Dashboards

You built the right Cost Explorer view once — group by team, filter to the production cluster, Fully Loaded, last 30 days — and now finance wants to see it every Monday morning. Dashboards are how you save that query, share it with the org, and give your teammates a stable URL to land on.

What a Dashboard is

A Dashboard is a Cost Explorer query, given a name. It captures:

• The current group-by chip (Workload, Namespace, etc.).
• The full filter set across the workload, node, and cluster axes.
• The cost mode (Fully Loaded vs Workload Only).
• The projection toggle state.
• The relative time-range preset (Last 7 days, Last 30 days, Month to date).

It deliberately does not capture absolute date ranges. Opening a saved Dashboard always re-evaluates against "now" — that's the whole point. A "weekly per-team spend" Dashboard is useful because next week it shows next week's numbers, not last week's.

The three lists

The Dashboards tab inside Cost Explorer splits saved views into three lists:

• All — every Dashboard you can see (your own plus anything shared with the org).
• Mine — Dashboards you created. The shortlist for the views you maintain.
• Shared with the organization — Dashboards anyone in your org made visible.

The default landing list is All; the others narrow to ownership.

Share with organization

Every Dashboard has a Share with organization toggle in the save dialog and on the Dashboard row. Toggle on, every member of your tenant can see and open the Dashboard. Toggle off, only you see it.

A few rules:

• There is no public share link. A Dashboard URL pasted to someone outside your tenant won't work — they'll get the dashboard's login wall, not the view.
• There is no per-user sharing. Either the whole org sees it, or only you. RBAC restricts what data anyone can see inside the Dashboard (a viewer who can't see cluster prod-east won't see its rows), but the Dashboard's visibility itself is a single org-wide flag.
• Only the creator can edit. A shared Dashboard is read-only to non-owners. To customize, Duplicate to a copy you own, then edit that.

The inaccessible-clusters banner

A Dashboard might reference a cluster the current viewer can't see — either because their RBAC role doesn't include it, or because the cluster was disconnected after the Dashboard was saved.

When this happens, the Dashboard renders with a banner at the top: "Some clusters in this view aren't visible to you. Numbers exclude those clusters." The KPIs and table reflect only the rows the viewer is allowed to read. The Dashboard configuration itself is unchanged — granting access later will start showing the missing rows immediately, no re-save needed.

When to use Dashboards (vs ad-hoc)

The boundary is "will I want to open this again?". If yes, save it; if no, don't clutter the list.

What Dashboards aren't

• Not a scheduling system. Dashboards don't email you on a cadence. For period-anchored, scheduled-delivery output, the Reports archive is the home — and the (forthcoming) Reporting Rules layer will wire scheduling onto it.
• Not the source of truth for a closed period. A Dashboard re-evaluates against the latest data every time, including back-fills, team renames, and label updates. The current month's Dashboard for "spend by team" can drift if you re-attribute workloads. The closed-period Report is the version that won't move.
• Not a chart library. Each Dashboard is one Cost Explorer query — one chart, one table. To build a multi-chart page, save multiple Dashboards and arrange them on your end.

Next steps

• Cost Explorer — build the query you want to save.
• Reports — the time-anchored archive Dashboards aren't.
• Cost monitoring overview — which surface for which question.