 2.0, which has a fixed schema and nested columns for tags, cost categories, product details and discounts. You deliver it to Amazon S3 and query it with Athena.
Cost Explorer is great for a quick look. But the moment someone asks which team's pods drove last month's S3 bill, or what one customer actually costs you, the answer is in the Cost and Usage Report. It's the raw ledger behind almost every serious FinOps process.
This guide covers what CUR contains, how CUR 2.0 differs from the legacy report, how to set it up, and the queries and pitfalls worth knowing on day one.
What is the AWS Cost and Usage Report?
The Cost and Usage Report is a file-based export of every line item on your AWS bill. Each row records a service, usage type, account, time period and cost, and, if you ask for it, the individual resource ID. It's the most granular billing dataset AWS offers, available at hourly granularity.
That depth is the point. Cost Explorer summarises. CUR lets you build your own views: chargeback by team, cost per customer, commitment coverage by account, or the data behind a unit economics model.
CUR vs CUR 2.0 vs FOCUS
AWS now offers three formats through Data Exports. They are not the same thing.
| Format | Best for | Key traits |
|---|---|---|
| Legacy CUR | Existing pipelines | Columns vary month to month; tags and categories are separate columns |
| CUR 2.0 | Most new AWS-only work | Fixed schema; four nested columns; extra columns such as account names |
| FOCUS 1.2 | Multi-cloud reporting | Open, vendor-neutral schema |
CUR 2.0 was launched at November 2023 together with AWS Data Exports. Its main improvements are a consistent schema and nested data. Four columns, product, discount, resource tags and cost categories, are collapsed into key-value columns, which avoids the hundreds of sparse columns that legacy CUR could produce. It also adds bill_payer_account_name and line_item_usage_account_name.
Legacy CUR remains available, but AWS recommends moving to CUR 2.0. Two caveats are worth knowing. CUR 2.0 doesn't currently support pro forma data for AWS Billing Conductor billing groups. And split cost allocation data for EKS and ECS is available in CUR 2.0 but not in FOCUS 1.2, so container-heavy teams should stay on CUR 2.0 unless they need multi-cloud standardisation. Our post on split cost allocation for EKS explains that data in detail.
How to set up CUR 2.0 in six steps
- Sign in to the management account of your AWS Organization, so the report covers every linked account.
- Open Billing and Cost Management, then Data Exports, and choose to create a standard data export.
- Select the CUR 2.0 table.
- Choose your options: include resource IDs, include split cost allocation data if you run EKS or ECS, and pick hourly, daily or monthly time granularity. Hourly with resource IDs is the most useful and the largest.
- Pick an S3 bucket and prefix, and Parquet as the format, since it's smaller and faster to query.
- Create the export and wait. The first delivery can take up to a day.
You can run up to five CUR 2.0 exports at once, which is handy for separate finance and engineering datasets. If you have existing legacy queries, Data Exports lets you write a SQL statement that unpacks the nested columns so the output matches your old schema. That's the safest migration path.
Querying CUR with Athena
Point a Glue crawler at the S3 prefix, or create the table from the DDL in the AWS documentation, then query with standard SQL. Two starting points follow. Column names are from the CUR 2.0 table, so check them against your own export before relying on the results.
Spend by service for one month:
SELECT line_item_product_code, SUM(line_item_unblended_cost) AS cost FROM cur2 WHERE line_item_usage_start_date >= TIMESTAMP '2026-10-01 00:00:00' AND line_item_usage_start_date < TIMESTAMP '2026-11-01 00:00:00' GROUP BY 1 ORDER BY cost DESC LIMIT 20;
Spend by team tag:
SELECT resource_tags['user_team'] AS team, SUM(line_item_unblended_cost) AS cost FROM cur2 WHERE line_item_usage_start_date >= TIMESTAMP '2026-10-01 00:00:00' AND line_item_usage_start_date < TIMESTAMP '2026-11-01 00:00:00' GROUP BY 1 ORDER BY cost DESC;
The second query shows why tagging discipline matters: every untagged resource lands in an empty team bucket. Our guides on AWS cost allocation and resource tags cover how to shrink that bucket.
Columns worth learning first
- line_item_usage_start_date: the time bucket for the usage.
- line_item_usage_account_id: the account that used the service.
- line_item_product_code and line_item_usage_type: what was used.
- line_item_resource_id: the specific resource, if you enabled it.
- line_item_line_item_type: usage, tax, credit, refund, fee and commitment lines.
- line_item_unblended_cost: the cost as charged on the line.
- pricing_public_on_demand_cost: what the usage would have cost at On-Demand rates.
- resource_tags: the nested tag map.
Five CUR pitfalls
1. Mixing cost types
Unblended, amortized and net cost answer different questions. Finance usually wants amortized, so upfront commitment fees are spread over the term, while engineers often look at unblended. Agree on one definition per report and label it.
2. Double counting commitment lines
Savings Plans and Reserved Instances produce several kinds of line item, including covered usage, recurring fees and negation lines. Summing every row without filtering by line item type gives wrong totals. Use the effective cost columns for commitments instead.
3. Forgetting to activate cost allocation tags
A tag only appears in billing data after you activate it as a cost allocation tag. Activate early so your history has no gaps.
4. Expecting an exact invoice match
CUR includes credits, refunds and tax lines, and your invoice may group them differently. Reconcile by line item type, not by hunting for one number.
5. Querying without partitions
Athena charges by data scanned. Partition by billing period and select only the columns you need, and a query that scanned gigabytes will scan megabytes.
What to build once CUR is flowing
Once the data lands, the useful outputs are fairly predictable:
- A cost-by-owner dashboard, based on tags and account groupings.
- A commitment view comparing On-Demand-equivalent cost with what you paid.
- Resource-level hunts for idle assets, as in our walkthrough on finding idle VPC endpoints with CUR queries.
- Unit cost metrics, such as cost per customer or per transaction.
- Anomaly checks at the resource level, which complements the managed tooling described in our anomaly detection guide.
Frequently asked questions
What is the difference between CUR and Cost Explorer?
Cost Explorer is an interactive console tool for quick analysis. CUR is a detailed export you query yourself, down to individual resources and hours, which makes it the right base for custom allocation and unit economics.
Is the legacy CUR being retired?
Legacy CUR is still available, and you can keep both formats side by side. AWS does recommend CUR 2.0 for its fixed schema, nested columns and extra account name fields.
Does CUR cost money?
Creating the report has no direct charge. You pay for the S3 storage it uses and for any Athena queries, so Parquet and partitioning keep that small.
How often is CUR updated?
AWS delivers the export on a recurring schedule, so the files refresh through the month. You choose hourly, daily or monthly line-item granularity when you create it.
Final thoughts
Set up CUR 2.0 once, with resource IDs on, and every later FinOps question gets cheaper to answer. Start with the two queries above, fix tagging gaps they reveal, and build outward from there. The real value shows up when billing data meets business data.
If you'd rather skip the pipeline work, book a demo with TruCost.Cloud and we'll show you how allocation and unit economics look on your own AWS data.
Related reading

AWS Cost Per Transaction: How to Measure Cloud Unit Economics and Prove FinOps Value
AWS cost per transaction links cloud spend to business activity. Learn how to calculate it, which data you need, and how unit economics proves FinOps value.

Agentic AI Cost Governance: How to Control the Hidden Costs of Autonomous AI Agents
Autonomous AI agents make many model, tool and API calls per task. Learn how to assign ownership, set limits, track cost per workflow and keep agent spend accou…

AI Cloud Cost Allocation: How to Track and Attribute AI Infrastructure Spending
AI spend hides inside one cloud bill. Learn how to allocate model, GPU and shared infrastructure costs to the teams and applications responsible for them.
See your own bill allocated to 100%
Thirty minutes with a FinOps engineer, on your AWS data if you like. No slideware.