Understanding email bounce rates: hard vs soft bounces and how to fix them

Every bounce is a receiving server telling you it could not or would not accept your message. A few bounces are normal, even on a perfectly run list: people change jobs, abandon mailboxes and mistype addresses. The problem is a rate that climbs and keeps climbing, because mailbox providers treat it as a direct measure of how well you look after your list.

This guide explains what bounce rate is, the difference between hard and soft bounces, how to read the codes, what a healthy rate looks like and what to do to bring yours down.

What bounce rate means and how to calculate it

Bounce rate is the share of sent messages that the receiving server rejected.

bounce rate = (bounced messages / messages sent) x 100

Example: 180 bounces / 12,000 sent x 100 = 1.5%

Some platforms divide by messages sent, others by messages attempted after excluding suppressed addresses. Either is fine as long as you are consistent. More useful than the headline number is the split: hard vs soft, and by mailbox provider.

What happens when a message bounces

When you send, the receiving server can do three things:

1. Accept the message. Delivery succeeds (though it may still land in spam).
2. Defer it with a temporary 4xx response. Your server queues the message and retries, usually for up to a few days.
3. Reject it with a permanent 5xx response. Your server gives up and generates a non-delivery report (NDR).

Sometimes the rejection happens during the SMTP conversation (a synchronous bounce). Sometimes the server accepts the message and sends an NDR back later (an asynchronous bounce). Your ESP collects both and classifies them for you.

Hard bounces vs soft bounces

Hard bounces

A hard bounce means delivery to that address will not succeed. The most common is 550 5.1.1, user unknown. Others include 5.1.10 (recipient address not found, often seen at Microsoft), 5.1.0 (bad address) and 551 or 553 (address not accepted).

Policy rejections such as 5.7.1 are technically hard bounces too, but they are different in kind. The address may be fine; the receiver is refusing you. Suppressing the address will not fix that. You need to fix whatever caused the block (see below).

Soft bounces

A soft bounce means the address exists but delivery failed this time. Common examples:

• 4.2.2 and 452: mailbox full or insufficient storage.
• 421 and 4.7.0: service unavailable or temporarily deferred, often throttling.
• 450 and 451: mailbox temporarily unavailable or a local processing error.
• 4.7.1: greylisting or temporary policy deferral.
• 4.4.1: connection timed out.
• 4.7.28 and 4.7.26: Gmail rate limiting and authentication-related deferrals.

How to read a bounce code

Most bounce messages carry two codes: a basic SMTP reply, such as 550, and an enhanced status code in the form class.subject.detail, such as 5.1.1 (RFC 3463).

550 5.1.1 The email account that you tried to reach does not exist.
^   ^ ^ ^
|   | | +-- detail: bad destination mailbox address
|   | +---- subject: addressing
|   +------ class: 5 = permanent
+---------- basic SMTP reply

• The first digit tells you what to do. 2 is success, 4 is temporary (retry), 5 is permanent (stop).
• The middle number groups the problem: 1 addressing, 2 mailbox, 3 mail system, 4 network and routing, 5 protocol, 6 content, 7 security and policy.
• The last number narrows it down.

Servers do not always follow the standard exactly, so read the human text too. Our bounce code pages explain each common code with provider-specific notes.

Reading bounces by mailbox provider

The same problem can look different depending on who is rejecting you. A few patterns are worth knowing.

• Gmail uses clear enhanced codes. 5.1.1 means the account does not exist. 5.7.26 and 4.7.26 mean authentication failed. 4.7.28 and 421 mean rate limiting, usually because your reputation does not support your current speed.
• Microsoft often returns 5.1.10 for unknown recipients and 550 5.7.1 with an internal reference for reputation blocks. Deferrals with 4.7.0 on new IPs are common during warmup.
• Yahoo uses 421 and 4.7.0 style deferrals heavily when it wants you to slow down, and 554 for hard policy blocks.
• Corporate filters vary widely. Some return 5.7.1 for everything from blocklist hits to recipient-level rules, so read the text.

This is why per-provider reporting matters. A 1.5% overall bounce rate looks fine until you see that 40% of it came from one provider in one afternoon.

A simple suppression policy

Write your rules down, automate them and apply them consistently.

The key distinction is between bounces caused by the address and bounces caused by you. Suppress the first kind. Investigate the second.

What counts as a good bounce rate

Context matters. Transactional mail such as receipts and password resets usually bounces less than marketing mail, because the address was used moments before. A first send to an old list will bounce far more than a weekly send to active subscribers. Many ESPs also enforce their own limits and may pause an account that crosses 5% or more on a single campaign. For sector comparisons, see our deliverability benchmarks by industry.

What causes bounces

Causes of hard bounces

• Addresses that do not exist. Typos at sign-up, fake addresses entered to get a discount, closed employee accounts.
• Domains that cannot receive mail. Misspelled domains (gmial.com), expired domains or domains with no MX records. You can check any domain with the MX Lookup.
• Purchased, scraped or appended data. These lists are full of dead addresses and traps.
• List decay. Business lists in particular lose a meaningful share of valid addresses each year as people change jobs.

Causes of soft bounces

• Full mailboxes. Often genuinely temporary, but persistent ones usually mean an abandoned account.
• Server problems. Outages, maintenance or overload on the receiving side.
• Throttling and greylisting. The receiver is deliberately slowing an unfamiliar or suspicious sender.
• Message size. Oversized messages can be refused with 552.

When bounces are really blocks

If a large share of bounces come from one provider with 5.7.x or 4.7.x codes, the addresses are probably fine and the problem is you. Typical triggers are a blocklisting, failed authentication or a reputation drop. Check your listings with the Blacklist Checker, your records with the Sender Reputation Checker, and follow the blacklist removal guide if you find a listing.

How to reduce your bounce rate

1. Validate addresses at sign-up

Add real-time checks to every form: correct syntax, a domain with working MX records, typo suggestions for common domains and, if it fits your audience, blocking of disposable addresses. Most bad data can be stopped before it enters your list.

2. Use confirmed opt-in

A confirmation click proves the address works and belongs to the person who signed up. It is the most effective single step to keep hard bounces near zero. See our double opt-in guide.

3. Verify lists before you send them

Any list that is imported, older than a few months or has not been mailed in a while should go through a verification service first. Dedicated verifiers such as ZeroBounce and NeverBounce are the most thorough, and SendPulse’s email verifier is a convenient option if you already send through that platform. Our verification tools comparison explains how they differ, and the catch-all verification guide covers the addresses no verifier can fully confirm.

4. Suppress hard bounces immediately

Configure your platform to add every hard-bounced address to a permanent suppression list on the first failure, so it can never be mailed or re-imported by accident. Most reputable ESPs do this by default, but check, especially if you send through your own SMTP server.

5. Retry soft bounces, then sunset them

Let the platform retry for a day or two. After three to five consecutive soft bounces across campaigns, suppress the address.

6. Re-engage or remove inactive subscribers

People who have not opened or clicked in six to twelve months are drifting toward abandoned accounts. Run a short re-engagement series and remove those who do not respond. The sunset policy guide has a ready-made flow.

7. Monitor by source and provider

Break bounce rates down by acquisition source (which form, which event, which partner) and by mailbox provider. A single bad source points to a form or data problem. A spike at one provider points to a block.

Triage: what to do when bounces spike

When a campaign comes back with an unusual bounce rate, work through these questions before the next send.

1. Hard or soft? Split the bounces by class. A spike in 4xx deferrals is a timing or reputation issue. A spike in 5xx rejections is a list or policy issue.
2. Address or policy? Among the hard bounces, separate unknown-user codes (5.1.x, 5.2.1) from policy codes (5.7.x, 554). Only the first group is about your data.
3. Where from? Group by mailbox provider. If one provider accounts for most of the spike, you are dealing with a block at that provider.
4. Which source? Group unknown-user bounces by acquisition source or import date. One bad source usually stands out.
5. What changed? Compare with the previous campaign: new segment, new template, new sending domain, higher volume, DNS changes.

Then act on what you find. Suppress bad addresses, pause the problem source, fix authentication or blocklist issues, and send the next campaign only to recently engaged subscribers until rates are back to normal. If the spike was a policy block, confirm your status with the Blacklist Checker and the Sender Reputation Checker before resuming full volume.

How bounces damage sender reputation

Mailbox providers read your bounce rate as a proxy for list quality. A high rate says your list was bought, scraped or never cleaned. That lowers your sender reputation, which pushes more mail to spam, which lowers engagement, which pushes reputation down further. The spiral is hard to stop once it starts.

There is also a spam trap link. Abandoned addresses that keep bouncing are sometimes recycled by providers into traps. If you never suppressed them, the next send hits a trap instead of bouncing, and trap hits are a common route onto blocklists.

The good news: bounce rate is one of the easiest reputation signals to control. Clean data in, automatic suppression, regular verification, and it stays low. For the broader picture, work through the deliverability playbook and the list hygiene guide.