A quick introduction to Ads Data Hub

Editor's note: This post was originally published in June 2020 and updated in September 2026 to reflect changes to Ads Data Hub privacy checks, audience features, and Google's Data Transfer redaction policy.

Ads Data Hub (ADH) gives you a way to answer questions about your Google media that no standard report can reach — cross-platform reach, custom attribution, audience overlap — while keeping user-level data inside Google's environment. It lets you query Campaign Manager 360, Display & Video 360, Google Ads, and YouTube Reserve data with the same SQL you already write in BigQuery.

Ads Data Hub matters most when you compare it to what Data Transfer can still give you. Plenty of agencies and advertisers pull raw event-level Data Transfer files from Campaign Manager 360 and Display & Video 360, and those files still arrive. What has changed is how much of the user ID column survives. Google stopped populating encrypted UserID and PartnerID for YouTube inventory globally, then extended redaction to users in California, the EEA, and more recently Florida, Texas, Oregon, and Montana.

The practical result is a file that still works for impression and click reporting but has a shrinking, geographically uneven base for anything that depends on stitching a user across events. Ads Data Hub is where that user-level join still happens, under privacy rules that govern what comes back out.

How analysis works in Ads Data Hub

You can conduct analysis in Ads Data Hub by running query jobs in the Ads Data Hub environment with your BigQuery table as a destination for the results. You can either utilize pre-built queries or write custom queries on top of Ads Data Hub-provided as well as your own data. Table schemas in ADH are very similar to the ones produced by the data transfer service and can be reviewed directly from Ads Data Hub's UI or in more detail in Google's  documentation.

In comparison to running queries in BigQuery, two additional settings are available within Ads Data Hub:

• Parameters, which allow running the queries with an argument from the UI or API
• Filtered row summaries, which tally the rows that privacy checks removed so your totals stay consistent even when individual breakdowns get dropped

Most queries need roughly 50 unique users per result row. Queries that touch only clicks and conversions can report on about 10. If you use noise injection instead of difference checks, that drops to around 20. Events with null user IDs don't count toward the threshold at all, which is the part that surprises people the first time a breakdown they expected comes back mostly empty.

How to build audiences in Ads Data Hub

• The result of an Audience query should be a list of user ids
• Instead of a destination table, your existing or new user list should be configured as a “destination”

Setting up and running query jobs

• Query Type (Analysis or Audience)
• Query Name (to select the query that was written)
• Ads data from (selecting an account providing the data)
• Destination Table (location within your BigQuery where the results will be stored)
• Start Date
• End Date
• Time Zone (to "calibrate" the midnight time of start and end date)
• Arguments (if parameters were added to the query)

Audience queries will include user list fields in place of the destination table. This gives you the option to update an existing user list or create a new one.

You can then track the query execution within the Ads Data Hub UI.