I'm guessing the rep's id is not unique from within the set, and your final result is not granular enough to reduce the number?
For clarity, a blend is different from a join, a join will join 1:1, 1:many or many:many depending on the join predicates, but generally as an analyst, you will join sufficiently enough to create a 1:1 join (if this is required), where blending on the other hand, will pre-aggregate the secondary source to meet the partition schema of your viz, and then join, so if your viz is set to yearly and you are choosing a non-unique field, Tableau will sum the value of this field before adding it to the viz, only by drilling to a deeper granularity, will Tableau reduce the number.
Without seeing your data or workbook (even if the data is sanitised), we can only give advice, my suggestion is to ensure the fields are truly unique, at least as far as the numbers you want to show, and that you are setting the relationships to the correct level of granularity, which is to say, ensuring only those paperclips in the secondary source are selected to match the desired granularity.
The rep's IDs are unique and I've verified that they both match on the primary and secondary data sources.
I'm using a date field as a filter and noticed that it doesn't show up on the secondary source. For example, in my viz, Calls is secondary and Emails is primary. On verifying the data, I've created another worksheet where Calls is the primary source and Emails is the secondary. I verified that the rep IDs match. However, I noticed that when I pull the month and year that I'm also using as a filter, they are reported. When I look back at the worksheet where Calls is secondary and Emails is primary, the month year pulled from the Calls data source is no longer listed. See below:
Calls Secondary, Emails Primary
Calls Primary, Emails Secondary