Enable spmassassin to learn from what you mark as spam

Is there a way that we can check whether this is working? I have, for example, messages from Square Receipts after a sale is made on my website that are continually marked as Spam (SPF headers fail) despite me repeatedly moving them to my inbox and/or another non-Spam folder.

Are there logs for sa-learn that I can observe? Do I need to add it to a cron job like @swauzie mentioned above?

It appears that @daveteu runs sa-learn manually. Is that necessary?