Load your Excel file into your job using a preset schema, then test using a tMap to see if all expected columns are populated. Load the results into a tHashOutput. If any column is missing (or holding the incorrect type), either fail the job with a message (using a tDie) or simply do not progress to the next SubJob. The next SubJob will take your data added to the tHashOutput and carry on the processing using a tHashInput to read it back in.