Skip to main content

​
What is a Schema metric?

Schema metrics monitor the structure of a data asset — whether the expected columns are present, whether data types match, and whether the schema has drifted from its baseline. Schema changes are one of the most common causes of silent downstream failures in data pipelines. Type: Automatic — Schema metrics are enabled on all connected assets by default as part of the Essential observability profile. No configuration is required to activate them. Schema metrics answer the question: Is the shape of the data what consumers expect? Prizm captures a schema snapshot on the first profiling run and establishes it as the baseline. On each subsequent run, the current schema is compared against the baseline. Any structural difference triggers a schema change detection event.

​
Metrics

​
What Schema monitoring detects

Prizm tracks eight categories of structural change: For each detected change, Prizm records the before and after state — column name, data type, length, and constraint changes (PK, FK) — so you can see exactly what changed without querying the data platform directly.

​
Example use cases

Detecting an accidental column drop: A database migration script drops a column that downstream dbt models depend on. Prizm fires a Schema alert immediately on the next scan — before any dbt jobs fail — allowing the team to roll back the migration before impact propagates. Catching an upstream attribute resize: A source system changes transaction_amount from NUMERIC(10,2) to VARCHAR during a system upgrade. The column appears in queries but downstream aggregations silently return nulls. Prizm’s Schema alert fires on the resize before any dashboard shows incorrect totals. Surfacing unannounced column additions: An upstream team adds new columns to a shared source table without notifying consumers. The Schema alert surfaces the addition immediately — allowing data consumers to assess whether to incorporate the new columns in their models.
Image
Schema alerts fire on any detected structural change. Planned schema changes (intentional column additions, DDL deployments) will also trigger alerts. Acknowledge expected changes promptly from the Alert page so alerts remain meaningful for actual unintended drift.

Configure

Set thresholds, understand the diff view, acknowledge schema changes, and configure alerts.