On the duplicate step when we change the matching rules I wonder if two records are empty this is going to be considered an exact match? Is there a way of not considering an empty field?
Data Studio generates Blocking Keys based on record values to identify candidate pairs of potential duplicates, and these pairs are processed through the scoring Rules. Rules & Blocking Keys.
As a result, Blocking Keys are only generated for populated values, meaning records containing mostly blank fields are unlikely to be scored or matched. This is all completely configurable though, so if certain fields are known to commonly be blank, you can choose to define keys that mostly don't use those fields.
Similarly, Rules can be configured to allow matching even when one or both fields in the record pair are blank. See for example the OnePopulated and NonePopulated comparators.
Find Duplicates config can be tricky. If you have specific scenarios you need to match (or not match) let us know!