Dashboard
Executive spend, top offenders, and operational efficiency for the connected BigQuery projects.
The dashboard is the home screen after a scan. It answers three questions: what you spent, who and what drove it, and whether the last hours look healthy.
Executive summary
One screen that answers "how are we doing?" for the selected range.
Range. Pick 24H, 7D, 30D (default), MTD (month to date) or 90D at the top right. Every section of the dashboard follows it, including Top offenders. Month-to-date spend, the forecast and the commit always use the calendar month. Changes compare the range with the same number of days just before it. If there is not enough history for a full earlier period, the comparison is hidden.
Health badge. Next to the title: Healthy, Needs attention or Critical. Hover the badge to see why. It turns critical when the end-of-month forecast is more than 20% over your commit (or last month, if you have no commit), or when an anomaly alert is still open. It needs attention when daily burn rose more than 10%, the failure rate rose 2 points or more, or an anomaly alert in the range is now back to normal.
Top row.
- Total potential savings — from open findings, per month. Overlapping rules (partition and cluster the same table) never sum past the spend in that bucket.
- Month-to-date spend and the end-of-month forecast.
- Commit used — how much of your spend commit is used and where you will land by the end of the period. Owners set the commit in Settings → Billing → Spend commit (per month, or per year on a plan billed yearly). Without a commit, the tile shows how this month is pacing against last month. When you pick a cost center, the tile becomes Budget used for that cost center only: its monthly budget, or Set a budget when it has none. It never compares one cost center's spend with the account commit.
- Pricing model — on-demand, capacity or both, with the compute spend for each.
KPI tiles. Burn rate (average spend per day), data size (logical bytes and growth), incidents (open findings and anomaly alerts in the range — opens Findings in a new tab), active users, active tables, and failure rate.
Chart tabs.
- Spend mix — daily spend stacked by cost category, with a dashed 30-day moving average. Categories with no spend fold into a "+N empty" chip in the legend.
- Health — small charts for spend, data scanned, storage and failed jobs.
- By cost center — daily spend per cost center. Shown once you define cost centers.
- SKU — your billed BigQuery SKUs from the billing export. Shown once a billing export is connected.
Spend groups. Beside the chart: query compute, storage, streaming & components, and data movement, each with its range total, change, and trend line.
Cost center filter. When cost centers exist, a picker next to the range limits the whole dashboard to one cost center (or Unallocated). Under a cost center the daily spend chart shows compute only, because storage has no per-job owner. Storage for that cost center shows as its current size.
Storage appears as a daily cost: the monthly storage price divided by the days in the month.
Dates use your account timezone, not the browser timezone.
Top offenders
Queries, users, datasets, and tables ranked by cost, for the range at the top of the page. A second block ranks savings by finding type so you know which rule class is worth a sprint.
Hover a top query and click Optimize to get an AI rewrite in Cognis (Starter and up). See AI query optimizer.
Click through to Usage or a finding when a row is actionable.
Operational efficiency
Trends for cost per query, waste, slots, failures and storage growth. Ranges up to 7 days show hours. Longer ranges show days, in your account timezone. These charts read pre-aggregated metrics, not a raw dump of every job.
Anomalies
Every scan checks each project and region for unusual rises in spend, bytes, slots, failures and storage, and for missed or late scheduled jobs. The feed shows one row per anomaly, with a Why? link to the cause. A red bell marks the same anomaly on the charts.
Full details: Anomalies.
Email report
The Email report button next to the range buttons emails you an Excel file of what you see, with your range, project, region and cost center, and only the report parts you get. Nothing downloads in the browser. The file has one tab per dashboard section, plus Findings.
It works on every plan. See Excel report and Report emails.
If numbers look low
- The scan window may be shorter than a full month on a brand-new connection.
- History depends on your plan: 30 days on the trial, Starter and Growth, 60 on Pro, 90 on Enterprise. A 90D range on a 30-day plan shows only 30 days of jobs.
- Excluded projects do not contribute.
- On-demand vs capacity math depends on reservation metadata. Grant reservation IAM if that block is empty.