Hard bounces: when should an address come off the suppression list?

The usual rule is simple: an address hard bounces once, you suppress it everywhere and never send to it again. Soft bounces get a few days of retries, and an address that keeps soft bouncing campaign after campaign ends up treated as hard. Every sending platform works more or less like this, and for good reason: sending to dead addresses is one of the fastest ways to lose sender reputation.

The problem is that not every hard bounce is a dead address. A small business forgets to renew its domain, every email to it bounces as “domain not found” for four days, then they renew and everything works again. A receiving server is misconfigured for an afternoon and rejects everything with a permanent error. Suppress those forever and that customer never hears from you again, even if they order next month. Make it too easy to come back and you are mailing dead addresses again. This post is about where to draw the line.

What a bounce code tells you

When a mail server rejects a message it answers with a status code. The first digit is the one that matters most: 4 means “not now, try again” (soft bounce) and 5 means “no, and don't retry” (hard bounce). The next two numbers say why. These are the ones you will see most:

Notice that only the first two say anything definitive about the person. The domain-level codes describe the domain, and 5.7.1 describes your reputation with that receiver, not the recipient. Treating all five of the 5.x.x rows the same way is where most suppression lists go wrong.

A suppression policy that holds up

1. Mailbox-level failures stay suppressed for good

5.1.1 and 5.1.10 mean the receiver looked for the mailbox and it was not there. People do not get their old address back. Keep these suppressed permanently, with no exceptions and no manual override. If you only apply one rule from this post, apply this one.

2. Domain-level failures go to quarantine, with a deadline

5.1.2 and 5.4.4 go into a separate bucket, not the permanent list. Once a week, look up the MX record of each quarantined domain. If it resolves for a week straight, release the addresses, but only into a single low-stakes send: an order confirmation, a plain-text note, never the full campaign. If that bounces again, they go to the permanent list. And cap the quarantine: a domain that has been dead for two months is not a renewal mix-up. Stop checking and suppress.

3. Look at bounces by domain and time, not one by one

Twenty addresses at the same domain all bouncing within the same hour is almost never twenty dead mailboxes. It is the receiver being broken. When bounces cluster like that, reclassify the whole batch as soft and retry later, whatever code came back. The code says what the server reported; the pattern says whether to believe it.

4. A new signup or order proves the address, not the consent

When someone who bounced shows up again with the same email, by signing up on a form or placing an order, you know the address works now. That is a good reason to send the receipt. It is not, on its own, a reason to put them back on the newsletter: a working mailbox and a person opting back in are two different things. Let the checkbox at checkout or the signup form decide the marketing part, the way it would for any new contact.

5. Manual unsuppression: one at a time, with a paper trail

Should a store owner be able to take an address off the list by hand? Yes, for the real cases: a customer who calls and says their domain was down last week. Keep it to one address at a time, require a reason, and log who did it. Never offer a bulk option. The moment there is a button that unsuppresses in bulk, someone uses it on three thousand dead addresses to “grow the list” the week before Black Friday, and the reputation damage lands on every campaign after that.

How Mailchimp handles it

If you send with Mailchimp, most of this is decided for you. A hard bounce moves the contact to cleaned, and so does a run of soft bounces on consecutive campaigns. Cleaned contacts do not count toward your bill and cannot be emailed. You cannot flip them back by hand or by importing them again; the only way back in is the person signing up again through one of your forms, which is also the right test, because it proves the address works now. We cover the details in what cleaned means in Mailchimp.

That design matches the policy above on the points that matter: dead mailboxes stay out, a new opt-in brings people back, and there is no bulk unsuppress button to misuse. What it does not give you is the quarantine for domain-level failures, so a customer whose domain lapsed for a few days stays cleaned until they sign up again. The practical fix is to look at your cleaned list by domain every month: a cluster of cleaned contacts at one domain, all from the same campaign, is a receiver that broke, not a group of customers who left, and it is worth a short note asking them to sign up again.

The list that costs you more than bounces

Bounced addresses are the easy part: the platform catches them and stops sending. The contacts that quietly hurt your delivery are the ones that still accept mail and never open it. acadion finds those in your Mailchimp audience, with your rules and protections, and archives them on a schedule, reversibly. See what it would do with a free audit.