Cloud Data Loss Prevention (Cloud DLP) is now a part of Sensitive Data Protection. The API name remains the same: Cloud Data Loss Prevention API (DLP API). For information about the services that make up Sensitive Data Protection, see Sensitive Data Protection overview.

Metrics reference

This page lists and describes all metrics that are gathered in data profiles.

There are three types of data profiles—project data profiles, table data profiles, and column data profiles.

Project data profiles

Each project data profile has the following fields. The values for these fields are aggregated based on the resources profiled within the project.

Insights

Project data profiles provide the following insights:

Data risk: Level of risk associated with the data at its current state. For more information, see Sensitivity and data risk levels.
Sensitivity: Score indicating the sensitivity level for this project. For more information, see Sensitivity and data risk levels.

Metadata

Project data profiles provide the following metadata:

Last profile generated: Date and time the profile was last generated.
Project ID: ID of the project that was profiled.
Resource name: Fully qualified name of the data profile.
Status: Icon that indicates the status of the profiling operation.

Table data profiles

Each table data profile has the following fields:

Insights

Table data profiles provide the following insights:

Data risk: Level of risk associated with the data at its current state. For more information, see Sensitivity and data risk levels.
Sensitivity: Score indicating the sensitivity level for this table. For more information, see Sensitivity and data risk levels.

Metadata

Table data profiles provide the following metadata:

Database: The database containing the table that was profiled. This field applies only to Cloud SQL discovery.
Dataset ID: ID of the dataset that contains this table.
Encryption: Whether encryption for this table is managed by Google or by your organization.
Expiration time: Optional. The time when this table expires.
Failed column count: The number of columns skipped in this table because of an error.
Inspect config snapshot: Snapshot of the inspection template that was used when the profile was generated. For more information, see Data profile snapshots.
Instance: The instance containing the table that was profiled. This field applies only to Cloud SQL discovery.
Last profile generated: Date and time the profile was last generated.
Latest update in BigQuery: Date and time this table was last modified.
Project ID: ID of the project that contains this table.
Public: Whether this table is available to all users or restricted to certain users.
Resource labels: Labels that the table had at the time the profile was generated.
Resource name: Fully qualified name of the data profile.
Row count: Number of rows in this table when the profile was generated.
Scanned column count: The number of columns profiled in this table.
Service account: Number of service accounts with IAM permissions to access this table.
Status: Indication of whether the profile succeeded in generation.
Table ID: ID of this table.
Table creation time: Date and time the table was created.
Table size: The size of this table when the profile was generated.
Type: The type of discovery performed.

Column data profiles

Each column data profile has the following fields:

Insights

Column data profiles provide the following insights:

Data risk

Level of risk associated with the data at its current state. For more information, see Sensitivity and data risk levels.

Sensitivity

Score indicating the sensitivity level for this column. For more information, see Sensitivity and data risk levels.

Predicted infoType

If a single built-in or custom infoType clearly predominates over others in the column, Sensitive Data Protection sets this field to that infoType. Otherwise, this field has no value.

To view a list of all infoTypes detected in the column, refer to the Other infoTypes field.

Sensitive Data Protection scans for only the infoTypes that you specified in the inspection template. Thus, only those infoTypes can appear in the Predicted infoType field. For example, if the column has email addresses, but you didn't include the EMAIL_ADDRESS infoType detector in your inspection template, then this field doesn't contain EMAIL_ADDRESS.

If the column data predominantly matches several closely related infoTypes that belong to the same general category, Sensitive Data Protection sets this field to the more general infoType. For example, if the column predominantly has a mix of PASSPORT, AUSTRALIA_PASSPORT, and CANADA_PASSPORT infoTypes, the Predicted infoType field is set to PASSPORT. The Other infoTypes field shows the more specific infoTypes and their estimated prevalence.

Other infoTypes

InfoTypes detected in the column that don't have a strong enough signal to be considered that column's predicted infoType. In this document, see Predicted infoType.

For data profiles generated after October 13, 2022, each infoType listed in this field has an estimated prevalence. The estimated prevalence is an approximate percentage of non-null rows in which the infoType was detected.

For example, suppose you have a column that has the following metrics:

Predicted infoType: FDA_CODE
Other infoTypes: PERSON_NAME (2%), STREET_ADDRESS (1%)

In this example, there is a strong indication that the column contains FDA codes. Sensitive Data Protection also determined that approximately 2% of non-null rows in the column might contain person names and 1% might contain street addresses.

Sensitive Data Protection scans for only the infoTypes that you specified in the inspection template. Thus, only those infoTypes can appear in the Other infoTypes field. For example, if the column has email addresses, but you didn't include the EMAIL_ADDRESS infoType detector in your inspection template, then this field doesn't contain EMAIL_ADDRESS.

Estimated null proportion

Approximate proportion of null values in this column, categorized as high, medium, low, or very low. This value is high if a large proportion of entries in this column is null.

Estimated uniqueness

An estimate of how much of the data in this column is unique, categorized as high, medium, or low. A high uniqueness level suggests that the column contains distinct values. A high presence of unique values can indicate that the column contains identifiers.

A low uniqueness level suggests that the column contains many common values such as enums or boolean values.

If Sensitive Data Protection determines that there aren't enough rows in the table for it to calculate this metric, this value is blank.

Free text score

The probability that this column contains freeform text. A value close to 1 indicates the column is likely to contain freeform or natural-language text. Possible values range from 0 through 1.

A high free text score can increase a column's data risk and sensitivity levels.

Metadata

Column data profiles provide the following metadata:

Database: The database containing the table column that was profiled. This field applies only to Cloud SQL discovery.
Data type: The data type of the contents of this column.
Dataset ID: ID of the dataset that contains this table column.
Field ID: Name of the column.
Instance: The instance containing the table column that was profiled. This field applies only to Cloud SQL discovery.
Instance location: Location of the instance containing the table column that was profiled. This field applies only to Cloud SQL discovery.
Last profile generated: Date and time the profile was last generated.
Policy tags: Indicates if a policy tag is applied to the column. For information on best practices for using policy tags, see Using policy tags in BigQuery.
Project ID: ID of the project that contains this table column.
Resource name: Fully qualified name of the data profile.
Status: Icon that indicates the status of the profiling operation.
Table ID: ID of the table that contains this column.