The match process depends on the match tokens in the match key table being current. If match tokens need to be updated (for example, if records have been added or updated during the load process), the match process automatically runs the tokenize process at the start of a match job. To expedite the match process, it is recommended that you run the tokenize process separately—before the match process—either by:

manually executing the Generate Match Tokens job, or
configuring the tokenize process to run automatically after the completion of the load process