Usage
Break BigQuery spend down by user, table, query, and failures, with shared date and dimension filters.
Usage has four tabs. Filters (date range and dimensions such as user, label, or job source) sit on each tab and apply in that page’s scope.
| Tab | URL | Question it answers |
|---|---|---|
| Table | /usage/table | Which tables are scanned or stored most expensively |
| Query | /usage/query | Which query fingerprints repeat and cost the most |
| User | /usage/user | Which emails and service accounts spend the most |
| Failures | /usage/failures | Failed jobs and the waste of retries |
How to read a row
Cost is estimated from BigQuery job statistics (bytes billed, slot-ms) using the list prices of the job's region (from Google's BigQuery price page) and, when present, reservation context. It is an analytics estimate, not your invoice line. Compare with Billing when you have Cloud Billing connected.
Query tab
Open a fingerprint to see SQL (formatted), users who ran it, and linked findings. Copy the query or job id when you take the item to the warehouse team.
Hover a query on the top-queries chart, or a job in the job list, and click Optimize to get an AI rewrite in Cognis (Starter and up). See AI query optimizer.
To see Looker cost, filter Job source = Looker.
Failures
Failed jobs still consume slot time and sometimes bytes. High failure rates often sit next to slot contention or bad scheduled queries.
Filters
Use the filter bar to slice by user, dataset, label, priority, workload source (dbt, Looker, scheduled query, and similar classifications from job ids and SQL comments), or cost center (shown once you define cost centers; includes Unallocated). Clear filters if a chart looks empty after a narrow slice.
On Starter and up, the filter panel also has a Cognis box: describe the jobs you want in plain words, for example "service accounts running dbt in prod", and Cognis picks the filters. You can edit them after. See Cognis chat.
Save a filter set as a segment to reuse it later.
History
Usage goes back as far as your plan's history: 30 days on the trial, Starter and Growth, 60 on Pro, 90 on Enterprise. Times use your account timezone.
Anomalies
How SlashPeak learns the normal pattern of each metric, flags unusual rises, finds the cause, and sends alerts. On every plan.
Cost centers
Group BigQuery spend by department, product or business unit with ordered match rules, give each group a monthly budget, then filter the dashboard and Usage by them.