im guessing you mean order deliveries table original and new record count after calculating the date difference dont match - is that correct?
the load distinct and left join should guarantee that there are no duplications.
can you share the QVW even with a reduced set where you observed records not matching? you can do this by grouping by factory ID and comparing counts before and after. select the factroty IDs where the counts dont match and reduce data set keeping possible values.
it would be helpful if you can provide the original albeit reduced data set prior to the calculation of the date diff similar to what you did when you added the last QVW