Toad Data Point Import wizard reads some sample records and creates a destination table databas schema based on the deducted data types from this sample. With a small sample this will often be wrong. In fact I think for the most part you should rather trust the original schema at create a destination schema as close to the original as possible.
When a deviant destination schema is created then you should warn the user about this and give the user an option to not deviate.
It should be possible to have an option if you want to base destination schema on sample data or create it as close to the original schema as possible.
It should be possible to declare the number of samples (in %?) to use.
Ref Service Request: 4629950