Yesterday’s campaign bounced like it always does. Today’s campaign bounced enough that you pasted the rate into Slack. Your creative did not do that. Something in who you mailed, or who you mailed through, changed.
Find that change before you delete half the list. A bounce spike is a sorting problem: dead addresses, a temporary deferral, or a rejection of you.
Last verified: 26 September 2026, against Klaviyo’s bounce documentation, RFC 5321, and the SES enforcement FAQ.
Sort the bounce rate spike before you touch the list
| What the bounce log shows | Likely cause | First move |
|---|---|---|
| 5xx, mailbox unknown, mostly on a segment you imported or collected this week | Bad addresses in a new source | Suppress those hard bounces. Stop that source. |
| 4xx, try again, rate, or greylist, clustered on one provider | Deferral, not dead mailboxes | Let retries finish. Do not ramp volume. |
| 5xx policy or authentication, on mail that worked last week | You, not the customer | Read SMTP 4xx vs 5xx. Fix the block before any suppression sweep. |
| Hard bounces spread across old engaged customers | The list did not rot overnight | Check whether the From domain, return path, or ESP changed |
Hard bounce vs soft bounce is the definition. Klaviyo already suppresses hard bounces. Your job in a spike is to stop the source that is creating new ones, not to re-suppress what Klaviyo has already held. The high bounce rate fix page has the step-by-step version if you want a checklist to work through.
The three sources that show up in your eCommerce account
A form or a file. Your checkout and popup fields will accept gmail.con all day if nothing validates them. A wholesale CSV from a partner is the same failure with a nicer filename. Purchased lists are the extreme case. A sloppy popup is the common case.
A provider throttle that your ESP calls a bounce. Gmail, Yahoo, and Outlook can defer or reject when your volume, complaints, or authentication change. The percentage looks like list decay. Your log shows one domain in the recipient column. If more than half of the new failures share one recipient domain, you are looking at a provider decision, not a rotten list. That split is a working heuristic, not a published rule, so read the SMTP text to confirm it. Decay does not start at 9 a.m. on a single provider.
A DNS or identity change. A new sending subdomain, a Klaviyo domain cutover, or an SES identity that is still unverified produces bounces that have nothing to do with your customer file. The domain change guide and the migration guide cover those cutovers. If the spike started in the hour DNS changed, stay on that trail.
What not to do when bounce rate spikes
Do not mail the bouncing segment again “to see if it clears.” A second pass at unknown mailboxes is how your list problem becomes a reputation problem. SES recommends a hard-bounce rate under 2%, can put your account under review at 5%, and can pause sending at 10%. Those thresholds are on the SES page, and they are SES rules, not a universal bounce law.
Do not treat Klaviyo’s in-product bounce colors as Gmail’s decision. Gmail’s published figure for bulk senders is a spam complaint rate, measured on mail that was delivered. Bounces and complaints move for different reasons. When both rise together, handle them as two incidents that share a clock, using bounce rate vs complaint rate.
Do not wait for a blocklist to confirm the spike. Lists are a lagging signal. The SMTP text is the current one.
InboxEagle
Measure the spike against your normal rate
Enter sends, hard bounces, and soft bounces for the spiking campaign and the one before it. If the hard-bounce side moved, the list or the source changed. If only soft bounces moved, look for a deferral.
Sources
- Understanding bounced emails, Klaviyo. How hard and soft bounces are classified and suppressed.
- SMTP reply codes, RFC 5321. Permanent versus transient failures.
- SES enforcement FAQ, AWS. How SES calculates bounce rate and when it reviews or pauses an account.