You need to make sure the data partition is configured correctly. And make sure the partition key accurately segments your data without omitting relevant data.

One of the things I watch closely in Domo ETL processing is duplicate records affecting joins. Number of rows can easily go whack. Normally expanding not loss. With less records I would recheck things like filters and formulas. And possibly calculations affected by NULLs. If values come out as null, the calculations could vacate. Review the use of COALESCE.