As I mentioned above the main difference is if you use the Multi Phse datapump in data transformations. This switches the datapump into row-by-row processing instead of a bulk load.

If you are able to accept the performance hit you can develop a pretty sophisticated data loadng routine that traps data conversion and key violation errors. The rows in error can be redirected to an exception table for later examination.

If you're loading a lot of data then a better approach would be to bulk load the file directly into a staging table that has all the fields defined as varchar. Then you can run validation routines on the data without affecting performance.