Bounce rate is the most under-monitored metric in email marketing, and one of the most consequential. Open rates are directional and increasingly distorted by privacy features. Bounce rate is a hard number that mailbox providers read as evidence about where your list came from.
The distinction that matters
Every bounce arrives with an SMTP response code, and the first digit tells you almost everything.
| Type | Code | Meaning | Correct action |
|---|---|---|---|
| Hard | 5xx | Address or domain does not exist | Suppress permanently, never retry |
| Soft | 4xx | Temporary — full mailbox, server down | Retry with backoff, then escalate |
Common hard bounces include 550 No such user and 553 Mailbox name invalid. Common soft bounces include 452 Insufficient storage and 421 Service unavailable.
The single worst mistake in bounce handling is retrying a hard bounce. The address will never exist. Every retry is another data point telling the provider you are not processing their responses — which is behaviour characteristic of bulk software that does not care.
What providers infer from your bounce rate
Think about it from their side. Someone emailing a list of people who signed up will have very few invalid addresses — a handful of typos, some staff who have left jobs. That is well under 1%.
Someone emailing a list they bought will have a great many invalid addresses, because purchased lists are typically old, scraped, or partly fabricated.
So hard-bounce rate functions as a proxy for list provenance. It is not a perfect signal, but it is a good one, and it is cheap for providers to compute. A sender arriving with a 12% hard-bounce rate has effectively announced where the list came from.
Targets worth holding to
- Under 2% total — healthy.
- Under 1% hard — what a maintained list looks like.
- 2–5% — investigate now; something is stale.
- Above 5% — expect filtering. Stop and clean the list.
- Above 10% — you are actively damaging the domain with every send.
Preventing bounces before you send
Most hard bounces are avoidable, and validation at upload is where the work belongs.
- Syntax checking. Catches typos and malformed addresses without any network calls.
- MX record lookup. Confirms the receiving domain is configured to accept mail at all. A domain with no MX record cannot receive email, and every message to it will bounce.
- Duplicate removal. Duplicates inflate volume and can mean contacting the same person twice in one campaign, which generates complaints.
- Suppression checking. Cross-reference against people who previously bounced or opted out, so re-importing an old list cannot resurrect them.
On Email.biz.pk these four run automatically when a CSV is uploaded, and invalid rows are reported before the campaign is queued rather than discovered afterwards from the bounce report.
Handling bounces after the fact
Hard bounces go into the suppression list immediately and permanently. Storing suppression entries as SHA-256 hashes rather than plain addresses is a small detail with a real benefit: you can reliably check whether an address has bounced without retaining readable contact data for someone you will never email again.
Soft bounces retry on a widening backoff. If they continue failing past roughly 72 hours, treat them as hard — a mailbox full for three days is realistically abandoned.
The list-decay problem
Email lists decay at roughly 20–30% per year through job changes, domain closures and abandoned addresses. A list that was clean eighteen months ago is not clean now, and nothing about it changed except time.
This is why re-engagement matters more than acquisition for established senders. If you have contacts you have not mailed in over a year, do not simply add them to the next campaign — send a small test batch first and watch the bounce rate. Discovering a 15% bounce rate on 500 messages is recoverable; discovering it on 50,000 is not.