Dashboard
The CostPilot dashboard gives you an at-a-glance view of your Kubernetes cost health. This guide explains every metric and how to interpret it.
Summary metrics
The four headline cards at the top of the dashboard summarise your cluster’s financial state over the selected time period.
Total cost
The total amount spent on compute resources (CPU + memory + storage) across all monitored clusters in the selected period. This is the fully-allocated cost — every pod’s cost is accounted for, including idle capacity distributed proportionally back to workloads.
Efficiency score
A percentage from 0–100% representing how effectively your requested resources are being used. An efficiency of 70% means 30% of what you’ve paid for is going to waste.
Efficiency is calculated from the ratio of actual resource usage to resource requests, weighted by cost. See Efficiency Scoring for the full calculation.
Target ranges:
| Score | Grade | Interpretation |
|---|---|---|
| 90–100% | A | Excellent — tight resource management |
| 75–89% | B | Good — minor optimisation opportunities |
| 60–74% | C | Fair — meaningful savings available |
| 45–59% | D | Poor — significant overprovisioning |
| 30–44% | E | Very poor — major waste |
| 0–29% | F | Critical — immediate action recommended |
Idle cost
The amount of money spent on cluster capacity that was not used by any workload during the period. Idle cost comes in two forms:
- Overprovisioned waste — Resources that pods requested but did not actually use
- Unallocated waste — Node capacity that no pod requested at all
See Understanding Idle Costs for how to reduce each type.
Pod count
The number of distinct pods observed over the selected period. This is a distinct count — a pod that restarted multiple times is counted once. Useful for understanding scale trends.
Cost breakdown table
Below the summary cards, the cost breakdown table shows your spend by dimension. By default it shows namespaces, but you can change this in the Cost Explorer.
Each row shows:
- Name — The namespace, workload, or label value
- Cost — Total spend for the period in USD
- Share — Percentage of total cluster spend
- Efficiency — Colour-coded efficiency indicator for this dimension
- Trend — Arrow showing cost direction compared to the previous equivalent period
Reading efficiency colours
- Teal — Efficiency ≥ 75%
- Amber — Efficiency 45–74%
- Red — Efficiency below 45%
Trends chart
The daily cost trends chart shows how spending has evolved over the selected time window. The chart breaks down each day’s cost into:
- Allocated cost — Resources actively used by pods
- Idle cost — Unallocated or unused capacity (shown as a separate area)
Look for these patterns:
| Pattern | Possible cause |
|---|---|
| Sudden spike | New deployment, autoscaler scale-out, or incident response |
| Gradual upward trend | Organic growth or resource request creep |
| Flat line with high idle | Overprovisioned node pool or static cluster size |
| Weekend dip | Workloads that scale down at weekends (opportunity for automation) |
Time range selection
The time range picker in the top-right corner controls the period shown across all dashboard panels:
| Option | Use for |
|---|---|
| Last 24 hours | Checking recent deployments |
| Last 7 days | Weekly review (default) |
| Last 30 days | Monthly reporting |
| Last 90 days | Quarterly trends |
The dashboard reflects costs with up to 2 minutes of latency. The agent ships metrics every 15 seconds; the processor ingests them in near-real-time.
Multi-cluster view
If you have multiple clusters connected, the dashboard shows an aggregate view across all of them by default. Use the Cluster filter in the top-left to scope the view to a single cluster.
Each cluster’s costs are calculated independently using provider-specific pricing for that cluster’s region and instance types.