Hi Nathe,
I am referring to this from the guide:
Selective mode offers intermediate protection between Never (Wildcard Only) and Add All Entities. Selective mode will suggest the addition of explicit entities to the policy if their attributes are higher ( or different) from attribute values specified in the wildcard.
This provides application owners with the flexibility to expand a policy if there are irregular entities that need precise protection measurements. In other words, Selective mode is suitable for applications containing entities which use similar or identical attributes.
But if some the entities need special handling, the policy can be expanded to include exceptional explicit entities just for those outliers. New entities that are created from accepting Selective based learning suggestions are created with their Staging checkbox enabled, so they can continue learning attributes that are specific only for this new inexplicit entity.