I am not sure you can do what you want using SQL. For one thing, I think you should have an extra [logical] field in both tables which is set to false for all records in both tables initially. Next, start a full table scan on one table then switch to the second table to start field comparisons among matching POs. Where you find matching records, set the flag fields in both to true.

The records left with a value of False will be the ones you need to reconcile. Because you have the potential of a many-to-many relationship, doing this relationally is of little help. Where it will help is when 1 record exists in one table but not the other.

The other problem is you do not have any way of uniquely identifying records that may have been matched up the previous time but do not exactly match this time. If at all possible, you are going to be better off enforcing a unique [key] field in the Excel source that you can carry over to the History table. If you do that, and get your History table synced with the Excell data, updates will be a breeze.