Kinda in the same line of this question, exactly how much traffic comes to us from other sources? This will give us a very specific view of how can we improve our stop-gap before users ask questions without the "minimum research effort" that the SE model ask for any of us.