Published: April 24, 2026 Updated: April 24, 2026 By Anony Botter Team

Anonymous Poll Approval in Slack: Moderation, Tone Coach, and Banned-Topic Blocks for Polls

Polls now get the full message treatment — approval queue, tone coach rewrites, and banned-topic auto-declines. No more silent drops when a poll violates policy.


Anonymous poll approval, tone coach, and banned topics in Slack

📖 What You'll Learn

  • Why polls were a moderation blind spot — and what the new parity means in practice
  • How Tone Coach evaluates poll questions and answer options pre-ack
  • The banned-topic auto-decline flow — what replaces the old silent-drop behavior
  • How to write anonymous poll questions that don't get auto-declined or tone-flagged
  • Five real poll-moderation scenarios, with mod and author perspectives

Anonymous polls used to be the loophole. A workspace could spend a quarter carefully tuning its approval queue, tone-coach settings, and banned-topic policy for messages — and then an employee would post an anonymous poll with "Is the CEO incompetent?" as option A, and none of those guardrails applied. Polls shipped as a separate flow, and the moderation stack was built around messages. That's fixed now.

As of this release, anonymous polls in Slack inherit the complete moderation pipeline that messages already had: the pre-ack Tone Coach rewrite modal, the post-ack banned-topic check (now an explicit auto-decline with a DM to the author and a mod heads-up — no more silent drops), and the three-button approval-queue workflow with optional AI Mod Summary. One flow for messages. One flow for polls. One bar for both.

The Problem: Polls Were a Parallel Universe

In older versions of Anony Botter, messages and polls shared the anonymity model but diverged on moderation. Messages had the approval queue. Messages had Tone Coach. Messages had the banned-topic check. Polls had none of that — when you submitted an anonymous poll, it either posted or, if it hit a banned topic, silently vanished. No DM to the author. No heads-up to mods. The author saw their modal close and assumed the poll posted. The channel never got it. The author never knew why.

The silent-drop behavior had an obvious problem: authors couldn't fix what they didn't know was broken. Someone would post a banned-topic poll, notice it never appeared, assume a bug, and repost the same poll through a different channel. The block was technically effective but operationally pointless. Parity with the message flow — which DMs the author on a block — was overdue.

📊 What Changed at a Glance

  • Approval queue: polls now show up alongside messages with the same three buttons.
  • Tone Coach: poll questions and options are checked pre-ack, with rewrites available in a modal.
  • Banned Topics: policy hits auto-decline the poll explicitly — DM to author, heads-up to mods, no identity leaked.
  • No more silent drops. Every block is recorded, every author is told, every mod has a paper trail.

The Three Moderation Gates, Applied to Polls

A poll submission now flows through three checkpoints — the same three a message goes through, in the same order.

Gate 1: Tone Coach (Pre-Ack)

When you submit a poll, Tone Coach runs on the question text and the answer options before the bot acknowledges the submission. If it detects hostility — in the question, in any option, or both — you get the familiar rewrite modal with a softer version of whichever part triggered the check. You can Send rewrite or Edit, exactly like the message flow. The 1.8-second timeout and fail-open behavior apply: if the check doesn't finish in time, the poll posts exactly as written.

Example: submitted poll question "Who's the worst-performing exec?" with options "CEO", "CTO", "CFO". Tone Coach modal:

Your poll:

Question: "Who's the worst-performing exec?"
Options: CEO / CTO / CFO

Suggested rewrite:

Question: "Which area of executive communication could be clearer for the team?"
Options: Strategy / Engineering / Finance

Gate 2: Banned Topics (Post-Ack, Auto-Decline)

After the bot acknowledges submission but before the poll reaches the channel, the banned-topic check runs against the workspace policy paragraph. This is the gate that changed most. The old behavior was a silent drop — the poll never posted, no one was notified. The new behavior:

  1. The poll is explicitly auto-declined.
  2. The author gets a DM with the original question and options quoted back, plus the policy reason.
  3. If an approval channel is configured, the mods see an auto-decline heads-up card with the blocked poll content and the reason — and no author identity.
  4. The author can rephrase and resubmit. No strike, no record tied to them.

💡 Auto-decline vs silent drop: The practical difference is trust. Silent drops left authors confused and mods blind. Auto-decline makes the block visible to the people who need to know (the author, to fix; the mods, to monitor) while keeping the content out of the target channel and anonymity fully intact.

Gate 3: Approval Queue (Human Review)

If your workspace has require-approval enabled on the target channel and the poll has passed Tone Coach and Banned Topics, it lands in the approval queue as a card. The mod sees the poll question, the options, and — if AI Mod Summary is on — a one-line summary and a suggested action. Three buttons:

Approve

Poll posts to the target channel as anonymous.

Decline

Poll does not post. Author gets a decline DM with no editorial feedback.

Decline with feedback

Poll does not post. Mod types a short explanation; author gets a DM containing their original poll and the mod's feedback so they can rephrase and resubmit.

Writing Anonymous Polls That Don't Get Declined (Employee Lens)

Most declined polls are easy to rewrite. The question, the options, or both carry the same patterns that get messages declined — and the fixes are the same. If your poll auto-declines or gets a Decline-with-feedback, the issue is almost always one of three things.

Issue 1: Named Individuals in Options

❌ Bounces

Q: "Who should we remove from leadership?"
Options: Sarah / Mike / Priya

✅ Lands

Q: "Which leadership area would most benefit from clearer communication?"
Options: Strategy / Engineering / People / Operations

Issue 2: Loaded or Leading Question Framing

❌ Bounces / Tone-Flagged

Q: "How badly did management mess up the pricing change?"
Options: Very badly / Extremely badly / Catastrophically

✅ Lands

Q: "How well was the recent pricing change communicated to the team?"
Options: Very well / Well / Neutral / Poorly / Very poorly

Issue 3: Banned-Topic Content in the Question or Options

❌ Auto-Declines

Q: "What's a fair salary for a senior engineer at this company?"
Options: $130K / $150K / $170K / $190K

✅ Lands

Q: "How confident are you that our senior-engineer pay band is market-competitive?"
Options: Very confident / Somewhat / Neutral / Somewhat doubtful / Not confident

💡 The poll-writing principle: Anonymous polls work best as sentiment or process questions, not verdict-assigning ones. "How well is X going?" will always survive moderation better than "Who should we blame for X?" — because the first produces a signal you can act on, and the second produces a liability.

Admin Setup for Poll Moderation

If you already have message moderation configured, polls inherit the exact same settings. There's no separate poll-moderation configuration — the goal was parity, not a second configuration surface to keep in sync.

  • Banned-topic policy paragraph: the same paragraph governs both messages and polls.
  • Approval channel: the same channel shows both message cards and poll cards, distinguishable by their content layout.
  • Tone Coach toggle: applies globally to both surface types.
  • AI Mod Summary: if enabled, summarizes both message and poll cards in the approval queue.

If you're new to moderation configuration, see the Slack Admin Guide for step-by-step setup.

Five Real Scenarios

1. The Sentiment Poll That Landed

A People Ops lead wants to gauge reaction to a new remote-work policy. They submit an anonymous poll: "How well did the recent WFH policy change land with you?" with a five-option Likert scale. Tone Coach finds nothing to flag; Banned Topics finds nothing to block; the approval queue approves on "suggested: Approve." The poll posts in six seconds. Thirty-two responses come in over the next hour. Useful signal, zero friction.

2. The Named-Exec Poll That Got Rewritten

An engineer, frustrated with a specific exec, submits "Whose strategy is failing us?" with three names as options. Tone Coach fires pre-ack with a rewrite modal suggesting "Which area of company strategy needs the most attention?" with area names instead of people. The engineer hits Send rewrite. The poll posts anonymously, gets genuine engagement from leadership, and the area named "roadmap clarity" becomes the subject of a real offsite agenda item.

3. The Salary Poll That Auto-Declined

Someone submits a poll asking about specific compensation figures. The banned-topic check catches it; the poll auto-declines. The author gets a DM within two seconds with their original poll quoted back and the reason ("specific compensation figures — not permitted under this workspace's pay-discussion policy"). The mod team sees the auto-decline in the approval channel with the blocked content — no author info. The author rephrases to a band-competitiveness question and resubmits successfully.

4. The Mod Declines with Feedback

A poll makes it through both AI gates but the mod on duty reads the question and thinks the framing will polarize the team. She hits Decline-with-feedback and types: "Can we rephrase to focus on what support the team needs rather than who's accountable?" The author gets the feedback DM, revises, and resubmits a cleaner version two minutes later. It posts.

5. The Silent-Drop Regression Test

A workspace admin, on purpose, submits a clearly banned-topic poll to verify the new auto-decline flow works. They get the DM with the original + reason within seconds; the approval channel shows the auto-decline card with blocked text and reason; no author identity appears anywhere. The old silent-drop behavior is confirmed gone. The admin documents the new behavior in their internal runbook.

Limitations

  • Tone Coach can't rewrite across meaning. A question that's fundamentally a witch-hunt can't be rewritten into a fair question — the modal offers a reframe that changes intent. The author has to accept the reframe or write something different.
  • Banned Topics runs on the whole poll. If any option trips the policy, the whole poll auto-declines. You can't have a ten-option poll where nine are clean and one is a policy hit that gets silently dropped.
  • Approval-queue latency applies to polls. If your queue has a backlog, polls wait in it like messages do. This is the reason AI Mod Summary exists — to keep queues flowing.
  • Poll-specific features like option limits still apply. Up to five options per poll; same as before.

Frequently Asked Questions

Do anonymous polls now go through the approval queue?

Yes. If your workspace requires approval, polls flow through the same queue as messages with the same three buttons: Approve, Decline, and Decline-with-feedback.

Does Tone Coach check poll questions too?

Yes. The poll question and the answer options are both evaluated pre-ack. If hostility is detected, a rewrite modal shows you softer alternatives for the question, the options, or both.

What happens when a poll hits a banned topic?

The poll is auto-declined — explicitly declined, not silently dropped like in older versions. The author gets a DM with the original quoted back and the reason, exactly like a banned-topic message. If an approval channel is configured, mods see the auto-decline notice with the blocked text and reason, but never the author identity.

What was the old behavior on policy hits?

Before this release, polls that violated banned-topic policy were silently dropped — no DM to the author, no record for admins. The new auto-decline flow replaces the silent drop with an explicit decline and a full paper trail.

Can I decline a poll with feedback?

Yes. The Decline-with-feedback button works the same way on poll cards as on message cards — the mod types a short explanation, and the author gets a DM with the decline reason plus their original question and options, so they can revise and resubmit.

Does AI Mod Summary run on poll cards?

Yes. If AI Mod Summary is enabled on your approval channel, poll cards get the same one-line summary + suggested action treatment as message cards — for example, "AI: Sentiment poll on remote-work policy; suggested action: Approve."

Related Reading

One Moderation Stack. Messages and Polls. Zero Silent Drops.

Install Anony Botter to get approval queues, Tone Coach rewrites, banned-topic auto-declines, and AI Mod Summaries — applied identically to anonymous messages and anonymous polls.

✅ Same Buttons

Approve / Decline / With feedback

✍️ Rewrite Modal

Tone-coach poll questions too

🚫 Auto-Decline

Policy hits get an explicit DM

🆓 Free Tier

Polls free forever on any plan

Conclusion: Parity Is the Feature

Every bit of moderation work a team invests — the policy paragraph, the approval channel, the mod runbook — is infra. It scales when it applies everywhere and fails when it applies only to some surfaces. This release closes the gap polls left: the same three gates now run on every anonymous submission, the silent-drop behavior is gone, and authors whose polls don't post finally know why. For admins, moderation is one configuration. For employees, it's one bar for everything they submit. That's the feature.