When you import Excel, using the old method, you get to define the data types.
This is clearly wrong. In a typical Excel table itself there are no column-related uniform data types - in an "intelligent table" as a later imitation of a database table in preparation for Power Query (internally a modification of SQL Server) there are.
With access via SQL / TransferSpreadsheet to a typical Excel table, Jet independently interprets the data types of the columns according to content, so you can't intervene yourself.
By default, this interpretation evaluates the first eight rows of data (registry key TypeGuessRows).
With this data type interpretation you are powerless and run into an error (created error import table) if, for example: there are numerical values in the first lines and an alphanumeric expression, i.e. text, much deeper.
But this was about importing from a text file, and the old method is much more flexible. This means you can integrate a link to the text file and import specification directly into a query, and with a query itself you can do a lot more than just blindly import everything as it is.