Data quality
The Data Quality page keeps an eye on the data behind your dashboards: define tests that assert what should be true, run them on demand, and let the AI scan surface issues you didn’t think to test for.
Create Test defines a rule against a model table/column. Six rule types:
| Type | Asserts |
|---|---|
| Null check | a column has no (or bounded) missing values |
| Unique | values are distinct — the classic primary-key check |
| Range | numeric values stay within bounds |
| Pattern | values match a regex (emails, codes, IDs) |
| Row count | a table’s size is in the expected range |
| Custom SQL | anything you can phrase as a query |
Run tests individually or Run All; each shows PASS / WARN / FAIL / ERROR with its last-run time. The page’s KPI cards summarize tests passed, tests failed, and AI anomalies at a glance.
AI scan & suggestions
Section titled “AI scan & suggestions”AI Scan analyzes your models and files issues it finds — the summary tells you how many models were scanned and issues found. Findings land under Suggestions and AI Anomalies for you to review, accept, or dismiss. (The same review queue holds AI-suggested model relationships.)
Monitoring views
Section titled “Monitoring views”Beyond tests, the tabs give you standing views of data health:
- Overview — the quality summary.
- Column Profile — per-column statistics and distributions.
- Issues — everything currently wrong, in one list.
- Freshness — stale-table monitoring: which tables haven’t seen new data.
- Scan History — past runs.
- Lineage — relationship health across the model.
Related
Section titled “Related”- Data models & relationships
- Troubleshooting connections — when the problem is upstream of quality.