Rules that read every post so your moderators do not have to
A single human reviewing everything works at small scale and collapses at large scale. Advanced Moderation runs your rules against every post and reply the moment it is submitted, so the obvious cases are handled before anyone opens the queue.
You get five ways to describe bad content. A keyword rule matches an exact word or a comma-separated list, so one rule can cover “buy now, click here, limited offer”. A regex rule runs a full pattern for the things a word list cannot catch, like phone-number formats or obfuscated spam. A link-limit rule triggers when a post carries more than a set number of links, which catches the classic five-to-ten-link spam post while leaving a genuine “here are some resources” reply alone. A new-user rule catches posts from authors below a trust level you set. A spam-score rule adds up points: two per link, three per matching keyword rule, five for an account less than a day old, and three for a member with no posts or replies yet. Because it is a points heuristic rather than a percentage, start with a low threshold around 10.
Four actions, least restrictive first
Matching content is only half the rule. The other half is what happens next, and you set that per rule. Flag lets the content publish but drops it into the moderation queue for a second look. Hold keeps it pending and invisible until a moderator approves it. Block rejects the post outright and shows the member an error. Spam hides it immediately.
The right default is the lightest action that solves the problem: Flag for borderline content, Hold for likely-bad content, and Block or Spam only for the clearly harmful. Because Advanced Moderation sits on top of the free trust-level system, which already rate-limits Trust Level 0 accounts, these rules are a content layer, not your only line of defense.
Scope and hit counts to keep it honest
Rules are global or space-scoped. Global rules enforce site-wide policy such as prohibited words or competitor spam. Space-scoped rules match one space’s norms, so a support space can block every link to stop phishing while another space allows them freely.
Every rule carries a hit count, and that number is how you tune. A rule that fires constantly and sends everything to spam is catching legitimate posts, and the count makes that obvious so you can loosen the threshold instead of quietly losing real members’ content.