Working with BetterReviews in Slack

Last updated: April 26, 2026

When you install BetterReviews, you get a private channel inside our Slack workspace where the BetterReviews team and our @BetterReviews bot show up. You can approve and reject reviews, see stats, search by topic, and ask us anything — all in a single channel.

:shield: BetterReviews will only ever message you in your dedicated channel from the verified BetterReviews app. Any other Slack app or DM claiming to be BetterReviews is phishing — please report it to your channel.

What you get

  • A private channel in the BetterReviews Slack workspace, just for your store.
  • The bot is in the channel and responds to @BetterReviews <command> mentions.
  • The BetterReviews support team is in the channel too — you can @ us directly when you have questions.
  • Notifications when high-marketing-potential reviews come in, when a review needs support follow-up, or when something gets auto-flagged for moderation.

How setup happens

When you install BetterReviews from the Shopify App Store, the channel is created automatically. Today we onboard each merchant by hand — your channel and welcome message are set up by a real person on the BetterReviews team, typically within a business day of install. You’ll be invited as a single-channel guest, which means you have access only to your own channel — not the rest of our workspace.

You can invite teammates as guests too. They’ll be able to see and use the bot, but with limits you control (see Roles below).

Talking to the bot

Mention the bot with @BetterReviews followed by a command. The bot replies in a thread so the channel stays readable.

@BetterReviews stats
@BetterReviews show me marketing-ready reviews from this week
@BetterReviews approve REV-1234

You can use exact commands (listed below) or just ask in plain English — the bot will figure out what you mean. If it isn’t sure, it asks you to rephrase or run help for examples, instead of dumping a wall of options.

Thread memory. Inside a thread, the bot remembers what you just asked for about 30 minutes. Follow-ups like “what about 5-star ones?” or “show me #2” pick up where you left off, as long as you @-mention the bot in the same thread.

Commands you can run

@BetterReviews approve <review id>      Approve a pending review
@BetterReviews reject <review id>       Reject a review
@BetterReviews pending                  List reviews waiting for moderation
@BetterReviews stats                    Review counts, ratings, last 7/30 days
@BetterReviews search <topic>           Find reviews mentioning a topic
@BetterReviews chat funnel              Where shoppers drop off in the AI chat
@BetterReviews email funnel             Email open / click / bounce + conversion
@BetterReviews help                     Show all commands

Review IDs look like REV-1234 and appear at the bottom of every review the bot shows you.

Or ask in plain English for things like “best photos for [product]” — see below.

Asking in plain English

The bot understands natural language for read commands. Multi-step questions and follow-ups both work:

  • “How many reviews came in this week?”
  • “Show me 5-star reviews”
  • “Any reviews mentioning shipping?”
  • “How are we doing on shipping complaints?” — the bot searches and summarises trends.
  • “What are people saying that’s critical?”
  • “How are our review emails performing?” — open / click / bounce rates plus conversion.
  • “Where are shoppers dropping off in the chat?” — engagement bucket breakdown.
  • “How many email-driven chats are no-shows?” — combines email origin with the no-show bucket.
  • “Best media and reviews for [product name]” — returns up to 3 photo + verbatim quote cards for a specific product, ready to drop into ad copy.
  • “3 best photos for [product name]” — same, terser phrasing.
  • “Reviews with photos for [product name]” — same surface, no “best” framing.
  • Inside a thread: “what about last week?”, “show me #2” — picks up the previous question.

Time windows. For funnel and stats questions, you can specify a window in plain English: “today”, “yesterday”, “the last 24 hours”, “this week”, “last 7 days”, “last 30 days”. The bot resolves these against today’s date in UTC and shows the exact window it queried at the top of every reply (Window: 2026-04-29 → 2026-04-30 (UTC) · 1 day(s)) so you can sanity-check it interpreted you correctly. Without a window, funnel queries default to the last 30 days.

Topic-search caveat. When you ask “any reviews mentioning shipping?”, the bot matches against keywords our analysis pipeline tagged on each review — not raw review text. Reviews submitted in the last few minutes, or with very rare/niche terms, may not be indexed yet. If the bot says “no matches” for a topic you’d expect to find, give it a few minutes for analysis to catch up, or browse the dashboard.

Product-name resolver caveat. When you ask for media or reviews for a specific product, the bot fuzzy-matches the name against your catalog (token-overlap, no semantic search yet). If your phrasing doesn’t match the title closely enough, you’ll see a “Did you mean:” card with up to 3 close candidates — pick one or retype. The bot always echoes the resolved product title in the lead-in (Found the *X*…) so you can catch a wrong match before reading the cards.

Approval and rejection commands always go through strict validation, not the AI — so a creative phrasing of “approve everything” will get rejected. Use the exact approve <id> form for moderation actions.

Roles — admin vs viewer

The bot supports two roles in your channel:

  • Viewer — can read everything: list reviews, see stats, search, view individual reviews.
  • Admin — viewer access plus can approve/reject reviews, send review request emails, and manage who has access.

Admins manage roles directly in the channel:

@BetterReviews add user @sarah as admin
@BetterReviews add user @marketing-team as viewer
@BetterReviews remove user @sarah

When you add or remove someone, the bot replies with the current count — for example “Sarah added as admin (admins in this channel: 3, viewers: 5).” That gives you an at-a-glance check on who has access without a separate audit step.

You can also use Slack’s built-in /who command in the channel to see the full list of members.

Notifications you’ll receive

The bot posts in your channel when something interesting happens:

  • Marketing-ready review — a high-quality review with marketing potential. Good for ad copy or social proof.
  • Support follow-up needed — a review where the customer mentioned a problem. Worth reaching out.
  • Outstanding review — exceptionally detailed, often with photos.
  • Spam blocked — auto-rejected. Just an FYI; you don’t need to do anything.
  • Flagged for review — possible spam or abuse our auto-moderation isn’t sure about. Click through to moderate manually.

Privacy & isolation

Each store gets its own private channel. The bot will only ever return data for the store that owns the channel you mention it in — cross-store leakage is structurally impossible.

When the bot interprets a typed command, it uses Anthropic’s Claude. Read replies (stats, lists, search summaries, funnel breakdowns) are also phrased by the same model — but every number, ID, and rating it quotes is checked against the structured data we sent it, and a reply that fails the check is dropped in favour of a structured stats block built directly from the API response. The model can’t quote a number it didn’t see, and you still get the underlying numbers either way.

For card responses (e.g. ad-asset suggestions), the bot uses templated layouts with verbatim quotes. Each rendered quote is guaranteed to be a literal substring of the original review (the snippet extractor never synthesizes or paraphrases text), and all merchant- and reviewer-supplied text passes through Slack-control-sequence neutralization before rendering — a reviewer can’t inject @channel pings or phishing links via review content.

Approval and rejection commands run through strict validation, not the AI — they never go through a language model. We strip personal data from internal records before storage, and customer email addresses are encrypted at rest.

Limits

Heavy use is throttled — if your team hits a limit, the bot tells you and resets within a minute. Messages over 500 characters get truncated, and replies are kept short by design (typically a sentence or two).

The bot replies in threads to keep channel history readable. If you need a permanent record of a moderation decision, the answer also lands in your BetterReviews dashboard.

If you suspect phishing

The legitimate BetterReviews bot:

  • Lives only in your dedicated channel — not in DMs, not in other channels.
  • Shows the verified BetterReviews app name and icon next to every message.
  • Never asks for your Shopify password, store API token, or payment details.

If anything else claims to be BetterReviews — a different Slack app name, a DM from “BetterReviewsHQ,” an email asking you to re-authorize — assume it’s phishing. Report it in your channel and we’ll investigate.

FAQ

Can I leave the channel?

Yes. The channel is yours. If you leave, you’ll stop receiving notifications and won’t be able to message the bot — but your reviews keep working as normal. Reach out via help@betterreviews.app if you want back in.

Can I add a teammate later?

Yes. Any admin can run add user @them as viewer (or as admin) once the teammate is in the channel as a guest. If they’re not in the channel yet, ping us — we’ll send them a single-channel guest invite.

What if the bot stops responding?

Try @BetterReviews help — if you get nothing back within a minute, the bot is probably temporarily down. Message the BetterReviews team in your channel; we get pinged automatically and respond as soon as we can.

Why are some replies prose and others cards?

Stats, search, and topic questions get a short voiced sentence — the answer is a number or a summary, and prose reads better. Product-asset questions (“best photos for X”, “reviews with photos for X”, “ad for X”) return Block Kit cards because the answer is a list of artifacts: photo, verbatim quote, rating, and a deep link to open the review in your admin. The cards are designed for copy-paste into ad copy.

Can I see how my review emails are performing?

Yes — ask the bot @BetterReviews email funnel (or “how are our review emails performing?”). You get aggregate counts and rates over the last 30 days: sent, opened, clicked, bounced, complained, plus a per-trigger breakdown (created / paid / fulfilled / delivered) and conversion (sent → started chat → submitted review). It’s read-only and aggregate-only — no per-customer data, no email addresses.

How do I see chat drop-off?

Ask @BetterReviews chat funnel (or “where are shoppers dropping off in the chat?”). The bot returns engagement buckets — no-show (chat opened, never typed), one-and-done, mid-flow, completed, long, and failed — broken down by origin (email vs widget vs API). Combine the two to diagnose specific friction: “How many email-driven chats are no-shows?” surfaces the gap between email click and chat start.

How do I get help if the bot can’t answer?

Just @ the BetterReviews team in your channel — that’s literally why we’re there. Founder-led support is part of every plan, no enterprise tier required.