How Octonity's moderation policy hides hate and spam automatically, why negative comments are deliberately never auto-hidden, and how to layer your own comment rules on top.
Leave a comment section unattended for a week and you can watch it rot in a predictable order. First the crypto spam and the "DM me for collab" bots. Then the abuse — because an unmoderated thread signals that nobody's watching. And then the quiet damage: real customers scroll past your post, see a comment section full of junk, and conclude that the brand behind it isn't paying attention. Comments are the most public part of your presence, and unlike a bad ad, a toxic thread is user-generated, around the clock, and attached to your name.
In short: a comment moderation tool reads every incoming comment and acts before you do — hiding the universally unwanted (hate, abuse, spam) automatically and flagging the judgement calls for a human. In Octonity that's two systems working together: a default moderation policy that runs on everything, and Comment Automation rules you write yourself for anything specific to your brand.
Octonity's moderation policy lives at Analytics → Moderation and runs
automatically on 100% of incoming comments on your Instagram and Facebook
posts. Every comment is scored by our classifier before it ever reaches
your inbox. There are three settings, and the defaults are opinionated on
purpose:
Alongside the classifier there's a simpler keyword layer: a blocklist at
Settings → Moderation blocklist — a plain word list with a severity per
word — for the terms you never want to see regardless of context.
One boundary worth stating plainly: the policy acts on comments, not DMs. Meta's API cannot hide a direct message, so no tool — ours included — can "moderate your DMs" in the hide-it-automatically sense. Anyone claiming otherwise is describing something else.
Here's the part most moderation tools get wrong, and where we made a deliberate product decision: negative and critical comments are never auto-hidden. Not at any threshold, not with any toggle.
"This arrived broken and support ignored me" is not hate and it's not spam — it's a customer with a problem, in public. A tool that hides it does you no favour. Hiding genuine criticism tends to escalate: the customer notices, gets angrier, and takes the complaint somewhere you can't hide it. And Meta penalises the pattern — accounts that systematically suppress critical comments are playing a game the platform is built to detect.
So instead of hiding criticism, Octonity flags it. Critical comments land in your Inbox with a severity badge, so your team sees the worst first. If something flagged genuinely crosses a line, you can hide any flagged item in one click — but that click is yours. The classifier surfaces; a human decides.
This is the honest trade: a moderation tool that promises to make all negativity disappear is selling you a reputation risk with a nice dashboard. One that hides the indefensible and escalates the rest is actually protecting the brand. If your audience writes in several languages, the same flagging logic applies across them — we've covered that in our guide to multilingual AI moderation.
The default policy handles what every brand agrees on. For everything
specific to your brand — product names, campaigns, languages, support
patterns — there's a second system: Comment Automation, at
Inbox → Automation.
Every rule has the same shape: when a comment arrives on one of your Instagram or Facebook posts, if it matches a condition, do one or more actions.
Conditions can match on brand, channel, a specific post, a keyword or phrase, whether the commenter is new or returning, comment length, or language. Actions are independently checkable per rule — one rule can do several things at once:
{commenter_name}
supported)support_case tagThree rules we see work in practice:
support_case. The customer gets an
instant pointer; your support team gets a tagged queue instead of a
scavenger hunt through comment threads.If you've already set up comment-to-DM flows from our Instagram auto-reply guide, this will feel familiar — same event, different intent: auto-replies turn comments into conversations, automation rules keep the comment section worth reading.
The short version:
| Use the default policy for… | Use Comment Automation for… |
|---|---|
| Universally unwanted content: hate, abuse, spam | Anything specific to your brand, product line, campaign, or language |
| "Nobody wants this under any post" | "Under this post / for this keyword / in this language, do X" |
| Set once, tune with one threshold | As many rules as you have patterns |
And they stack. The policy runs on everything as the floor; your rules add brand-specific behaviour on top. A comment can sail past the policy as perfectly civil and still trigger your refund rule.
No — deliberately. Negative and critical comments are never auto-hidden, at any setting. They're flagged in the Inbox with a severity badge, and you decide, one click per item, whether anything crosses the line. Hiding genuine criticism tends to escalate it, and Meta penalises accounts that make a pattern of it — so Octonity won't do it for you automatically.
No. The moderation policy acts on comments on your Instagram and Facebook posts only. Meta's API simply cannot hide a DM, so automatic DM hiding isn't something any compliant tool can offer. DMs still arrive in your inbox for humans to handle.
Yes. Actions in a Comment Automation rule are independently checkable, so
a single rule can hide the comment, reply publicly, DM the commenter, like
the comment, and tag it for human review in any combination. In practice
you'd rarely hide and reply publicly to the same comment — but pairing a
DM with a support_case tag, as in the refund rule, is a common combo.
Adjust MinScoreToHide at Analytics → Moderation. The classifier scores
each comment's severity, and only comments above the threshold are hidden
— the default is 30. Raise the number to be more lenient (fewer comments
hidden), lower it to be stricter. For individual words you never want to
see, add them to the blocklist at Settings → Moderation blocklist
instead.
AutoHideSpam ships off by default because spam is a per-brand judgement
call — promotional comments that are noise for one account are leads for
another. When you're ready, it's a single toggle at
Analytics → Moderation, and the same severity threshold applies.
You can't out-type the bots and you shouldn't be the one deleting slurs at midnight. Turn on the default moderation policy, leave criticism visible where you can answer it well, and write one or two automation rules for the patterns your brand actually sees. The goal isn't a silent comment section — it's one where the real conversations are the ones left standing.
Plan, create, and publish across every channel from one workspace. Free plan available — no credit card.
A practical guide to Instagram auto-replies: how comment-to-DM automation works, keyword matching vs. intent detection, setting up your first rule in Octonity, what Instagram's rules actually permit, and the situations where a human should answer instead.
دليل شامل عن بوت انستقرام: الفرق بين البوت الرسمي المبني على واجهة إنستغرام وبوتات السبام التي تهدد حسابك، أهم استخدامات الرد التلقائي، وخطوات عملية لتشغيل أول بوت لحسابك خلال دقائق.
The safe way to give an agency access to your Facebook Page and Instagram in 2026: Task Access vs Partner access step by step, no passwords shared, no admin handed over — plus a security checklist to run before you approve any request.