Find the cloud bill gotchas in your own AWS, Google Cloud or Azure bill.

CUR 2.0 (Data Exports)

In the Billing and Cost Management console, open Data Exports and create a standard data export (CUR 2.0) with resource IDs included, delivered as CSV with gzip to an S3 bucket. Download the files for the last two or three months and drop them in together. To see Kubernetes pod costs, turn on split cost allocation data first.

Legacy Cost and Usage Report

If you already have a legacy CUR delivered to S3, download the .csv.gz parts for the months you want and drop them in. Columns are matched by name, so either column style works.

FOCUS (Data Exports)

In Data Exports, create a FOCUS export with all columns, as CSV with gzip. AWS's FOCUS export keeps its usage types (x_UsageType, or SkuMeter in FOCUS 1.2), so most checks can run. It has no split cost allocation data.

Cost Explorer CSV (fewer checks)

In Cost Explorer, set the last three months, monthly granularity, group by Usage type, and choose Download as CSV. It has no resources, accounts, tags or commitment line items, so these checks can't run on it: Discounting your waste, Kubernetes requests vs. reality, Logs nobody reads, and Nobody owns the bill. Grouped by Service, only the CI/CD and AI checks can run.

Cost Management export (amortized)

In the Azure portal, search for Cost Management, choose your billing scope, then open Exports and select + Create. On the Datasets tab, select + Add export and choose the type of data Cost and usage details (amortized). Amortized cost is the one that shows unused reservations and savings plans. On the Destination tab, choose CSV and Gzip compression, then Review + create. Download the files for each month from the storage account and drop them in together. Enterprise Agreement, Microsoft Customer Agreement and pay-as-you-go column names all work.

FOCUS export

The same steps with Cost and usage details (FOCUS) as the type of data also work. It combines actual and amortized cost in one file.

Downloaded usage file

Without an export: on a Microsoft Customer Agreement, open the billing scope overview, then Download usage and prices, then Azure usage file, Prepare and Download. On an Enterprise Agreement, open Usage + charges and download the usage details. On pay-as-you-go, open Invoices and download the usage details. The pay-as-you-go file has no charge types, so the commitment check can't run on it.

Billing export to BigQuery, then one query

Google Cloud sends billing data to BigQuery, not to a file. If it isn't on yet: in the console, open Billing, then Billing export, and on the BigQuery export tab enable the Standard usage cost export to a dataset. A multi-region dataset gets data from the start of the previous month; a regional one only from the day you enable it.

Then open BigQuery, start a new SQL query, paste the query below, and put your own table name in the FROM line (gcp_billing_export_v1_ followed by your billing account ID). Run it, then choose Save results and save it as a CSV file on your computer. BigQuery limits a CSV saved to your computer to 10 MB; if yours is bigger, save it to Google Drive and download it from there. The query flattens labels and credits, because BigQuery can't save nested fields to CSV, and groups by month, so the file stays small. It reads only your billing table.

-- Cloud Bill Gotchas Checker: Google Cloud billing, last 3 full months.
-- Put your own table in the FROM line, run it, then Save results as CSV.
WITH line_items AS (
  SELECT
    invoice.month AS invoice_month,
    service.description AS service_description,
    sku.id AS sku_id,
    sku.description AS sku_description,
    project.id AS project_id,
    cost_type,
    currency,
    usage_start_time,
    usage_end_time,
    usage.amount_in_pricing_units AS usage_amount_in_pricing_units,
    usage.pricing_unit AS usage_pricing_unit,
    cost,
    IFNULL((SELECT SUM(c.amount) FROM UNNEST(credits) c WHERE IFNULL(c.type, '') != 'PROMOTION'), 0) AS credits,
    IFNULL((SELECT SUM(c.amount) FROM UNNEST(credits) c WHERE c.type = 'PROMOTION'), 0) AS promotion_credits,
    IFNULL((SELECT SUM(c.amount) FROM UNNEST(credits) c WHERE c.type IN ('COMMITTED_USAGE_DISCOUNT', 'COMMITTED_USAGE_DISCOUNT_DOLLAR_BASE')), 0) AS cud_credits,
    IFNULL((SELECT SUM(c.amount) FROM UNNEST(credits) c WHERE c.type = 'FEE_UTILIZATION_OFFSET'), 0) AS fee_offset_credits,
    IFNULL((SELECT STRING_AGG(CONCAT(l.key, '=', IFNULL(l.value, '')), ';' ORDER BY l.key) FROM UNNEST(labels) l), '') AS labels,
    IFNULL((SELECT STRING_AGG(CONCAT(l.key, '=', IFNULL(l.value, '')), ';' ORDER BY l.key) FROM UNNEST(project.labels) l), '') AS project_labels,
    EXISTS(SELECT 1 FROM UNNEST(system_labels) s WHERE s.key = 'compute.googleapis.com/is_unused_reservation' AND LOWER(IFNULL(s.value, '')) != 'false') AS unused_reservation
  FROM `my-project.my_dataset.gcp_billing_export_v1_XXXXXX_XXXXXX_XXXXXX`
  WHERE invoice.month >= FORMAT_DATE('%Y%m', DATE_SUB(DATE_TRUNC(CURRENT_DATE(), MONTH), INTERVAL 3 MONTH))
    AND invoice.month < FORMAT_DATE('%Y%m', DATE_TRUNC(CURRENT_DATE(), MONTH))
)
SELECT
  invoice_month, service_description, sku_id, sku_description, project_id,
  cost_type, currency, labels, project_labels, unused_reservation,
  MIN(usage_start_time) AS usage_start_time,
  MAX(usage_end_time) AS usage_end_time,
  SUM(usage_amount_in_pricing_units) AS usage_amount_in_pricing_units,
  ANY_VALUE(usage_pricing_unit) AS usage_pricing_unit,
  SUM(cost) AS cost,
  SUM(credits) AS credits,
  SUM(promotion_credits) AS promotion_credits,
  SUM(cud_credits) AS cud_credits,
  SUM(fee_offset_credits) AS fee_offset_credits
FROM line_items
GROUP BY invoice_month, service_description, sku_id, sku_description, project_id,
  cost_type, currency, labels, project_labels, unused_reservation
ORDER BY invoice_month, SUM(cost) DESC

Amounts are cost after credits, in your billing account's currency. Promotional credits are kept in their own column and left out, the way AWS credits are, so the checks show what your usage costs once a promotion runs out. The standard export has no resource names, so the checks show services, SKUs and projects. The column names follow Google's "Structure of Standard data export" page in the Cloud Billing docs.

FOCUS export (preview)

Google's FOCUS export also lands in BigQuery. Its labels and credits are nested, so a CSV of it carries no tags or commitment data, and those two checks say they can't tell. The query above covers more.