Data Source Manager

The Data Source Manager connects your BigQuery dataset to BPP. It reads your dataset's schema, lets you decide which tables and fields matter, and activates the platform so audiences, signals and AI models can use your data.

Copilot drafts the whole configuration; you review it in a 5-step wizard.

The 5-step wizard

Step	What you do
Step 1 — Classify Tables	Mark tables as Event, User, or Unused
Step 2 — User Identifiers	Select identifier fields and assign ID types
Step 3 — User Model	Build the unified user model with aggregation rules
Step 4 — Event Mapping	Label, describe and hide event table fields
Step 5 — Review & Activate	Review everything, activate

How it works

• Personal data stays in your perimeter. Identity resolution runs inside your own BigQuery project. BPP keeps no copy of your identifiers in its own database, and every ad platform receives emails and phone numbers hashed in the format it requires, at send time.
• _bpp tables are a faithful copy. Each _bpp table in your dataset is a copy of its source table, without exact duplicate rows, plus four platform fields. Nothing is hashed in your warehouse.
• The timestamp is chosen where it is used. Each signal (and feature) picks its own timestamp column. No column has to be called event_date.
• One unified user model. Step 3 builds the user_model_bpp table: one row per person, each field merged from ranked sources with a rule that fits its type.
• Your own identifier types. Create the ID types your business uses in Step 2.
• Editable after activation. The user model stays editable, and new columns of tables you already use can be added without onboarding again.

After activation

Once the platform is active, you can manage your tables at any time from the Datasource Manager page (see Tables Configuration). This is where you edit field labels, descriptions, visibility and the user model, without onboarding again.

To add new BigQuery tables, or new columns of tables you already use, use Add Tables, which has its own review flow.