Cold email bounce rate: what is normal and 7 fixes

A good cold email bounce rate is below 2% of the emails you send. Between 2% and 5%, clean your list before the next send. At 5% or above, stop sending until you find the cause. Most bounces come from old or unchecked addresses, so the fix usually starts with the list, not the copy.

One warning about this topic: most of the numbers come from companies that sell email tools, and that includes several sources below. The rules from mailbox providers and sending platforms carry more weight, so this post uses both and says which is which.

What is a normal cold email bounce rate

Three large datasets give three answers. Belkins counted a 1.71% bounce rate across 7.5 million cold emails sent in 2025. Smartlead, with more than 850 million sends, puts the median sender at 1.54%, the top quarter below 0.73% and the 75th percentile at 3.21%. Hunter reports an average of 3.6% across 31 million emails sent in 2025 and says that unverified contact information is the typical cause.

The danger lines agree more than the averages do. Amazon SES, a sending platform, tells its customers to keep the bounce rate below 2%, reviews an account at 5% and might pause sending at 10%. Belkins writes that industry guidance treats below 2% as acceptable and under 5% as manageable. You may not send through Amazon SES, but its lines are useful because a platform published them for its own customers. Read together, they give a simple scale. The scale is one reading of these sources, not a published standard:

• below 2%: healthy, keep sending
• 2% to 5%: clean the list before the next send
• 5% or above: stop and find the cause

Your market also moves the number. In the Belkins data, May peaked at 2.26% and October was lowest at 1.26%, while India reached 3.26% and Finland stayed at 0.71%. Compare your number with your own past sends first.

1. Count it the same way each time

Divide the bounced emails by the emails you sent, for each campaign. Do not divide by the people on your list, because follow-ups change that count. Our guide to a good cold email reply rate shows how the same problem distorts reply numbers.

Small sends swing a lot. Two bounces out of 50 emails is 2 / 50 = 4%, and one more is 6%. So read the number across your last few campaigns, and look at each bounce by name.

2. Read the bounce message

Every bounce says why it happened, and the reason decides the fix. Mailchimp defines a hard bounce as a permanent reason, such as an address that does not exist, and a soft bounce as a temporary issue. Google publishes its full list of SMTP errors. Three of those errors are common in cold email:

• 550 5.1.1: the account does not exist, so remove the address
• 550 5.2.1: the account is inactive, so remove the address
• 450 4.2.1: the person gets email too quickly, so try again later

A code that starts with 5 is usually permanent, and a code that starts with 4 is usually temporary. If many messages fail with a reason about your domain or your reputation, the list is not the problem. Our guide on how to keep cold emails out of spam covers that case.

3. Remove hard bounces the same day

An address that does not exist will bounce again in your next follow-up, so one bad address can count several times. Google tells senders to automatically unsubscribe recipients who have multiple bounced messages. Mailchimp goes further and removes most hard-bounced addresses from an audience immediately. If you send from a normal mailbox, nothing does this for you. Stop the sequence for that person and mark the address as dead in your sheet, so it does not come back in a later list.

4. Put a limit on soft bounces

A full inbox or a busy server can fix itself, so one soft bounce is no reason to remove someone. But an address that keeps failing is not coming back. Mailchimp allows 7 soft bounces for an address with no subscriber activity before it treats the address as a hard bounce. A newsletter with years of history can afford that patience. A founder with a small cold list cannot, so set a much lower limit and remove the address after a few failed attempts.

5. Check addresses when you send, not when you build the list

People change jobs, and companies close. Smartlead says that contact data decays at 25 to 30% a year and tells senders to check again any address not verified in the last 90 days. Smartlead does not show a study behind that decay figure, so treat it as an estimate. The direction is the useful part: a list that was clean in spring is not clean in autumn.

LitePal gives a founder the decision-maker to write to with a checked email for each company that fits, and the founder sends from their own account. Whatever tool you use, run the check close to the send date. Our steps for finding the decision maker at a company show how to do the search by hand.

6. Treat accept-all domains as a separate list

A check cannot confirm every address. Hunter explains that an accept-all domain accepts email even for mailboxes that do not exist, so no verification tool can fully confirm delivery. A later bounce from one of these addresses looks like a failure of your check, but the check could not know. Smartlead reports that some lists heavy with these addresses produce bounce rates of 8% or more.

Keep these addresses apart from the confirmed ones. Send to the confirmed list first. Then send to the accept-all list in small batches, and only to people you have strong reason to believe work there, such as someone whose address appears on a public page.

7. Slow down when bounces start

Sending more does not get you past bad data. Google's advice is to reduce the sending volume when messages start to bounce or get deferred, and then increase slowly. Hunter also names aggressive sending rates as a cause of bounces. So when your number passes 2%, pause the campaign, read the messages, fix the cause and then send a small batch. Do this for each mailbox if you send from more than one, because an average across mailboxes can hide the one with the problem.

A worked example: an old list of 500 addresses

Take a founder with 500 addresses collected six months ago. This example assumes that the 25 to 30% yearly decay above is correct and spreads evenly across the year. Here is the arithmetic:

• six months of decay: 12.5% to 15% of the list
• stale addresses: 500 x 0.125 = about 63, up to 500 x 0.15 = 75
• if all of those bounce: 63 / 500 = about 13%, above the 10% line of Amazon SES
• the 2% line for a clean send of 425 addresses: 425 x 0.02 = about 8 bounces

Not every stale address bounces, so the first send would likely land below the worst case. It still shows the cost. Removing 75 dead addresses loses nothing, because a dead address never replies. At the 3% reply rate Hunter reports for sales outreach, the 425 live addresses still give 425 x 0.03 = about 13 replies.

Paste your site into LitePal and get companies that fit, the person to write to at each with a checked email, and a first line drafted for every one.