Cost monitoring

There can be a lot of questions a FinOps engineer may ask.

Sometimes the question is one-off ("why is the bill up this week?"), sometimes it's the same question every month ("here's the per-team breakdown"), and sometimes it's "send me the formal report".

Kubeadapt splits cost monitoring into three surfaces so each question lands on the right tool.

The three surfaces

Cost Explorer — for the question you have right now

Cost Explorer is the ad-hoc query surface. You pick a time window, choose one of ten group-by dimensions (Workload, Namespace, Controller Kind, Node, Pod, Container, Deployment, StatefulSet, DaemonSet, Instance Type), layer filters across workload, node, and cluster axes, and read the result.

Use Cost Explorer when:

• You have a one-off question and just need the number.
• You're debugging a spike and want to drill from cluster → namespace → workload.
• You want to export the rows for a spreadsheet or share-with-finance moment.

Cost Explorer reads the same underlying cost data as the rest of the product — the chart, the table, and any export you take all reflect the same numbers.

Dashboards — for the view you re-open

Saved Dashboards live inside the Cost Explorer "Dashboards" tab. A Dashboard is a Cost Explorer query — group-by, filters, time-range preset, cost mode — given a name and (optionally) shared with the rest of your organization.

Use Dashboards when:

• You're asking the same question on a schedule (weekly per-team rollup, monthly per-namespace, anomaly hunt).
• You want a teammate to land on the same view you did, without dictating the exact filter chips over Slack.
• You want a stable URL to paste into a runbook.

Dashboards keep three lists — All, Mine, Shared with the organization — so you can find your own work without scrolling past everyone else's.

Reports — for the formal archive

Reports is the time-anchored archive. Daily, weekly, monthly, quarterly, and yearly reports get finalized as the period closes, captured at a snapshot you can come back to a year later, and surfaced as a heatmap by cadence and metric.

Use Reports when:

• You need the close-of-month number for an invoice review.
• You're presenting trend data to leadership and want the version that won't change because someone re-attributed a team.
• You're hunting for the period when a regression first appeared — the heatmap surfaces deltas across every closed period.

Reports also hosts the (forthcoming) Reporting Rules scheduling layer. Today the button is a placeholder; see the Reports page for status.

How they fit together

A typical month uses all three:

1. Finance pings you about last Tuesday's spike. Open Cost Explorer, group by namespace, scope to that day. Find it in two minutes.
2. The investigation produces a useful view — say, daily cost by team for the production cluster. Save it as a Dashboard, share with the org.
3. End of month, the closed monthly Report is the version everyone agrees on. The Dashboard might shift if a team gets renamed; the Report doesn't.

What ties the three together

• Same cost data. Cost Explorer, Dashboards, and Reports all read from the same per-day cost numbers — broken down by cluster, namespace, workload, pod, node, and container. See How costs are computed.
• Same cost modes. Fully Loaded (the default — capped compute + idle + overhead) and Workload Only (compute only). The mode toggle is in Cost Explorer; Dashboards inherit the saved mode; Reports default to Fully Loaded.
• Same UTC boundaries. All windows are inclusive-start, exclusive-end, in UTC. The same week that closes in your Cost Explorer view is the same week the Report captures.

Next steps

• Cost Explorer — the deep-dive on group-by, filters, and cost modes.
• Dashboards — save and share views.
• Reports — the time-anchored archive.