What is a Distribution metric?
Distribution metrics capture the character-level and value-level composition of a column — nullness, character types, whitespace, distinct values, and numeric sign. They reveal structural quality issues that aggregate counts and row-level checks miss entirely. Type: Configurable — Distribution metrics are part of the Structural (Advanced) profile. They are enabled per column by mapping the asset and attribute through the Configure Structural Metrics panel. Distribution metrics answer the question: What does the content of each column actually look like?Metrics
How distribution monitoring works
On each profiling run, Prizm profiles every configured attribute and records the distribution snapshot. Anomaly detection compares each metric against its historical baseline. For null-related metrics, absolute thresholds can also be configured — for example, “alert if Null Percentage exceeds 5%.”Example use cases
Email column validation: Acustomer_email column that should always contain a valid email address begins showing an increase in Special Character counts and a decrease in Alpha Numeric counts after a source system migration. Prizm alerts on both metrics, pointing to a format change in how emails are stored — discovered before the CRM team notices bounce rates increasing.
PII completeness monitoring: A tax_id column is legally required for all corporate accounts. Prizm monitors Null Percentage on the column with a threshold of 0%. Any null in the column immediately triggers an alert, allowing the compliance team to trace the record back to the source and remediate.
Distribution metrics run at the Attribute (column) level. They must be explicitly enabled for each column via the Configure Structural Metrics panel — they do not activate automatically on all columns.
Configure
Enable Distribution metrics per column, set thresholds, and configure alerts.