Hi Raffaella,
Before we proceed further here - the first thing I would try is to create some simple tabular Insight and put there the attributes/facts which they claim that are missing as a single rows or columns without any further metrics or aggregations.
That way, you will see the raw column and it’s values and you can compare it with the columns from the source data. If you do that, do you truly see some data missing compared to the redshift data source?
This way, we can confirm/exclude that option. Can you please give it a try and let me know about the result?