Bigabid source
The Bigabid source connector syncs daily advertising spend by app, campaign, country, ad set, ad, and publisher site.
Setup
- Obtain a reporting API token from Bigabid.
- In your workspace, add a new Bigabid source.
- Enter the Bigabid API Token and connect the source.
- Create a connection, select the report stream and your destination, and configure the backfill window and schedule.
Authentication
Extract requests GET https://rep.bigabidserv.com/reporting/{token}. The API token is part of the URL path, rather than a query parameter or authorization header. Treat the full reporting URL as a secret.
When connecting the source, Extract requests the previous UTC day's report and checks its CSV headers. A report with valid headers and no data rows is accepted.
Configuration
| Field | Scope | Required | Description |
|---|---|---|---|
| Bigabid API Token | Source | Yes | Reporting API token provided by Bigabid. |
| Backfill (Days) | Connection | Yes | Number of recent days to re-pull so finalized reporting data is refreshed. Defaults to 7. |
Sync behavior
Extract requests one day at a time, with both start_date and end_date set to that day. Reports use date as the partition key. Subsequent runs re-read the configured backfill window, and the stream cursor advances only after all requested days finish successfully.
Reports are downloaded and parsed as a stream, including gzip-compressed responses, so the entire report does not need to fit in memory. Requests allow up to three retries for transient failures before report streaming begins. A stalled read times out after 60 seconds; this is not a limit on total report duration. If a download or CSV parsing fails after streaming begins, the run fails without advancing the cursor.
With an Append destination, such as Google Sheets, backfill can append rows that were already loaded. Retrying a failed run can also repeat rows written before the failure. Use a destination and load mode that support date-partition replacement when you need refreshed reports without appended duplicates.
Report fields and validation
CSV headers are normalized to snake case: for example, app-id becomes app_id. The report includes date, app_id, media_source, campaign_id, campaign, geo, adset, ad, ad_id, site_id, spend, and currency. IDs remain strings, spend is numeric, and currency values are preserved without conversion. Extract also adds _singular_extract_timestamp.
Every expected column must appear exactly once after normalization. Extra columns are ignored, but each row must have the same number of columns as the header. Each row's date must match the requested day, and spend must be a finite number. Invalid headers, dates, spend values, or row widths fail the run. A header-only report is valid; an empty response without headers is not.