Auburn,
The first thing you may wish to do is look at what you must have. In our case, we needed three things: 1) data to support our MRP system; 2) a record of method for setting the standard; and, 3) a record of results.
The first we needed to conduct business. The second was to c.y.a. in case our methods were challenged. The third was to allow the development of a forecasting model to be used by engineering and marketing (How often can you please BOTH of these areas?). The first two being corporate required residence where many could access them while the third, being departmental, needed only to reside where those in our department could get at it.
Old data obtained using good methods is still good as long as there is decent information on the details under which it was obtained. Consequently, if no one knows who, when, why, or how, you probably want to replace these first because they are indefensible.
On the other hand, even poor results to which you have the answers to those questions has more value.
I suggest you get a copy of Maynard's IE handbook if you don't already have one. There is a lot of info on various facets of your problem.
One last item. You might want to enlist someone familiar with developing relational databases. While they may not know much about doing your job, they should know the questions to ask of a software provider to help you get what you pay for.
Okay, off soapbox.
Griffy