# What to tell a customer whose reply to your support address bounced

> A bounce on a customer reply goes to them, not you. Ask for the bounce text, sort the cause into typo, your setup or theirs, then reply with a clear next step.

- Published: 2026-11-10
- Author: Koltrix Team
- Category: Customer support
- Source: https://koltrix.com/blog/customer-reply-to-support-address-bounced

A customer says: "I replied to your email and it came back as undelivered." You check the support inbox and find nothing, because a bounce for an inbound message is sent to the sender. Your system may never have seen the message at all. That makes this different from the usual bounce question, where you are the one sending.

This post is the support-side playbook for that case: what to ask for, how to tell whether the fault is a typo, your setup or theirs, and what to write back.

## Why don't you see the bounce yourself?

A bounce for an inbound message goes to the person who sent it, so only the customer sees it. A bounce is a delivery status notification, a message from a mail server that says a message could not be delivered. [RFC 3464](https://www.rfc-editor.org/rfc/rfc3464) defines the format. Often the refusal happens during the SMTP conversation itself, and the customer's own mail system then writes the notice. Your mailbox log may have no trace.

So the first step is always the same: get the bounce text from the customer.

## What should you ask the customer for?

Ask for the full bounce message pasted as text, the address they typed, and roughly when they sent it. A screenshot of the top half hides the lines you need.

> Thanks for telling us. I want to look at the exact error. Could you paste the whole returned message here as text, and tell me which address you replied to? That tells me whether the problem is the address, our side or yours.

Then find three things in it:

1. **The address that failed.** Is it spelled the way you publish it?
2. **The status code.** A code starting with 4 is temporary and the sender's server keeps retrying. A code starting with 5 is permanent. The [code classes are defined in RFC 3463](https://www.rfc-editor.org/rfc/rfc3463).
3. **The server that wrote the error.** Look for a line such as "Remote server returned" or "Diagnostic-Code", which names the system that refused the message.

If you want the code basics in plain words, [explaining a bounced email to a customer](/blog/explaining-a-bounced-email-to-a-customer-in-plain-words) covers the common ones.

## Is it a typo, your setup or their setup?

The failure falls into one of three groups. This table maps the usual signals to a cause.

| What the bounce says | Most likely cause | Whose side |
| --- | --- | --- |
| 5.1.1 or "user unknown" for your address | The address was typed or edited wrong, or the mailbox does not exist | Usually the sender, sometimes you |
| 5.1.2 or "domain not found" | The domain in the address is misspelled, or your domain has no working DNS | Sender, or you if the domain is yours |
| "No MX" or "unable to route" | Your domain has no mail routing records | You |
| 5.7.x policy or authentication refusal | Your server rejected their message, or their server refused to send it | Check which server wrote the text |
| 4.x.x delayed | A temporary problem, such as a busy server | Wait first |

### Did they reply to the address you think they did?

Check the address in the failed message against the one in the original email. Replies go to the Reply-To address if one is set, and to the From address otherwise, as [From, Reply-To, envelope sender and Return-Path](/blog/from-reply-to-envelope-sender-return-path-explained) explains. A common cause is a transactional or notification email sent from an address that cannot receive mail, which is one more reason to [avoid noreply addresses](/blog/noreply-address-case-against).

If the address in the bounce is not one your team owns, fix the Reply-To or From on the sending side. That is a configuration bug, and the customer did nothing wrong.

### Is your domain able to receive mail?

If the error says the domain has no mail server, or the message cannot be routed, your MX records are the first suspect. MX records are DNS records that tell the world which servers accept mail for a domain. Check them from outside:

```bash
dig +short MX yourcompany.com
```

An empty answer means no server is listed for the domain. A wrong answer means mail goes somewhere you do not control. If you recently changed DNS or moved providers, this is the likely cause. [MX records are explained here](/blog/mx-records-explained), and the setup steps for a support mailbox on your own domain are in [setting up a support shared mailbox on a custom domain](/blog/set-up-support-shared-mailbox-custom-domain).

Test it yourself, from an outside account. Send a short message from a personal mailbox at a different provider to the support address and see whether it arrives. If your test works and the customer's fails, the cause is more likely in the customer's setup.

### Did your side reject their message?

A rejection by your receiving server contains a policy reason in the text. If the line names your server and a policy such as authentication, the customer's own domain may be failing a check, for example a strict policy on a domain whose mail is sent through a service that is not set up for it. That is a problem to fix on their side, and the bounce text will usually say what failed. Do not loosen your own checks to let one message through without understanding why it was refused.

### Was it only delayed?

A notice that says the message is delayed, not failed, means the sending server is still trying. Tell the customer not to send it again, because duplicates make threads harder to read. If a final failure follows, start the steps above.

## What do you write back?

Be specific about whose side the cause is on, and give one next step. Three short templates cover most cases.

**Typo or wrong address:**

> The returned message shows it was addressed to "support@exmaple.com", with the letters swapped. Our address is support@yourcompany.com. Please send your reply there, or just reply to this message again.

**A problem on your side:**

> Thanks for letting us know. The error shows our address did not accept your message, and that is on our side. I am fixing it now. In the meantime I have noted your question here, and I will answer it in this thread as soon as I can. If you would like, send it to me at [another address].

**A problem on the customer's side:**

> The returned message says your mail system refused to send to us, with the reason "[quote one line]". That is a setting on your side that your IT contact can check. Until then, you can reach us at [other channel], and I will pick up from there.

Always give the customer a way forward in the same message. A person whose reply bounced has already had one bad experience, and a second round of "send us more details" will feel like a runaround.

## How do you stop it happening again?

- **Publish one support address** and use it everywhere, so the one customers see is the one that works.
- **Send a monthly test** from an outside mailbox to every public address, including the ones on invoices and in email footers.
- **Check the Reply-To** on every automated message that a customer might answer.
- **Keep DNS changes on a checklist** that ends with a test message to the support address.
- **Keep a short reply template** for "my reply bounced", with the three cases above.

## Bottom line

When a customer's reply to your support address bounces, you will not see the bounce, so ask for its text. Read the address, the code and the server that wrote the error, and sort it into typo, your setup or their setup. Then answer with the cause in one plain sentence, one next step, and another way to reach you while it is fixed.
