Tables Configuration
Tables Configuration — the Datasource Manager page once onboarding is complete — is where you manage the active schema: every table and field active in BPP, with their labels, descriptions and visibility.
After activation the source schema is add-only. You can edit labels, descriptions and visibility, and add new tables via Add Tables. You cannot rename, retype or remove the fields of your source tables. The fields of the user model follow their own rules: see Editing the User Model After Activation.
How to access it
Open the system menu (the gear in the header) and choose Datasource Manager. The entry appears once the initial DSM Onboarding has been activated.
The page shows, from top to bottom: the Datasource & BigQuery Connection details (datasource, connection, GCP project, region, external and internal dataset), the Unified User Model, the Source Tables (collapsed until you open one) and, once the platform has created any (AI model results, for example), the Platform Generated Tables.
What you can do
| Action | Description |
|---|---|
| Edit field label | Change the display name of a field as it appears in audience and signal filters. |
| Edit field description | Add or update a short description to help users understand the field's meaning. |
| Show / hide fields | Toggle field visibility to keep the filter panel clean. Hidden fields are not deleted — they can be shown again at any time. |
| Edit the user model | Change fields, sources, priority and aggregation rules. See Editing the User Model After Activation. |
Personal data
There is no per-field PII setting. What protects personal data is where it lives and how it leaves:
- It stays in your BigQuery project. The
_bppcopies and the identity resolution tables are written to your own datasets. BPP does not copy your identifiers into its own database. - Ad platforms receive it hashed. Google Ads, Meta Ads, Microsoft Ads and OpenAI Ads get emails and phone numbers normalized and SHA-256 hashed at send time, in the format each platform requires. A value already hashed in your warehouse is sent as it is, never hashed twice.
- External enrichment is closed for identifiers. Every column declared as a user identifier starts closed to enrichment providers and language models, whatever its type is called. Bytek staff can open a column on request, and a decision already taken is kept.
Identifiers
Mark as identifiers only IDs of people. The ID of a case, an opportunity or an order is not an identifier: leave it out of the identifiers. Click ids (gclid, fbclid, msclkid…) are never identifiers either: they change on every visit. Activation refuses a column whose name or ID type says it is a click id.
Every identifier type is treated the same way, whatever its name: BPP does not need to know whether a type holds emails, cookies or CRM IDs. To keep a shared device, a shared login or a shop email from merging different people, BPP applies the same limits to every type: see Quarantines.
Email addresses are recognised by their value. For every identifier value shaped like an email, BPP keeps the SHA-256 of each spelling an advertising platform accepts, and sends each platform the spellings it takes, its preferred one first:
| Destination | First | Then |
|---|---|---|
| Google Ads, Google Data Manager | Basic | |
| Meta, OpenAI Ads | Basic | |
| Microsoft Advertising | Microsoft | the other spellings |
The Microsoft spelling goes to Microsoft Advertising only. Each platform limits how many values it takes per person (for example Meta 5, OpenAI Ads 3, Microsoft offline conversions 1), so BPP sends the preferred spelling of each email before any second spelling. Advertising platforms never receive an address in clear. A column that holds emails in clear shows the Multi-hash badge; see Emails in clear (multi-hash) for the spellings.
When a destination uses an email column that is not an identifier, BPP hashes it with that destination's rule only.
How to save changes
After making edits, click Save changes at the bottom of the page. Changes take effect immediately — no reconciliation is required.
Adding new tables
If your BigQuery dataset has grown and you need to bring new tables into BPP, use the Add Tables flow. Click Discover New Tables on the Datasource Manager page to start the process. The same flow also picks up new columns detected on tables you've already onboarded — see Column drift on existing tables.
Editing the user model
Field names, sources, and aggregation rules from Step 3 aren't locked in after activation — see Editing the User Model After Activation.
Starting over
Once the datasource is onboarded there is no Start over on this page. To onboard again from scratch, ask Bytek: a superuser can use Reset datasource on the Identity Resolution page. It removes the configuration of every table and field and the Data Source Manager analysis, and takes the customer back to the beginning of the Data Source Manager flow. Profiles keep their BPP user ids, and the waterfall is paused until the new onboarding is completed. The button is disabled while a reconciliation is running.
Warning icons on tables and fields
You may occasionally see a warning icon next to a table or field on this page. This means BPP's Schema Integrity Check has detected a mismatch between what's registered here and your live BigQuery dataset. Hover over the icon for details, and see the Schema Integrity Check page for what it means and what to do.