Here's how I'd handle each. The third one needs a decision from you before anything goes out.
1. "Ordered 2 weeks ago and still nothing??"
Type: Complaint, and likely a fair one. Two weeks is about 10 business days, which is past your 5 to 7 day window. Don't hide it or argue.
Public reply:
Two weeks is too long, and that's on us, not you. Send us a DM with your name and order number and we'll track it down today. If it's lost, we'll make it right.
Private follow-up (after they DM):
Hi [name], thanks for reaching out, and I'm sorry this has dragged on. Could you send the email or name on the order and the order number? I'll check where it is. [Status]. If it's lost or stuck, I can send a replacement or refund you, whichever you'd rather. Either way I'll update you [timeframe you can commit to]. You can also reach us at [support email].
Alternate tone (more formal):
You're right to expect it sooner. Our standard shipping is 5 to 7 business days, and your order is past that. Please DM us your order details so we can investigate and resolve it.
Don't say:
- Don't defend the "free shipping in 5 days" line, and don't blame the carrier before you know what happened.
- Check what the post says. Your policy is 5 to 7 business days, so if the post says "5 days," that's worth fixing.
- Don't promise a delivery date you haven't confirmed.
2. "This brand is a scam... they steal your card info"
Type: Misinformation, and possibly spam. The account has no posts, so it may be fake or a drive-by troll. The claim is serious, but there's no evidence in it.
Options:
- Report it first. Use the platform's report flow for false or misleading content or fake accounts. Check the current options, since they change. Hide or delete it only if your written guidelines allow that. Don't remove it just because it's negative.
- If you reply, keep it short and calm:
We take card security seriously. Payments are processed through [payment processor], and we never see or store your card details. [Verify this is true first.] If you've had a problem with an order, DM us and we'll help.
Alternate tone (playful but direct):
Whoa, that's a big claim! Payments go through [processor], so we never see your card number. If something's actually gone wrong with an order, DM us and we'll dig in.
Don't say:
- Don't call them a fake or a bot, and don't get defensive.
- Don't claim security facts you haven't confirmed. Fill in the processor line only if it's accurate.
- Don't reply at all if you'd rather not feed it. Reporting and leaving it is a valid choice.
3. "My daughter broke out in a rash... I'm talking to a lawyer."
Type: Crisis. It involves a possible injury to a minor and a legal threat. I'm not drafting a reply to improvise. Whatever you say could affect a real child's care and a possible legal process.
First steps:
- Don't reply substantively yet. Don't offer a refund, an apology that admits cause, or a guess about ingredients or allergies.
- Tell the owner or decision-maker now. This isn't a community-management call.
- Save the DM (screenshot) and note the time received.
- Check your records for the lotion's batch or lot, and check whether any other reports have come in. Don't contact the customer about this until you've decided what to say.
- Consider talking to your insurer or a lawyer about what to say and how to preserve records. I can't give legal or medical advice.
- Check your local rules on reporting adverse reactions for cosmetics.
Optional holding line, only if someone with authority approves it. It makes no claims:
Thank you for telling us. We're sorry to hear your daughter is unwell, and we want to understand what happened. [Owner name] will contact you directly at [contact].
Please make sure whoever replies knows the child's wellbeing comes first. It's reasonable to suggest in the follow-up that they seek medical advice if the rash persists. Have that wording approved too.
Don't say:
- "It's not our product."
- "Our lotion is safe."
- Anything that blames the child or the parent.
- Anything about cost, refunds, or compensation.
Placeholders to fill in
- Timeframe for updating the customer in #1.
- Payment processor name in #2, and confirmation that you never see card data.
- Who decides and who replies on #3, and the contact route.
- The wording of the "free shipping in 5 days" post, if it exists.
If you tell me what the post actually says, I can adjust #1. I can also build a reusable template set if that would help.
Fictional skincare shop. The model treated message 3 (a possible injury to a child plus a legal threat) as a crisis and drafted no reply, only first steps and an optional holding line to be approved by the owner. It used placeholders for facts it was not given, such as the payment processor and the update timeframe. Nothing was posted.
What it does
Helps a brand, creator or small business answer people online in their own voice. It sorts each message (praise, question, complaint, misinformation, troll or abuse, spam, crisis), picks a response pattern (acknowledge, own your part, act, move it to private), and drafts a short public reply, a more detailed private follow-up and an alternative tone, with a note on the traps to avoid. It never invents facts such as refunds, policies or causes; it asks, or leaves clearly marked placeholders. For threats of legal action, injury or safety, harassment, self-harm or viral crises it does not draft a reply to improvise: it explains what to do first and who must decide. It can also build a reusable template library, community rules and an escalation list.
Output
The message type, a public reply, a private follow-up, an alternative tone, pitfalls, an escalation note when needed and the placeholders you must fill.
Good for
Small businesses, creators and community managers who handle comments, DMs and reviews.
Low risk: pure instructions with no scripts, no network access and no file writes. It writes drafts and does not post, delete or moderate anything or connect to any account. It does not know your real policies or what actually happened, so check every placeholder and fact before replying. Replies about injury, legal threats, safety or data breaches need a person with authority first, and it gives no legal, medical or safety advice. Test-run once with a fictional skincare shop; the model correctly escalated the injury and legal message. Nothing was posted. Original by AIBars (MIT).