Reliability
The Reliability page monitors the health of your data and alerts you when something breaks — before it affects decisions.
How it works
OnlyMetrix continuously runs 4 automated checks on every monitored metric:
| Check | What it detects | Example |
|---|---|---|
| Freshness | Data hasn't been updated within the SLA | "invoices table is 6 hours stale (SLA: 2h)" |
| Schema drift | Tables or columns have been renamed/removed | "column country no longer exists" |
| Volume anomaly | Row count dropped >50% or grew >5x | "orders table dropped from 100K to 12K rows" |
| Null spike | Null rate jumped significantly | "email null rate spiked from 2% to 45%" |
Metric statuses
- Healthy — All checks passing, data is trustworthy
- Degraded — Warnings present, data may be incomplete
- Unreliable — Critical issues, data should not be used for decisions
Alerts
The Reliability page shows a summary banner with counts of healthy, degraded, and unreliable metrics. Metrics needing attention are listed first with their specific issues.
Notifications
Click Notify me when ready on any unreliable metric to get alerted when it recovers. Supports:
- Webhook (HTTPS)
- Slack
Integration with the AI agent
The agent automatically checks reliability before every query. If a metric is degraded, the agent warns you in its response. If it's unreliable, the query is blocked entirely with an explanation.