For Server Owners
A practical guide to moderation.
How to set up roles, filters, and logs so problems stay small. Ten minutes of setup saves hours later, and this page walks you through all of it: the first-week checklist, the everyday tools, the escalation ladder, and the hard calls.
Moderation is setup, not firefighting.
The best moderation decisions are the ones you make before anything goes wrong.
When people picture moderation, they picture the dramatic moment: the spammer, the argument, the ban. In a well-run community those moments are rare, short, and boring, because the real work happened earlier. Channel permissions were set so the announcement channel could not be flooded. Filters were on, so the spam never reached human eyes. The staff had already agreed on what earns a warning and what earns a ban, so nobody had to improvise under pressure.
That is the philosophy behind this guide: moderation is mostly structure, decided in advance. Structure is cheap to build when your server is quiet and expensive to retrofit during a crisis. If you are setting up a new server, work through the five steps below before you invite anyone. If your server is already busy, the same steps still apply; you will just be applying them with more information about who your members are and where friction actually shows up.
This page is the how. The moderation tools page is the what, with a feature-by-feature tour of everything in the toolbox. The owner handbook covers the broader job of running a community, of which moderation is one part. And the roles and permissions page goes deeper on the permission system this whole guide is built on.
Two rulebooks, one floor
Every community on Vronify.chat operates under two sets of rules. The platform-wide community guidelines are the floor: they apply to every server, and no server rule can go below them. Your own server rules are yours to write, and they can be as strict as your community needs. A knitting server and a competitive gaming server will tolerate very different tones, and that is fine. What matters is that your rules are written down, visible, and short enough that people actually read them. Three to seven clear rules beat a twenty-point legal document every time.
Who does this work
On a small server, you are the moderation team. That is normal and it works. As the server grows, you will hand pieces of the job to people you trust, and the permission system is how you hand over exactly the pieces you mean to and nothing more. Everything below assumes a team of one to start with, and scales from there.
The first week: five moves.
Do these in order. None of them takes more than a few minutes.
Define your roles
At minimum: owner, moderator, member. Write one line per role about what it is for, and resist inventing ten ranks on day one. Extra roles feel like structure, but every permission-bearing role is another thing to audit later. You can always add more when growth demands it; removing a role people feel entitled to is much harder.
Set channel permissions
Lock announcements to staff, keep general open, and give every channel an obvious purpose. Per-channel overrides handle the exceptions. A rules channel should be readable by everyone and writable by no one. A staff channel should be invisible to members entirely, so your team can discuss incidents candidly without an audience.
Turn on filters and slow mode
Let automation absorb the spam before a human sees it. Slow mode on busy channels keeps arguments from snowballing. The point of automated filters is not to catch everything; it is to catch the obvious, high-volume junk so your moderators spend their attention on the judgment calls only a human can make.
Agree on escalation
Decide as a team what earns a warning, a timeout, and a ban, before the first incident. Consistency is what members experience as fairness. Write the ladder down in your staff channel, keep it to a paragraph, and update it when a real case shows you a gap. An escalation policy invented mid-argument convinces no one.
Review the audit log weekly
Ten minutes a week keeps everyone honest and catches patterns early. It is also the record that settles any dispute. If a member claims a moderator deleted their message unfairly, you do not have to reconstruct events from memory; the audit log shows who did what and when, and the conversation becomes short.
Roles and permissions, done properly.
The permission system is your moderation policy, written in a form the server can enforce.
Start with three roles
Owner, moderator, member. The owner holds the keys: server settings, role management, and the final say. Moderators keep conversations healthy: they need to manage messages, apply timeouts, and remove people who will not stop. Members talk. That is the entire org chart for most servers under a few hundred people, and plenty of larger ones never need more.
Give moderators what the job needs, and no more
The principle is least privilege: every permission a role carries is something that can be misused, by accident or on purpose, so a role should carry only what its job requires. A moderator does not need to edit server settings, rename channels, or manage other people's roles to do moderation. Keeping those out of the moderator role is not distrust; it is what makes the role safe to hand out. When a moderator account is compromised or a moderator has a bad day, the blast radius is limited by design.
Defaults handle the norm, overrides handle the exceptions
Set your server-wide defaults for the common case: members can read and write in most places. Then use per-channel overrides for the exceptions: announcements are staff-post-only, rules are read-only, the staff channel is staff-only. Working this way keeps your permission setup legible. If you find yourself writing overrides on most channels, your defaults are wrong; fix the default and delete the overrides.
Roles as identity versus roles as power
Communities love roles as identity: region roles, interest roles, notification tiers. Those are healthy and you can have as many as you like, because they carry no dangerous permissions. Keep a clear mental line between decorative roles and permission-bearing roles, and audit the second group. When you review your setup, the question is not "how many roles do we have" but "how many roles can delete a message or remove a member, and does everyone holding one still need it".
When you add a new staff rank
Growth sometimes justifies a helper tier: trusted regulars who can apply slow mode or handle timeouts in their time zone without holding full moderator power. Add it when your existing team is visibly stretched across hours of the day, not before. Every new rank needs its own one-line job description, its own spot on the escalation ladder, and its own entry in your weekly audit habit. The full mechanics of role hierarchy and channel overrides are on the roles and permissions page, and if you are starting from zero, the create a server guide walks through initial setup end to end.
The everyday toolbox.
Six tools you will actually use, and what each one is really for.
Slow mode
A per-channel cooldown between messages. It is your first response to almost everything: spam waves, arguments picking up speed, event channels moving too fast to read. Slowing the room is gentler than silencing anyone, and it often ends a fight by itself because nobody can fire back instantly.
Automated filters
Filters catch the obvious junk before a human sees it, around the clock. Tune them to your community: strict in a server for kids or a classroom, lighter where adults are talking among friends. Review what they catch occasionally; a filter that silently eats legitimate messages erodes trust.
Audit log
Every staff action, recorded: who deleted what, who changed which permission, when. It protects members from moderator overreach and protects moderators from false accusations, in both directions at once. Make the weekly review a habit; patterns show up in the log long before they show up in public. Details on the audit log page.
Pinned messages
Pin the rules, the current event, the answer to the question everyone asks. A good set of pins prevents a surprising amount of moderation work, because most "violations" in a healthy server are just people who never saw the rule. Prune pins when they go stale; a pin wall nobody reads is the same as no pins.
Threads and replies
When a tangent takes over a channel, move it into a thread instead of killing it. The main conversation gets its channel back, the tangent gets room to breathe, and nobody was told to shut up. Threads are the moderation tool that does not feel like moderation.
Categories and structure
Channel structure is passive moderation. When every channel has one obvious purpose, off-topic posting mostly corrects itself, because members can see where things belong. A server with three clear channels needs less policing than a server with fifteen vague ones.
Tools that hold up at scale.
Moderation advice is worthless if the platform buckles under a big community. Ours does not.
The incident playbook.
Six situations every moderator eventually meets, the reflex response, and what actually works.
Every incident is different, but the shape of a good response is the same: slow the room down, act on the behavior, keep it as private as it can be, and write it down.
Moderate the behavior, not the person.
Our ban model backs you up on that.
On Vronify.chat, banning someone removes them from your community without confiscating their data. A ban stops participation, nothing more: the banned member keeps their account, every message they ever sent, and the DMs they received, unless the sender removed them. They can export their data any time, and if they want to leave completely, they can delete their account at any time, which anonymizes it. The full picture is on the account ownership page.
That distinction changes how banning feels, on both sides. On platforms where a ban locks someone out of their own history, moderators hesitate, because the punishment is wildly out of proportion to most offenses. Here, a ban ends the relationship between a person and your community; it does not end the relationship between a person and their own words. That lets you act firmly without feeling like a judge handing down a life sentence, and it lets you act sooner, before a bad situation gets worse, because the cost of the decision is proportionate.
It also changes the conversation afterward. A banned member has nothing to beg you for: their data was never yours to withhold. Appeals, when you choose to hear them, can be about behavior and whether it will change, which is the only question that ever mattered. The full toolbox is on the moderation tools page.
What a ban does
- Ends participation in your server, immediately
- Applies to your server, on your team's judgment
- Leaves a record your staff can review later
- Lets you protect your members without destroying anyone's history
What a ban never does
- Confiscate the banned member's account
- Take away the messages they sent, which remain theirs
- Touch their DMs, which stay private
- Hold their data hostage: export and account deletion stay available to them
Hard calls and edge cases.
The situations no checklist fully covers, and honest answers about where the limits are.
Harassment that moves to DMs
Direct messages on Vronify.chat are private. We don't read them, and no moderation tool reaches them, so you cannot either. That is a deliberate design choice we explain on the privacy page, and we will not pretend it is free: it means DM harassment cannot be moderated the way channel behavior can. What you can do is real, though. Help the target block the sender. Act on anything the harasser does in your channels, where you have full authority. And treat a pattern of independent reports about the same person as evidence in its own right; you do not need to read the messages to notice that three unrelated members are describing the same behavior.
Why public channels are different
Community channels are visible to their community and its moderators, and that is also deliberate: it is exactly what makes moderation possible. Moderators can see, remove, and act on what happens in public spaces, which is what keeps those spaces safe for everyone in them. The line is clean and worth explaining to your members: private conversations are private; shared spaces are governed by the people who run them.
Moderating your friends
The hardest moderation action you will ever take is against someone you like. This is where the ladder you wrote in step four earns its keep: when the rules and consequences were written down before the incident, applying them to a friend is following policy, not betraying a friendship. The moment you visibly bend rules for friends, every past and future decision you make becomes suspect. Most communities do not die from trolls; they die from insiders who were never held to the same standard.
When the moderator is the problem
It happens to good teams. Least privilege limits the damage a moderator can do, and the audit log means nothing they do is invisible. Look at patterns, not single actions; anyone can misjudge one case. Talk privately first. If the behavior repeats after a clear conversation, remove the role and say so plainly to your team. Protecting a moderator at your members' expense costs more trust than any single incident ever will.
Burnout is a moderation problem
A tired moderator makes bad calls, and one person covering every hour of every day will burn out on any platform. Keep the job small enough to be sustainable: let automation absorb the volume, keep the escalation ladder simple enough that decisions are fast, and add helpers across time zones before your existing team is drowning rather than after. A mostly bored moderation team is not a sign you over-hired; it is what success looks like.
What automation adds next
We are building a bot engine directly into the platform, and it is coming soon. For moderators, the interesting pieces are the message log, which will record edits and deletions to a private staff channel; tickets, which will give reports and appeals their own space instead of a busy channel; and welcome cards, which put your rules in front of every new member the moment they arrive. It is not live yet, so build your setup on the tools above, and watch the changelog for the release. Nothing about your setup will need to be redone; the engine arrives as one switch in your server settings.
Questions moderators actually ask.
How many moderators do I need?
There is no magic ratio. Think in coverage, not headcount: you want someone reachable during the hours your community is actually active, which usually matters more than raw numbers. A focused server can run for a long time on one or two people plus good filters and slow mode. Add people when incidents are going unhandled for hours, not because the member count crossed a round number.
Do I really need my own server rules?
Yes, even three lines. The platform community guidelines are the floor for every server, but they cannot know your community's culture: what is friendly teasing in one server is a bannable insult in another. Short rules, pinned where everyone sees them, do two jobs at once: they prevent most violations, and they give your warnings something specific to point at.
Can I moderate direct messages?
No. DMs are private, so no moderation tool reaches them and server staff cannot read them; as a matter of policy, Vronify does not read them either. That is a hard boundary, not a settings option. What you can do: help members block senders, act on behavior in your channels, and take repeated independent reports about one person seriously as a pattern.
What happens to a banned member's account and messages?
A ban only stops participation in your server. The member keeps their account, every message they ever sent, and the DMs they received, unless the sender removed them. They can export their data any time, and they can delete their account at any time, which anonymizes it. The reasoning is laid out on the account ownership page.
Where can I see what my moderators have been doing?
The audit log records staff actions: deletions, permission changes, removals, with who and when attached. Pair it with server analytics to see the health of the community those actions are protecting. A ten-minute weekly review of both is the cheapest oversight system you will ever run.
Is there automated moderation beyond filters?
Automated filters and slow mode are available today. The built-in bot engine, coming soon, adds a message log for edits and deletions, tickets for reports and appeals, welcome cards, custom commands, and more, all running on our infrastructure with nothing for you to host. It is not live yet; the release will be announced in the changelog.
What if a situation is bigger than my team can handle?
Stabilize first with the tools you have: restrict posting, apply slow mode, remove the accounts involved. Then contact support. Anything that violates the platform community guidelines is our problem as well as yours, and we would rather hear about a serious situation early than late.
Your community. Your rules. Your data.
Create a server, invite your people, and see what chat feels like when your account actually belongs to you.
Free to start. Live in under a minute. No install required.