Learn what your spam looks like
Collect a week of spam and note the senders, subjects and links. Patterns such as the same text from many addresses are easy to catch.
Filter the obvious cases only
Block exact repeats and known bad senders first. Leave borderline messages in the inbox until you are sure of the pattern.
Keep a place to recover mistakes
Send filtered mail to a spam view rather than deleting it, and check it once a week so a real customer is never lost.
Reduce bot traffic at the source
Add a simple check to any contact form, such as a hidden field or a rate limit, so bots stop before they create tickets.
Keep spam out of your metrics
Mark spam so it does not count towards volume or response time, otherwise your reports will look worse than your service is.
Review the rules every month
Spammers change tactics. A short monthly check keeps the filters useful without growing into a maze.