Bounces and complaints

Bounces and complaints are the feedback your sending generates, and they are the earliest warning you get that something is wrong. This page explains what each type means, what happens automatically, and what the numbers on a campaign report should be before you send the next one.

Hard bounces#

A hard bounce is a permanent failure. The receiving server has told you this address will not accept mail now or later. The usual causes are a mailbox that no longer exists, a domain that no longer exists, and a typo that was never a real address in the first place.

Hard bounces are handled for you. An address that hard bounces is suppressed automatically, so later campaigns do not attempt it again. You do not have to find them and remove them by hand, and you should not re-add an address that has hard bounced in the hope it comes back.

The reason this matters beyond the wasted send is what a hard bounce tells the receiving provider about you. Legitimate senders mail people who signed up recently enough that their addresses still work. A sender producing a high proportion of hard bounces looks like someone mailing a list they did not build, and that judgement is applied to the whole send, not to the failing addresses.

Soft bounces#

A soft bounce is a temporary failure. The address exists but the message could not be delivered at that moment. Common causes are a full mailbox, a server that was down or overloaded, a message larger than the recipient's limit, and temporary rate limiting where the receiving provider is deliberately slowing you down.

Soft bounces are retried, so a single soft bounce is not a signal about the address. What is a signal is the same address soft bouncing on every campaign for weeks. That is usually an abandoned mailbox that filled up and stayed full, which is functionally a dead address even though it never returned a permanent failure. Repeated soft bounces are eventually treated the same way as a hard bounce.

A cluster of soft bounces all at one domain in a single send is a different problem entirely, and it usually means that provider is throttling you rather than that those mailboxes are full. Slow down and check your complaint rate before sending again.

Spam complaints#

A complaint is a recipient pressing the spam or junk button in their mail client. Their provider records it and, where a feedback loop exists, reports it back. Complaints on the platform's own sending are processed automatically, and a complaining address stops receiving your mail.

Complaints are the most damaging signal in email, and the numbers involved are much smaller than people expect. The working ceiling is 0.1 percent, meaning one complaint per thousand delivered messages. Sustained rates approaching 0.3 percent are enough for a major provider to route your mail to the spam folder for every recipient, including the ones who never complained.

Put concrete numbers on that. On a 30,000-message send, thirty complaints is at the ceiling. Ninety is a serious problem. Those are small enough numbers that a single send to a list you have not mailed in a year can produce them, which is why the advice about stale lists is so insistent.

Complaints outweigh their volume because of what they tell a provider. A bounce says an address is wrong. A complaint says a real person, with a working mailbox, looked at your mail and decided it was unwanted. No amount of authentication argues with that.

The usual causes, in order:

• Time. People forget they signed up. The longer since the last contact, the higher the complaint rate.
• Mismatch. The mail is not what they expected when they gave you the address.
• A hidden unsubscribe. If leaving is harder than clicking spam, some people click spam. A visible one-click unsubscribe is a complaint-reduction feature.
• Frequency. Going from monthly to daily without saying so.

Reading the figures on a campaign report#

After a send, open the report for that campaign and read the bounce and complaint figures next to the number delivered. Two habits make this useful.

First, work in percentages, not counts. Thirty bounces means nothing until you know whether the send was 1,000 messages or 100,000.

Second, compare against your own previous sends, not only against absolute thresholds. A jump from your normal 0.4 percent bounce rate to 1.5 percent is worth investigating even though 1.5 percent is not alarming on its own. Something changed: a new import, a different segment, an older part of the list included for the first time.

What to do after a bad send#

Work through this in order.

1. Identify the source. Was the bad segment an old import, a purchased file, or a group that had not been mailed for a long time? A single import is usually responsible.
2. Remove it from future sends. Suppress that group rather than hoping it improves. Blacklist and suppression covers how addresses are held out of sends.
3. Verify what is left. Run the remaining addresses through verification to remove the undeliverable ones before your next send.
4. Send small and engaged. The next campaign should go to people who opened something recently. Rebuild the pattern of wanted mail before returning to full volume.
5. Set a cadence for hygiene so the same list does not decay again. List hygiene describes what to remove and when.