Soft Bounce vs Hard Bounce: What to Do About Each
Direct answer: A soft bounce is a temporary delivery failure, the mailbox exists and the domain is reachable, but something blocks delivery right now (full inbox, oversized message, a brief server problem), and most mail servers keep retrying for 24 to 72 hours before giving up. A hard bounce is permanent: the address or domain does not exist, so retrying never helps and the address should be removed immediately. The distinction matters because your sending reputation is scored almost entirely on the hard bounce number. A soft bounce that resolves itself costs you nothing; a soft bounce that never resolves and keeps getting retried, or a hard bounce that stays on your list, actively damages deliverability for every email you send afterward.
Table of contents
- Soft bounce vs hard bounce: the actual difference
- SMTP bounce codes, decoded
- How Gmail, Mailchimp, HubSpot, and Klaviyo each handle a soft bounce
- The bounce-rate thresholds you cannot cross in 2026
- Benchmarks: what a normal bounce rate looks like by industry
- Why email verification tools don’t fully solve this
- Bounce myths vs what the data supports
- A checklist to fix a bounce problem
- FAQ
Soft bounce vs hard bounce: the actual difference
A soft bounce means the receiving mail server was reachable and the address looked plausible, but delivery failed for a reason that could resolve on its own: the recipient’s inbox is full, the message is too large, the server is temporarily overloaded, or a spam filter held the message for review. Soft bounces are represented by SMTP codes in the 4XX range, and most mail servers (yours and the recipient’s) treat them as worth retrying.
A hard bounce means the message could not be delivered now, and will never be deliverable in the future without a change on the recipient’s side: the mailbox doesn’t exist, the domain doesn’t exist, or the server has permanently rejected all mail to that address. Hard bounces use SMTP codes in the 5XX range. There is no retry schedule that fixes a hard bounce, because the problem isn’t timing, it’s that the destination is gone.
The practical difference for you as a sender: a soft bounce is a data point to watch, a hard bounce is an action item. An address that hard bounces should be suppressed from your list immediately, on the first occurrence, not after a grace period. Continuing to send to a hard-bounced address doesn’t just waste a send, it actively signals to Gmail and Outlook that you’re not maintaining your list, which drags down deliverability for every other email in the same campaign.
SMTP bounce codes, decoded
Every bounce arrives with an SMTP status code, and reading it tells you immediately which category you’re in and whether it needs action.
| Code | Type | Meaning | Action |
|---|---|---|---|
| 421 | Soft | Server temporarily unavailable | Let it retry automatically |
| 450 / 451 | Soft | Mailbox temporarily unavailable, or a greylisting delay | Let it retry automatically |
| 452 | Soft | Mailbox full or over quota | Retry; suppress if it repeats 3-5 times |
| 550 5.1.1 | Hard | User unknown, mailbox does not exist | Suppress immediately |
| 551 / 552 | Hard/Soft | User not local, or exceeded storage allocation | Treat 551 as hard, 552 as soft |
| 553 | Hard | Mailbox name not allowed (malformed address) | Suppress immediately |
| 554 | Hard | Transaction failed, often a policy or blocklist rejection | Suppress and investigate reputation |
The single rule worth memorizing: 4XX is temporary, 5XX is permanent. If you only remember one thing from this table, that split tells you whether to wait or to act.
How Gmail, Mailchimp, HubSpot, and Klaviyo each handle a soft bounce
Every mail system retries a soft bounce, but the retry window and the giving-up point differ by platform, and knowing your platform’s rule tells you how long to wait before treating a stuck soft bounce as a real problem. Most SMTP servers, Gmail and Outlook included, keep retrying a soft-bounced message for roughly 24 to 72 hours before the message finally fails for good.
| Platform | Soft bounce handling |
|---|---|
| Mailchimp | Converts repeated soft bounces to inactive after 7 attempts for unengaged contacts, 15 attempts for engaged ones |
| HubSpot | Similar repeat-attempt logic, adjusted by the specific 4XX code returned each time |
| Klaviyo | Weights by engagement; addresses with recent opens are retried longer than cold, unengaged addresses |
| General rule of thumb | 3 to 5 consecutive soft bounces on the same address, across separate sends, should be reclassified and treated as a hard bounce |
That last row is the one most senders miss. A single soft bounce is noise. Five soft bounces in a row on the same address, across five separate campaigns, is not a temporary problem anymore, it’s a dead address wearing a temporary-failure code, and it should be suppressed exactly like a hard bounce would be.
The bounce-rate thresholds you cannot cross in 2026
Since Gmail and Yahoo enacted their bulk-sender rules in February 2024, bounce rate stopped being a soft metric you glance at occasionally and became a hard gate that decides whether your mail gets delivered at all. The requirements apply to anyone sending more than 5,000 messages a day to Gmail or Yahoo addresses: a spam complaint rate under 0.30 percent, a bounce rate trending below 1 percent, and DMARC enforcement set to at least p=quarantine. The hard-enforced number in those rules is the spam complaint rate, not a separate bounce ceiling (the widely cited 0.35 percent figure is Klaviyo’s internal guidance for when to intervene on bounces, not a Gmail or Yahoo rule), and most major ESPs (Mailchimp, Klaviyo, ActiveCampaign, SendGrid among them) will pause or suspend your sending if a single campaign’s hard bounce rate exceeds 2 percent, specifically to protect their shared sending infrastructure from your list problem becoming their reputation problem.
| Total bounce rate | Status | What it means |
|---|---|---|
| Under 2% | Safe | Normal, no action needed beyond routine hygiene |
| 2% to 5% | Warning | List hygiene and re-verification needed before your next send |
| Above 5% | Critical | Active sender-reputation damage; pause sending and clean the list first |
Benchmarks: what a normal bounce rate looks like by industry
An analysis of Mailchimp send data found an average hard bounce rate of 0.21 percent and an average soft bounce rate of 0.70 percent across billions of sent emails, meaning well-maintained lists sit under 1 percent total bounces almost by default. The broader 2026 cross-industry average bounce rate is closer to 2.48 percent, but that average hides a real spread: IT and technology senders run leaner, closer to 0.90 percent, while construction and manufacturing senders, industries with older, less frequently updated contact data, run closer to 2.20 percent. If your bounce rate is meaningfully above your industry’s benchmark, the fastest diagnosis is your data source, not your email copy.
Why email verification tools don’t fully solve this
Running a list through a verification tool before sending is standard practice, and it genuinely helps, but reviewers of the category consistently flag a specific blind spot: catch-all domains. A catch-all mail server accepts every address thrown at it regardless of whether a real mailbox exists behind it, which means a verifier can only guess at deliverability rather than confirm it, and that guess is where the false positives live.
ZeroBounce carries the largest footprint in the category, rated 4.7 out of 5 across roughly 1,361 G2 reviews, but a February 2026 benchmark test published by Allegrow, a competing verification vendor (so read it with that in mind), ran a controlled set of real and fabricated contacts through it and found its standard verification tier returned an 8.1 percent false-positive rate, marking clearly fictional addresses as valid specifically because the domains tested were catch-all. Its higher tier, marketed as Verify+, caught 96 percent of genuinely real contacts but pushed the false-positive rate up to 33.1 percent in the same test, trading missed real contacts for a much larger share of confidently wrong “valid” labels.
An 8.1% false-positive rate on standard verification, rising to 33.1% on the higher-accuracy tier, both driven by the same catch-all domain blind spot
Allegrow benchmark test, February 2026, controlled real/fake contact set, cross-referenced with G2 review sentiment
NeverBounce, rated 4.2 out of 5 across 141 G2 reviews, draws a different complaint pattern: reviewers consistently describe its credit-based pricing as feeling “nickel-and-dimed” rather than flagging accuracy as the core issue, and independent testing puts its real-world accuracy in the 93 to 97 percent range. Between the two, the lesson holds regardless of which tool you pick: verification reduces bounce risk, it does not eliminate it, and a catch-all-heavy list will still produce bounces a verifier told you were safe. Budget for an ongoing bounce rate near, not at, zero.
Diagnose your bounce problem in three checks
Before applying the fix checklist, three quick checks tell you which category your problem actually falls into, since the fix for a data-quality problem is different from the fix for a reputation problem.
Check 1: is it concentrated in a few domains, or spread across your whole list? Pull your bounce report and group by recipient domain. If 80 percent of bounces trace back to two or three domains, you likely have a source-list problem (a specific data provider or list import), not a broad reputation problem. If bounces are evenly spread across hundreds of domains, that points toward a sending-reputation issue instead.
Check 2: is it mostly 4XX or mostly 5XX codes? A list dominated by 5XX hard bounces means the data itself is stale or was never accurate, point back to how and when the list was built. A list dominated by 4XX soft bounces that never resolve, even after 72 hours, often signals a reputation or authentication problem instead, since receiving servers sometimes use soft-bounce codes as a polite way of temporarily rejecting a sender they don’t yet trust.
Check 3: did the bounce rate spike suddenly, or has it been climbing gradually? A sudden spike usually traces to a single event, a new list import, a broken authentication record, or a jump in sending volume. A gradual climb over weeks or months is closer to natural list decay, the ordinary rate at which B2B contact data goes stale as people change jobs and roles, and it calls for a re-verification cycle rather than a one-time fix.
Why bounces compound instead of staying flat
The reason a 2 percent bounce rate this month can turn into a 5 percent bounce rate a few months later, with no change in your list, is that mailbox providers respond to reputation signals by throttling, not by rejecting outright at first. A slightly damaged sender reputation doesn’t just get more emails bounced, it gets more emails routed to spam or held for delayed delivery, which increases the odds that a legitimate address ends up producing a soft bounce it wouldn’t have produced from a trusted sender. That soft bounce, retried and failing again, nudges reputation down further, which increases throttling on the next send. The practical implication: bounce rate is easiest and cheapest to fix while it is still small, since a small elevated rate left alone tends to compound rather than plateau.
Bounce myths vs what the data supports
| Myth | What the data actually supports |
|---|---|
| A soft bounce is harmless, only hard bounces matter | A soft bounce that repeats 3-5 times on the same address behaves exactly like a hard bounce and should be suppressed the same way |
| Any bounce rate under 5% is fine | 2026 industry benchmarks put well-run lists under 1% total bounces; 2-5% is already a warning zone, not a safe zone |
| Email verification tools guarantee a clean send | Allegrow’s February 2026 benchmark found false-positive rates as high as 33.1% on catch-all-heavy lists, even from category-leading tools |
| Bounce rate is a vanity metric, not worth watching closely | Since Feb 2024, Gmail and Yahoo expect bulk senders to keep bounces trending below 1%, with a hard-enforced 0.3% spam complaint cap; sustained high bounces risk inbox placement for your entire domain |
| All industries should target the same bounce-rate benchmark | IT/tech senders average 0.90% while construction/manufacturing average 2.20%; benchmark against your own industry, not a blended average |
A worked example: is your bounce rate actually a problem?
Bounce rate math is simple, but the benchmark you compare it to changes the verdict entirely. Take a campaign of 10,000 sends with 250 total bounces, a flat 2.5 percent bounce rate. Against the blended 2026 cross-industry average of 2.48 percent, that reads as roughly normal. Against the IT/tech industry benchmark of 0.90 percent, the same 2.5 percent is nearly three times the expected rate for that sector, a real problem worth investigating rather than a rounding error. And against the below-1-percent trend line Gmail and Yahoo expect from bulk senders, that same campaign is already well past the point where inbox placement risk kicks in for the whole sending domain, regardless of what the blended average says.
The lesson is to always benchmark against the strictest relevant standard, your mailbox provider’s enforcement threshold, not the friendliest available average, since the friendliest average is the one most likely to tell you everything is fine when it isn’t. If the 250 bounces in that example break down as 220 hard (5XX) and 30 soft (4XX), the hard-bounce share alone, 2.2 percent, already exceeds most ESPs’ 2 percent single-campaign suspension trigger, meaning this specific campaign is a candidate for an account-level warning even before the broader bounce-rate conversation starts.
A checklist to fix a bounce problem
- Pull your last campaign’s bounce report and split it by SMTP code, not just a raw percentage.
- Suppress every 5XX (hard bounce) address immediately; do not resend to any of them.
- For repeated 4XX (soft bounce) addresses, apply the 3-5 rule: if the same address has soft bounced 3 to 5 times across separate sends, suppress it as if it were a hard bounce.
- Run the remaining list through a verification tool, but treat “valid” on a catch-all domain as a maybe, not a guarantee.
- Check your total bounce rate against your industry benchmark, not the blended cross-industry average, before deciding whether you actually have a problem.
- If your bounce rate is trending above 1%, pause volume growth until the list is clean; adding senders or volume on top of a dirty list compounds the reputation damage.
- Verify addresses at the point of capture going forward so this becomes a maintenance task, not a recurring cleanup project. A verified email finder that checks deliverability before an address ever enters a sequence prevents most of this problem at the source, rather than diagnosing it after a send.
Bounce management is really one piece of a larger deliverability picture that also covers authentication, list hygiene, and sending volume together, covered in full in our guide to why emails land in spam or promotions. A sales engagement platform that auto-pauses a sequence on repeated bounces, rather than relying on a rep to notice, closes the gap between having this checklist and actually running it every single send.
Verify before you send: try Salesgear free
FAQ
What is the difference between a soft bounce and a hard bounce?
A soft bounce is a temporary failure, the address is reachable but delivery failed for a reason that can resolve itself, like a full inbox. A hard bounce is permanent, the address or domain does not exist, so retrying never works and the address should be removed right away.
How long do mail servers retry a soft bounce?
Typically 24 to 72 hours before the server gives up and the message fails for good. Most major ESPs then apply their own escalation rule on top, converting repeated soft bounces to inactive or suppressed after a set number of attempts.
What bounce rate is considered safe in 2026?
Under 2% total bounce rate is considered safe, and well-maintained lists typically run under 1%. Gmail and Yahoo’s 2024 bulk-sender rules expect bounces trending below 1% (the hard-enforced number is a 0.3% spam complaint rate), and most ESPs suspend sending if a single campaign’s hard bounce rate exceeds 2%.
Should I remove an address after one soft bounce?
No, a single soft bounce is normal and usually resolves on retry. Only suppress it once the same address has soft bounced 3 to 5 times across separate sends, at which point it is behaving like a hard bounce and should be treated as one.
Do email verification tools eliminate bounces completely?
No. Verification tools reduce bounce risk but cannot fully solve catch-all domains, where a mail server accepts every address regardless of whether a real mailbox exists. A February 2026 Allegrow benchmark found false-positive rates as high as 33% on catch-all-heavy lists even with category-leading verification tools.
Why does bounce rate matter so much for deliverability?
Since February 2024, Gmail and Yahoo hold bulk senders to published thresholds: bounces trending below 1%, spam complaints under 0.3%, and full authentication. Crossing their thresholds can affect inbox placement for your entire sending domain, not just the individual campaign that triggered it.
Vinodh Kumar is the Co-Founder and Director of Engineering at Salesgear, where he leads engineering, platform architecture, and product development. He previously held engineering roles at Adapt.io, Owler, and Aricent, building scalable SaaS products and distributed systems. He writes about AI in sales, MCP, deliverability, and the engineering behind outbound at scale, with one focus: fixing outbound with better data, smarter automation, and real conversations.
