The reliable order is field cleanup first, schema mapping second, barcode helper setup third. If the source value is not stable, the generated barcode will not be stable either. Start by making sure the PDF field names match the CSV headers, database columns, or JSON keys that will drive the workflow.
For QR and 1D barcode helpers, choose one mapped value. For PDF417, configure the labeled classes that should be included in the scan text. Keep class names consistent with the downstream scanner or import process so the code output is predictable.