Brand guidelines for responding to negative DMs are a short written policy that every agent follows when an unhappy customer messages your brand privately. The policy sets the tone rules, what an agent may offer without approval, which messages go to a manager, how to handle abusive DMs, and when to move a conversation to email or phone. Without one, each agent improvises, and a private reply can end up as a public screenshot. Below is what to put in the policy and a template to copy.
DMs are often treated only as the place a public complaint goes to be resolved. Our guide to handling negative feedback on social media covers moving complaints out of public view, and our guide on how to respond to negative tweets covers the public side on X. But many negative messages start in the inbox, and they need their own rules. If you want a team that already works this way, that is our social media support service.
Why negative DMs need written rules
A DM feels private, so agents relax. Anything an agent types can be screenshotted and posted, with the brand’s name on it and no context around it. A written policy protects the customer, the agent and the brand in three ways:
- Consistency. Two customers with the same problem get the same answer, whoever is on shift.
- Speed. Agents do not stop to ask a manager for things the policy already allows.
- Cover. When an agent refuses something, they are applying a rule, not making a personal call the customer can argue with.
What to put in the policy
1. Scope and channels
List every inbox the policy covers: Instagram, Facebook Messenger, X, TikTok and any others. Say who owns each one and when it is staffed. If you manage Facebook and Instagram together, Meta Business Suite’s Inbox brings messages from Messenger, Instagram and WhatsApp into one place, and also shows comments on your Facebook and Instagram posts (Meta Business Help Center). One queue makes a single policy easier to apply.
2. Response time targets
Set a first-response target for business hours and a separate one for nights and weekends, and publish your hours on each profile. A first reply can be short: “Thanks for messaging us. I am checking your order now.” Then keep to whatever time you promised.
Automation can cover the first touch. Meta Business Suite lets you set an away message, an instant greeting, and custom keyword replies, and it sends a keyword reply after 15 minutes unless someone on your team responds first (Meta Business Help Center). Use automation to acknowledge and route. Do not use it to answer a complaint. An upbeat automated reply to an angry message makes things worse.
3. Tone rules
| Do | Do not |
|---|---|
| Use the customer’s name and restate their problem in one line | Open with “As per our policy” |
| Apologize for the experience, clearly and once | Apologize repeatedly or grovel |
| Say what you are doing and when they will hear back | Promise an outcome you cannot guarantee |
| Match the customer’s formality, and stay calm if they are not | Mirror anger, sarcasm or capital letters |
| Sign with a first name | Reply as an anonymous brand |
| Explain a “no” with the reason behind it | Say “there is nothing we can do” |
4. Identity and data rules
Agents need enough to find the order and nothing more. State what they may ask for (order number, the email on the order, a photo of the item) and what they must never ask for in a DM (full card numbers, passwords, security codes). If a customer sends payment details unprompted, the policy should tell the agent not to repeat them and to move the conversation to a secure channel. Remember that a DM is not a secure channel. On X, deleting a message removes it only from your own account. The other person can still see it (the X Help Center).
5. What an agent may offer without approval
This is the section that saves the most time. Write down, by complaint type, what the front line can offer on their own and what needs a lead. For example:
| Complaint | Agent may offer | Needs a team lead |
|---|---|---|
| Late order | Tracking check, carrier claim, reship once lost | Refund plus reship, expedited shipping at your cost |
| Damaged or wrong item | Replacement or refund on a photo | A second replacement for the same order |
| Refund not received | Status check, refund reference number | Refund before the return arrives |
| Poor service from a previous agent | Apology, fix the original issue | Any goodwill credit above the agent’s limit |
| Subscription or billing error | Cancel, refund the latest charge | Refunds beyond the latest charge |
Put your actual limits in the internal version: the dollar amount of store credit an agent may give, and how many times a customer can receive goodwill in a year. Agents who know their limits say yes faster and say no more calmly.
6. Escalation triggers
Some messages leave the front line the moment they arrive. List them:
- A threat of legal action, or a message from someone who says they are a lawyer
- Any mention of injury, illness, an allergic reaction or a safety defect
- A journalist, or a customer who says they are going to the press
- A chargeback or dispute already filed with their bank
- A customer with several open cases, or who has escalated before
- Anything involving a minor
For each, name who takes it and how fast. The agent’s reply is the same every time: “Thank you. I am passing this to (a named role) now, and they will reply to you by (time).” The only change is the role and the time.
7. Abusive DMs
Criticism is not abuse. A customer who swears about a late order is still a customer. Abuse is a threat, a slur, sexual content, harassment of a named agent, or the same hostile message sent again and again.
The policy should give agents a clear sequence:
- Reply once, calmly, with the help you can give and a line on how you expect to be spoken to.
- If it continues, end the conversation politely and log it.
- Block or report through the platform, and tell a lead.
Each platform gives you tools for this. On X, you can report a single message or a whole conversation from inside the DM, and a blocked account cannot send you direct messages (the X Help Center). On Instagram, the Hide unwanted message requests setting sends requests that may contain scams or offensive words to a spam folder, and Hidden Words lets you add your own list of words and emoji to filter from message requests (Instagram Help Center). If a message contains a credible threat of violence, report it and contact local law enforcement. X notes it will work with law enforcement on reported threats.
Agents also need protection. Rotate the people who handle abusive queues, let an agent hand off a conversation that has turned personal, and never require an agent to use their full name.
8. When to move to email or phone
A DM thread is fine for a tracking number. It is the wrong place for anything that needs documents, several back-and-forths, or a sensitive detail. Move the conversation when:
- The fix needs an attachment, a signature or a form
- Payment or identity details are needed
- The thread has gone past three or four exchanges without a fix
- The customer asks for a call
Say it as help, not as a brush-off: “This will be quicker by email. I have just sent you a message from support@ with everything we have so far.” Then make sure the email arrives. A customer told to email who then waits two days is back in your DMs, angrier, and possibly in public too.
9. Closing and logging
End every negative DM with what happened and what the customer should expect next. Log the case in your help desk against the customer’s record, with the complaint type, what was offered and who approved it. That log is how you spot the same complaint arriving twenty times in a week, which is a product or operations problem rather than a support one.
A template you can copy
Paste this into a shared document and fill in the parts in parentheses for your own team.
Negative DM response policy for (brand)
- Channels covered: (Instagram, Facebook, X, TikTok). Owner: (team). Staffed (hours, time zone).
- First response: within (time) in staffed hours, by (time) next morning otherwise. Automation acknowledges only; it never answers a complaint.
- Tone: use the customer’s name, restate the problem, apologize once, say the next step and when, sign with a first name. Never argue, never use sarcasm, never quote policy as the first line.
- Data: ask only for order number, order email and photos. Never ask for card numbers, passwords or codes.
- Agent authority: (the table of what agents may offer by complaint type, with limits).
- Escalate at once: legal threats, injury or safety, press, chargebacks, repeat escalations, minors. Owner and response time for each.
- Abuse: reply once calmly, then end, block or report, and tell a lead. Agents may hand off any thread that turns personal.
- Move to email or phone: documents, payment or identity details, more than (number) exchanges, or the customer asks.
- Close and log: state the outcome and next step, then log type, offer and approver in (help desk).
When you publish it internally, the parts in parentheses are the decisions your team makes once, so no agent has to make them in the middle of an angry conversation.
Reply examples for common negative DMs
“Where is my order? This is ridiculous.” “Hi Jordan, I am sorry, it has taken far too long. I can see it left our warehouse on Monday and I have opened a trace with the carrier. I will message you here by 3 pm today with an update either way. Sam”
“Your product broke after a week. I want my money back.” “That should not happen after a week and I am sorry. Could you send a photo of the damage? Once I have it I can refund you straight away, or send a replacement if you prefer. Sam”
“Your last agent was useless.” “I am sorry you were left without an answer. I have read the whole thread, so you do not need to repeat anything. Your refund was approved yesterday and should reach your card within a few days. I will check again on Friday and message you here. Sam”
“I will be posting about this everywhere.” “I understand, and you are free to. I would still like to fix it for you first. I can send a replacement today or refund you in full, whichever you prefer. Sam”
Who should run your DM inbox
A negative DM policy only works if someone applies it every day, at the hours your customers message. Atidiv’s messaging support and social media teams work to each brand’s approved communication policies, and we have written about staffing Instagram and TikTok DMs at volume. Our teams deliver an average QA score of 98% and a 4.8 average CSAT, in 11+ languages, at up to 60% lower cost than an in-house team.
Frequently asked questions
Should a business reply to every negative DM?
Reply to every message from a real customer, even if the reply is only to say you are checking. Do not engage with spam, and after one calm reply, stop engaging with abuse. End the conversation, block or report the account, and log it.
Can we use saved replies for negative DMs?
Use saved replies for facts that do not change, such as return steps or tracking links, and write the opening and closing lines fresh for each customer. A reply that is obviously pasted will feel dismissive to someone who is already upset.
When should a DM move to email or phone?
When the fix needs documents, payment or identity details, when the thread has gone several exchanges without a fix, or when the customer asks. Tell the customer why, send the email yourself, and make sure it arrives.
Ben has a strong expertise in combining data and strategic thinking to help businesses that are in the process of growth, plan smarter, grow faster and make better business decisions. He has experience in economic research and managing operations, providing him with a broad & rounded analytical perspective from many angles. Curiosity is coupled with logical reasoning. Therefore he enjoys tackling complex issues and converting analysis/insights into action.