What is a Frequency metric?
Frequency metrics measure how often specific values, lengths, and value ranges appear in a column. They expose broken enumerations, boundary violations, and distribution shifts that generic completeness or uniqueness checks do not catch. Type: Configurable — Frequency metrics are part of the Structural (Advanced) profile. They are enabled per column through the Configure Structural Metrics panel. Frequency metrics answer the question: Are the values in each column conforming to their expected set and boundaries?Metrics
How frequency monitoring works
Prizm captures the value, length, and range distributions on each profiling run. Baselines are learned automatically. Alerts fire when new values appear in previously stable enumerations, when length distributions shift significantly, or when the proportion of out-of-range values crosses a configured threshold. For Regular Expressions, the user supplies one or more regex patterns. Prizm evaluates each pattern against the column and records the match rate per run. Alerts fire when the match rate drops below a configured threshold.Example use cases
Detecting a new order status: An e-commerce platform introduces a new fulfillment status"SPLIT_SHIPMENT". The Enum metric on orders.status fires an alert the first time this value appears — before any downstream order-processing pipeline has been updated to handle it.
Phone number format validation: A phone_number column should always match the pattern ^\+[0-9]{10,15}$. A Regular Expressions metric with this pattern alerts when a batch of records loaded without the country code prefix, dropping the match rate from 100% to 73%.
Configure
Enable Frequency metrics per column, configure regex patterns, set thresholds, and manage alerts.