Available Metrics

The following metrics are supported by Metric Monitors. Users can also define custom metrics using SQL. The metric names for Monitors as Code (MaC) config are listed in the MaC Name column for each metric.

Pipeline

Metrics that detect freshness and volume anomalies. Often used when at least one segmentation field is selected. The traditional Pipeline Observability monitors do not support segmentation, so segmented freshness and volume monitoring are addressed through these metrics.

Metric	Description	Threshold Types	MaC Name
Row count	Alerts to unusual spikes or drops in the count of rows	ML	ROW_COUNT_CHANGE
Relative row count (%)	Alerts to unusual spikes or drops in the count of rows for a particular segment, relative to the overall number of rows (learn more)	ML	RELATIVE_ROW_COUNT
Time since last change in row count	Alerts to unusually long periods of time since the last change in row count	ML	TIME_SINCE_LAST_ROW_COUNT_CHANGE

Uniqueness

Uniqueness metrics check for duplicates in unique keys like UUIDs, and for changes in cardinality.

Metric	Description	Column Types	Threshold Types	MaC Name
Unique (%)	Percentage of unique values across all rows	All	Manual & ML	UNIQUE_RATE
Unique (count)	Count of unique values	All	Manual	UNIQUE_COUNT
Duplicate (count)	Count of duplicate values, calculated as the difference between the total number of rows with non-null values and total number of distinct non-null values	All	Manual	DUPLICATE_COUNT

Completeness

Completeness metrics check for ways that data can be null or otherwise unpopulated.

Metric	Description	Column Types	Threshold Types	MaC Name
Null (%)	Percentage of rows where value is null	All	Manual & ML	NULL_RATE
Null (count)	Count of rows with null value	All	Manual	NULL_COUNT
Non-null (count)	Count of rows with non-null value	All	Manual	NON_NULL_COUNT
Empty string (%)	Percentage of rows where the value is an empty string	String	Manual & ML	EMPTY_STRING_RATE
Empty string (count)	Count of rows where the value is an empty string	String	Manual	EMPTY_STRING_COUNT
All spaces (%)	Percentage of rows where the text value is whitespace only	String	Manual & ML	TEXT_ALL_SPACE_RATE
All spaces (count)	Count of rows where the text value is whitespace only	String	Manual	TEXT_ALL_SPACE_COUNT
NaN (%)	Percentage of rows with a value of NaN (Not a Number), meaning the value is undefined	Numeric	Manual & ML	NAN_RATE
NaN (count)	Count of rows with a value of NaN (Not a Number), meaning the value is undefined	Numeric	Manual	NAN_COUNT
"none" or "null" (%)	Percentage of rows where the text value is a null keyword ("none", "null", "nil", "nothing" or "n/a")	String	Manual & ML	TEXT_NULL_KEYWORD_RATE
"none" or "null" (count)	Count of rows where the text value is a null keyword ("none", "null", "nil", "nothing" or "n/a")	String	Manual	TEXT_NULL_KEYWORD_COUNT

Distribution

Distribution metrics check for shifts in the numeric profile of data.

Metric	Description	Column Types	Threshold Types	MaC Name
Mean	Average value across all rows	Numeric	Manual & ML	NUMERIC_MEAN
Median	Median value across all rows	Numeric	Manual & ML	NUMERIC_MEDIAN
Min	Minimum value across all rows	Numeric	Manual & ML	NUMERIC_MIN
Max	Maximum value across all rows	Numeric	Manual & ML	NUMERIC_MAX
20th percentile	20th percentile of values	Numeric	Manual & ML	PERCENTILE_20
40th percentile	40th percentile of values	Numeric	Manual & ML	PERCENTILE_40
60th percentile	60th percentile of values	Numeric	Manual & ML	PERCENTILE_60
80th percentile	80th percentile of values	Numeric	Manual & ML	PERCENTILE_80
Zero (%)	Percentage rows with value equal to zero	Numeric	Manual & ML	ZERO_RATE
Zero (count)	Count of rows with value equal to zero	Numeric	Manual	ZERO_COUNT
Negative (%)	Percentage of rows that have a negative value	Numeric	Manual & ML	NEGATIVE_RATE
Negative (count)	Count of rows that have a negative value	Numeric	Manual	NEGATIVE_COUNT
Standard deviation	Standard deviation of values	Numeric	Manual & ML	NUMERIC_STDDEV
Sum	Sum of values across all rows	Numeric	Manual	SUM
True (%)	Percentage of rows where the value is true	Boolean	Manual & ML	TRUE_RATE
True (count)	Count of rows where the value is true	Boolean	Manual	TRUE_COUNT
False (%)	Percentage of rows where the value is false	Boolean	Manual & ML	FALSE_RATE
False (count)	Count of rows where the value is false	Boolean	Manual	FALSE_COUNT

Validity

Validity metrics check that values are honoring expected and usable formats, including common data entry errors.

Metric	Description	Column Types	Threshold Types	MaC Name
String length max	Maximum character length	String	Manual	TEXT_MAX_LENGTH
String length min	Minimum character length	String	Manual	TEXT_MIN_LENGTH
String length mean	Average character length	String	Manual	TEXT_MEAN_LENGTH
String length standard deviation	Standard deviation of character length	String	Manual	TEXT_STD_LENGTH
Integer (%)	Percentage of rows where the text value is an integer	String	Manual & ML	TEXT_INT_RATE
Not integer (count)	Count of rows where the text value is not an integer	String	Manual	TEXT_NOT_INT_COUNT
Float (%)	Percentage of rows where the text value is a floating-point number	String	Manual & ML	TEXT_NUMBER_RATE
Not float (count)	Count of rows where the text value is not a floating-point number	String	Manual	TEXT_NOT_NUMBER_COUNT
UUID (%)	Percentage of rows where the text value is a UUID (e.g., b391e7d2-80e0-4749-8c60-c76031c43dfe). The comparison is case insensitive.	String	Manual & ML	TEXT_UUID_RATE
Not UUID (count)	Count of rows where the text value is not a UUID (e.g., b391e7d2-80e0-4749-8c60-c76031c43dfe). The comparison is case insensitive.	String	Manual	TEXT_NOT_UUID_COUNT
SSN (%)	Percentage of rows where the text value is formatted as Social Security Number from the United States (e.g., 123-45-6789)	String	Manual & ML	TEXT_SSN_RATE
Not SSN (count)	Count of rows where the text value is not formatted as Social Security Number from the United States (e.g., 123-45-6789)	String	Manual	TEXT_NOT_SSN_COUNT
USA phone number (%)	Percentage of rows where the text value is a USA phone number (e.g., (123) 456-7890, +1 (123) 456-7890 etc)	String	Manual & ML	TEXT_US_PHONE_RATE
Not USA phone number (count)	Count of rows where the text value is not a USA phone number (e.g., (123) 456-7890, +1 (123) 456-7890 etc)	String	Manual	TEXT_NOT_US_PHONE_COUNT
USA state code (%)	Percentage of rows where the text value is a USA state code (e.g., TX)	String	Manual & ML	TEXT_US_STATE_CODE_RATE
Not USA state code (count)	Count of rows where the text value is not a USA state code (e.g., TX)	String	Manual	TEXT_NOT_US_STATE_CODE_COUNT
USA ZIP code (%)	Percentage of rows where the text value is a USA ZIP code (e.g., 94109)	String	Manual & ML	TEXT_US_ZIP_CODE_RATE
Not USA ZIP code (count)	Count of rows where the text value is not a USA ZIP code (e.g., 94109)	String	Manual	TEXT_NOT_US_ZIP_CODE_COUNT
Email (%)	Percentage of rows where the text value is an email address (e.g., [email protected])	String	Manual & ML	TEXT_EMAIL_ADDRESS_RATE
Not email (Count)	Count of rows where the text value is not an email address (e.g., [email protected])	String	Manual	TEXT_NOT_EMAIL_ADDRESS_COUNT
Timestamp (%)	Percentage of rows where the text value is an ISO-8601 format date or timestamp (e.g., 2023-01-12 16:50:11.045746 +00:00, 2023-01-12T16:50:11Z)	String	Manual & ML	TEXT_TIMESTAMP_RATE
Not timestamp (count)	Count of rows where the text value is not an ISO-8601 format date or timestamp (e.g., 2023-01-12 16:50:11.045746 +00:00, 2023-01-12T16:50:11Z)	String	Manual	TEXT_NOT_TIMESTAMP_COUNT
In past (%)	Percentage of rows where the value is a date or time occurring before the time when the metric is evaluated, with a granularity of seconds used for comparison	Date	Manual & ML	PAST_TIMESTAMP_RATE
In past (count)	Count of rows where the value is a date or time occurring before the time when the metric is evaluated, with a granularity of seconds used for comparison	Date	Manual	PAST_TIMESTAMP_COUNT
In future (%)	Percentage of rows where the value is a date or time occurring after the time when the metric is evaluated, with a granularity of seconds used for comparison	Date	Manual & ML	FUTURE_TIMESTAMP_RATE
In future (count)	Count of rows where the value is a date or time occurring after the time when the metric is evaluated, with a granularity of seconds used for comparison	Date	Manual	FUTURE_TIMESTAMP_COUNT
Unix time 0 (%)	Percentage of rows where the value is Unix time 0	Date	Manual & ML	UNIX_ZERO_RATE
Unix time 0 (count)	Count of rows where the value is Unix time 0	Date	Manual	UNIX_ZERO_COUNT

Custom

Users can define custom metrics using SQL. This is helpful to create a metric that:

Isn't available above, such asAPPROX_PERCENTILE(order_amount, 0.99)
References multiple fields at once, such as
- AVG(price + tax_collected)
- COUNT(DISTINCT(CONCAT(invoice_number, line_item)))/COUNT(*)

To create a custom metric, select Custom in the metric dropdown in the Define alert conditions section of a Metric monitor. Currently, only manual thresholds are supported for custom metrics.

To create a custom metric, provide a metric name and a SQL expression, then test the metric to confirm it works.

Relative row count

Relative row count will alert to sudden shifts in the distribution of categorical fields. It tracks the percentage of rows in a given segment compared to the overall number of rows, and alerts when the percentage for a segment suddenly spikes higher or drops lower. Because of this, segmentation is required when using this metric. Common examples:

An events table has an app_version field. The user would like to be alerted if any app_version is suddenly receiving a disproportionately high or low number of rows relative to overall volume, as it may indicate issues with that version.
An orders table has a country field. The user would like to be alerted if any country is suddenly receiving a disproportionately high or low number of rows relative to overall volume, as it may indicate an issue with processing order from that country.

Relative row count can also alert when new segments appear, so long as they represent a significant shift in the distribution of rows. However, to always alert when any new segment appears, we recommend configuring a Validation monitor with the is not in set operator.

Relative row count was formerly its own monitor type called Dimension Tracking. Starting in October 2024, Dimension Tracking monitors could no longer be created through the UI and users were guided to use relative row count instead.