Skip to main content

Add Tables

The Add Tables flow lets you bring new BigQuery tables into the platform, and lets you pick up new columns BPP has detected on tables it already knows about. Use it any time your dataset grows and you want the change available for AI models, audiences, and signals.

Video tutorial — A short walkthrough is available here: Watch on Loom (TODO!)

How to start​

Open the Datasource Manager page (system menu, gear in the header) and click Discover New Tables.

Discover New Tables button

BPP will scan your BigQuery dataset for tables not yet part of the active schema.

The flow​

Add Tables has its own review flow — it does not re-run the 5-step onboarding wizard. BPP analyses the new tables, you review them, and you activate. The Used ID Summary screen shows all identifier groups together (new and already active), and fields belonging to the new tables are marked with a New! badge so you can quickly verify compatibility. ID types already active keep their name; new ID types can still be renamed before you activate.

Used ID Summary with New! badges

Proposals to review​

The review of the new tables marks the proposals the analysis is not sure about (confidence below 75%) with a yellow badge, so you can check them first:

  • Type to review next to a table name: the proposed table type is uncertain.
  • To review under a field name: the analysis is not sure about what it proposed for that field. The table header also shows how many of its fields are to review, for example 12 fields · 3 to review.

Hover over a badge to see why the analysis made that proposal and how confident it is, for example (confidence 60%).

To see only these items, tick Only items to review above the tables. The checkbox shows how many tables have something to review. With the filter on, you see only those tables and, inside them, only the fields to review (a table flagged only for its type shows all its fields). Tables and fields with a validation error stay visible too, so you can always see what blocks Validate & Continue.

The badges are hints: you can continue without changing anything. They appear only on analyses that report a confidence.

Important: After clicking Activate and confirming, the reconciliation of the new tables starts. Click Back to the platform to keep working while it runs. Until it finishes, a banner says that data is loading and the Run button of audiences, signals and profile syncs is disabled. New event tables partitioned by date are read for the last 7 days first, and their history arrives with the next nightly update. See While the first reconciliation runs.

Column drift on existing tables​

Add Tables isn't only for brand-new tables. BPP also compares the live BigQuery columns of tables it already knows about against what's registered. When it finds new columns, the review screen shows a New columns detected on existing tables panel, grouped by table.

Each new column is reviewed like a field of a new table: label, description, user identifier and its ID type, and visibility, pre-filled with the analysis suggestion. Every column starts included; untick Add on the ones you don't need yet — they stay in the panel for later. Click Add n to schema on a table to add its included columns. This is a scoped update — it does not onboard the table again and does not touch the fields already active.

A table that gains a user identifier this way is fully re-reconciled at the next run, so its rows are linked to profiles through the new identifier too.

If a drifted column shows up unexpectedly, check the Schema Integrity Check page — the same mismatch is what that check reports in the notifications.

Edge cases​

No new tables found: If BPP does not detect any new tables, a "No New Tables Found" message is shown. Click Back to Datasource Manager to return.

Analysis error: If the analysis fails, an error screen is shown. Contact support and share the error details.

Reset: At any point during the review phase, click Reset to cancel the flow and return to the operational state without making any changes.

Tips​

  • You can run the Add Tables flow as many times as needed — repeat it whenever new tables are added to your BigQuery dataset.
  • Exports can't be run manually while the new tables are reconciled. Plan activations outside the hours when you export.
  • Tables added via this flow behave exactly like tables configured during the initial onboarding.
Last updated on