Exclude specific events and user properties from ads personalization

You can exclude specific events or user properties from being used to personalize ads, and thus reserve that data for measurement purposes only.

How audiences are affected

If you exclude an event or user property from ads personalization, then any audience that is based on that data is not eligible for export from Analytics to any of Google's advertising products (e.g., Google Ads, Display & Video 360, Search Ads 360). This also applies to any audience whose component audiences are based on that excluded data.

Audiences that include such events and user properties are still available within Analytics for use in reports, Analysis, and audience building, and can be exported to non-advertising products like Optimize and BigQuery.

How individual events and user properties are affected

The individual events and user properties excluded from ads personalization are available for export to advertising products, though they cannot be used by those products to personalize ads.

How to exclude or include

You need Edit permission to manage these settings.

To exclude an event or user property from being used to personalize ads:

  1. In the left pane, click either All events or User properties.
  2. In the row for that event or user property, click More.
  3. Click Mark as NPA (no personalized ads).

If you want to reinclude an event or user property:

  1. In the left pane, click either All events or User properties.
  2. In the row for that event or user property, click More.
  3. Click Unmark as NPA.

Marking an event or user property as NPA does not affect the data that has already been exported to Google Ads or Display & Video 360. You must login to Google Ads or DV360 to manage the ads-personalization uses of any exported event or user-property data.

Metrics derived from NPA events (e.g., LTV) are not automatically treated as NPA. Therefore, you must login to Google Ads or Display & Video 360 to manage the ads-personalization uses of derived metrics.

Was this helpful?
How can we improve it?