I'm wondering how others handle the situation of bringing in a wide (lots of columns) source table with very limited documentation. We may know that we need the table, hopefully identify the key, and probably have been told of a few columns that have data needed.
That may leave 50-100 columns that we really don't know much about. It would be easy to simply drop them on the satellite and move on - addressing them later if we find value. However if those columns are actually FKs then we missed building links up front and we're not representing the data correctly. It could take a lot of analysis time to go through what documentation we have and/or look at the values in the columns and the column name and try to guess what the FKs are.
Is it worth taking the time up front to do this - or do you in this case simply just drop all the columns onto the satellite and deal with links at a later time?