Duplicate Rows Across Partitions in Google Ads Transfer to BigQuery

[BamChocolate]

Hi,

I'm using the Google Ads Data Transfer Service to load data into a partitioned BigQuery table (p_ads_Ad_customer_id). According to the documentation, the transfer overwrites the partition for the same date to avoid duplicate rows, even when using a refresh window. However, I'm observing duplicate rows for the same ad_group_ad_ad_id and ad_group_id across multiple partitions.

Observations:

Partition Behavior:

Rows are appended across multiple _PARTITIONTIME values.
Each duplicate row is identical, with no hidden differences in fields or metadata.
Transfer Configuration:

Single Google Ads Transfer at the Manager Account (MCC) level
Scheduled daily at 01:00 UTC.
Refresh window: 30 days.

The data appears to re-export identical rows into new partitions during the refresh window.

Example:

For a given ad_group_ad_ad_id and ad_group_id, I see identical rows in _PARTITIONTIME values ranging from 2024-11-22 to 2024-11-28.

Question:

Is this behavior expected for the Google Ads Data Transfer Service?
Should the service be deduplicating rows during the refresh window to avoid appending identical data?
Is there a recommended way to address this behavior in BigQuery, or is this something that should be handled within the transfer configuration?

Thanks in advance for any insights or suggestions!