youtube bigquery transfer backfill on the same table on the same date, I got different counts when I run it in June comparing with run in Sept

Question

In June, I run a youtube bigquery transfer backfill for the data of 2017-04-10, I got 120,000+ records in total for "asset estimated revenues".

In September, I run the same youtube bigquery transfer backfill for the same date, I got about 98,000 records in total for "asset estimated revenues".

Should youtube bigquery transfer backfill data for 2017-04-10 be the same? Does not matter either I run this backfill in June or September?

When run youtube bigquery transfer backfill for the date 2017-04-10, did Google bigquery return the data they cached for 2017-04-10, in this case, does not matter when I request the bigquery transfer backfill, the data for 2017-04-10, the data should be the same.

Or each time when we run the bigquery transfer, such as for the date 2017-04-10, Google bigquery recalcuate the data that it would use to backfill for 2017-04-10 for us. So due to in June and in Sept. I have different assets under my content owner id. So each time the "re-calculation" is different?

Answer 1

OK. I figured it out.

It is youtube bigquery transfer work in progress.

In June, when we run youtube bigquery transfer backfill for "asset estimated revenue", the backfill included all the records with the revenue = 0.

In September, the same backfill, youtube bigquery transfer excludes the revenue = 0.

The data revenue != 0 is the same!