Comparisons

Volumetric and Metric checks share the same three comparison types. Each one interprets Min / Max differently, so choosing the right comparison determines what the check considers a signal.

Absolute Value

Absolute Value compares the raw measured value against a fixed range. It answers: is this value inside [Min, Max]?

• Anomaly rule. A measurement outside the range triggers an anomaly.
• Best for. Fields with an expected range that does not drift over time. Row counts that should sit between a lower and upper bound. Field metrics with a hard business limit (age BETWEEN 0 AND 120).
• Minimum measurements to render the chart. 2.
• Example. A daily_sales volumetric check with Min = 100, Max = 1000. A day recording 45 rows is out of range and raises an anomaly.

Absolute Change

Absolute Change compares each measurement against the previous one and measures how much the raw value moved. It answers: did the value jump or drop more than expected between scans?

• Anomaly rule. If the delta from the previous measurement falls outside [Min, Max], an anomaly is raised.
• Best for. Detecting sudden drops or spikes when the baseline itself drifts (a growing table, an evolving metric). The change matters more than the absolute value.
• Minimum measurements to render the chart. 3.
• Example. A weekly volumetric check with Min = -5000, Max = 5000. A week that adds 12,000 rows triggers an anomaly because the delta broke the +5,000 bound.

Percentage Change

Percentage Change compares each measurement against the previous one, but as a percentage rather than a raw delta. It answers: did the value shift by more than X percent between scans?

• Anomaly rule. If the percent change from the previous measurement falls outside [Min, Max], an anomaly is raised. Thresholds are decimals (0.2 means 20%).
• Best for. Comparing changes on data whose absolute scale varies over time. A percentage bounds behaviors relative to the current baseline rather than a fixed number.
• Minimum measurements to render the chart. 3.
• Example. A daily metric check on average order value with Min = -0.1, Max = 0.1 (±10%). A day where the average drops 25% breaks the -10% floor and raises an anomaly.

Choosing a comparison

Comparisons are set when the check is authored. Changing the comparison later requires editing the underlying check (Description, Tags, Metadata, comparison), which is handled from the Quality Checks flow, not from the chart-adjacent inline editors on the Observability page.