Cold email bounce rate: what is a good number and how to fix a high one

Bounce rate is the share of sent emails that were not delivered. Google and Yahoo publish no exact bounce threshold, but both tell senders to remove invalid recipients, and Google says to reduce volume when messages start bouncing. Treat every hard bounce as an address to remove for good, verify lists before sending, and pause a campaign when bounces climb instead of sending through them.

If you search for "good bounce rate", you will find a confident figure in nearly every article, usually "under 2 percent". It is a rule of thumb that vendors repeat. This post sets out what the mailbox providers actually state, how to read a bounce, and the order in which to fix a high rate. Statements about Google, Yahoo and Microsoft come from their own pages, read on 25 October 2026.

What is a bounce rate, exactly?

Bounce rate is bounced emails divided by emails sent, over a period. Say how you count: some tools divide by sent, some by delivered plus bounced. Our glossary entry on bounce rate gives the definition. Two kinds matter:

Google's SMTP errors page puts it in these terms. For 550 5.1.1 it says the account you tried to reach does not exist, and to check the address for typos and not resend until corrected. For the temporary 421, 450 and 452 deferrals it says to resend at a later time.

What do Google, Yahoo and Microsoft actually say?

None of the three gives a bounce percentage that makes you compliant. What they say:

• Google. Its sender guidelines say to automatically unsubscribe recipients who have multiple bounced messages, and: "If messages start bouncing or start being deferred, reduce the sending volume until the SMTP error rate decreases." Its separate spam rate rule is strict: keep spam rates in Postmaster Tools below 0.10%, and avoid reaching 0.30%. Senders over 5,000 messages a day to Gmail have extra requirements, including SPF, DKIM and DMARC and one click unsubscribe.
• Yahoo. Its best practices say to monitor hard and soft bounces and inactive recipients, and to remove invalid recipients promptly. It says sending to users who are not reading or who report spam harms your delivery metrics and reputation, and it recommends confirmed opt-in to reduce invalid recipients.
• Microsoft. Its Exchange Online page on non delivery reports lists the codes, for example 5.1.1 "Bad destination mailbox address", which can mean the recipient's address does not exist in the destination system. Its Outlook.com requirements for high volume senders, in effect since 5 May 2025 for domains sending over 5,000 emails a day, concern authentication rather than bounces.

Read together: the providers do not set a pass mark for bounces. They tell you to clean your list and slow down when errors rise. That is the standard to meet.

How do you read a bounce code?

A bounce message has a basic code (550, 421) and an enhanced code (5.1.1, 5.7.1). The first digit tells you the class: 5 is permanent, 4 is temporary. The ones that matter most for cold email:

Codes starting 5.7 are about you, not the recipient. A list full of 5.1.1 codes is a data problem. A run of 5.7.x codes is a sender problem. Mixing them up leads to the wrong fix.

A step by step fix for a high bounce rate

Step 1. Pause the campaign. Google's advice is to reduce volume until errors fall. Do not send through it.

Step 2. Export the bounces and sort by code. Count how many are 5.1.1 style (bad address), how many are temporary, and how many are 5.7 style (blocks).

Step 3. Remove every hard bounce permanently. Put them on the suppression list so they cannot be re-imported. A bounced address that returns in next month's import will bounce again.

Step 4. Find the source of the bad addresses. If they came from one import, one scraped list or one old file, that source is the problem. Check for typos in a column mapping. If you guessed addresses from a pattern such as first.last@, that is a likely cause.

Step 5. Verify the rest of the list. Run it through a verifier before the next send. Our post on email verification before send explains what a verifier can and cannot tell you, including why catch all domains stay risky.

Step 6. Check authentication. If you saw 5.7 codes, confirm SPF, DKIM and DMARC for the sending domain. Our post on Google and Yahoo bulk sender rules lists them.

Step 7. Restart small. Send a few dozen emails a day from each mailbox and raise slowly. Sellinger's email agent defaults to 50 sends a day per mailbox, with an optional warm up that starts at 5 a day and doubles every 3 days up to the limit. Our post on why warm up alone is not enough explains why volume is only part of reputation.

Step 8. Write a stop rule. For example (made up): "If more than a handful of a day's sends hard bounce, or any 5.7 block appears, pause the mailbox and review." Choose your own numbers and write them down before the next campaign.

A worked example (made up)

A team imports 1,000 contacts from an old event list and sends 200 emails on day one. The tool reports 38 bounces. Sorting them by code shows 33 are 550 5.1.1, 3 are mailbox full deferrals and 2 are 5.7.1 notices. The team pauses, removes the 33 permanent failures, retries the 3 deferrals in two days, and reads the 2 blocks, which both point at content with a tracking link. Then it verifies the remaining 800 addresses before sending the next batch. The point is the sorting: three different causes needed three different fixes.

How does Sellinger handle bounces?

From our fact register: the email agent checks the team's suppression list before every send, and it treats bounce notices, auto replies and similar automatic mail as machine mail, so they never count as a human reply and the agent does not answer them. Enrichment finds verified emails and phone numbers from a database of 700M+ contacts. None of this can guarantee a bounce rate, because that depends on your list and your domain.

The takeaway is the same as the providers': keep the list clean, slow down when errors rise, and fix the cause rather than the symptom.