We have a two stage application process, with a very simply Stage 1 and a more complex Stage 2.  This means that applicants are frequently adding new contacts on their Stage 2.  The way the system current works, while you can search for a contact match on the Stage 1 duplicate check, on the Stage 2 duplicate check you are forced to take what the system determines (i.e. if it doesn't find a match, and lists it as a new contact, even if you know they are in the system, you can't link to the correct contact).  And given that more than 50% of the time the system does not correctly link to an existing contact, this means we are creating many duplicate contact records that then need to be cleaned up afterwards, which is incredibly frustrating.  I am told this is by design, but it seems to me to be poor design.