## TL;DR
Point the query at the real table name, which ends with your billing account ID, for example gcp_billing_export_v1_[BILLING_ACCOUNT_ID]. List the dataset's tables with bq ls to see the exact name, then fix the query. Export table names are fixed by Google and always carry that suffix, so a hand-written name without it never matches.

## The error
```text
table not found: project.dataset.gcp_billing_export
```

## Fix it
1. Confirm the export is actually enabled: open Billing, then Billing export, and check the dataset name and location shown there.
   Expected: either a dataset is listed (export on) or the page says export is not enabled.
2. If it is enabled, list what is really in the dataset: `bq ls [PROJECT_ID]:[DATASET]`
   Expected: tables like gcp_billing_export_v1_[BILLING_ACCOUNT_ID] or gcp_billing_export_resource_v1_[BILLING_ACCOUNT_ID].
3. Rewrite the query to use the exact table name from step 2, including the billing account ID suffix.
   Expected: the query parses and the table resolves.
4. Re-run the query.
   Expected: rows return; if the table exists but is empty, the export was only just enabled and there is no backfill.

## Use this when
- A scheduled billing-export job starts failing with a table-not-found error
- Someone hand-wrote the export table name instead of copying it from the dataset
- The export dataset was recreated or moved and old queries still point at the old path

## Not for this skill when
- Billing export was never enabled (enable it first; this skill fixes the query, not the missing export)
- The error is a quota error on the query job (that is a quota problem, different fix)
- The query fails on permissions (the caller needs BigQuery Data Viewer on the dataset)

## Variant phrasings
- BigQuery billing export table does not exist
- Not found table gcp_billing_export_v1
- billing export dataset is empty after enabling
- where is the GCP billing export table name

## Why it happens
Google names export tables itself and appends the billing account ID, so the name is never just gcp_billing_export. Queries copied between accounts break because the suffix changes per account. Separately, the export only starts writing from the moment it is enabled, so a fresh export legitimately has no historical tables.

## Edge cases
- Standard usage export, resource-based export, and price export each write to differently named tables; make sure the query targets the right one
- The dataset location matters: a query running in the wrong region cannot see a regional dataset
- Export data lands with a delay (often several hours), so very recent days can look missing
- Deleting the dataset does not disable the export; re-enable it cleanly or exports fail silently

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_qmrL2eevYH5EXtY2O9eYwA