The sad fact is ODBC & Excel does not give you a database where each field is of a defined type.
The Excel ODBC driver scans the first 8 rows of each column and makes a "guess" at the data type so if you have a mixture in there it may cause a few glitches or become confused with what to do with a blank field.
I think you're on the right track with trying to manage what goes into the cell values (just bear in mind the recordset field types will be based on what's in those first 8 rows)