Every email platform has a bounce-rate policy. Almost none of them print the actual numbers, because the numbers invite gaming. But after enough deliverability post-mortems, the ladder is well known, and it's steeper than most senders expect.
First, the two kinds of bounce
A hard bounce is permanent: the mailbox doesn't exist, the domain is dead, the server says "never". A soft bounce is temporary: mailbox full, server busy, message too big. The ladder below is about hard bounces, the ones that say you don't know your own list. (Soft bounces matter too, but mostly as an early-warning light.)
The ladder
Under 2%: the green zone
This is normal decay territory. People change jobs, inboxes get abandoned; roughly 2–3% of a typical list dies every month. A freshly cleaned list bounces at a fraction of a percent; a list a few months old creeps toward 2%. Nobody bothers you here.
2–5%: the quiet penalties begin
Cross ~2% and two things start happening, neither of them announced:
- Your ESP notices. Automated warnings, then throttled sending, then a friendly-but-firm email from their compliance team. Shared-IP platforms are strict because your bounces poison the IP their other customers send from.
- Mailbox providers notice. Gmail and Outlook read persistent bounce generation as list-quality failure. Your deliverable mail starts landing in spam for everyone else on the list. It's the invisible penalty that makes people say "our open rates just died and we don't know why".
5%+: the cliff
At five percent, one send in twenty is hitting a dead address. In practice:
- Account review or suspension. Most major ESPs pause campaigns and demand an explanation (and usually proof of cleaning) somewhere between 4% and 8%. You lose sending ability at exactly the moment you had something to say.
- Blocklist exposure. Sustained bounce storms against providers' trap networks and dead domains are a classic path onto Spamhaus and friends, and a blocklisting outlives the campaign that caused it by weeks.
- Reputation debt. Even after you clean up, providers remember. Recovery from a 5% event takes consistent good behaviour over multiple sends. Think weeks, not days. (More on that mechanism in Reputation is a bank account.)
The arithmetic that surprises people: a list of 20,000 that's sat untouched for one year has accumulated roughly 4,800–7,200 dead addresses at normal decay rates. Sending to it cold isn't a 2% risk; it's a 25%+ bounce event, the kind that ends ESP relationships in one afternoon.
Why "remove bounces after the send" is the losing strategy
Plenty of teams treat bounces as self-cleaning: send, let the dead addresses bounce, let the ESP suppress them. The problem is sequencing: the reputational damage happens during the send. You paid the penalty to discover what a verification pass would have told you beforehand, privately, for cents an address. It's the difference between a blood test and exploratory surgery.
What to actually do
- Verify before every meaningful send, and on a schedule between them. Quarterly is the rule of thumb, with exceptions.
- Verify at capture so typos and fakes never enter the list at all.
- Watch soft bounces: a rising soft-bounce trend on a domain often precedes the hard-bounce wave.
- Keep the receipts. If your ESP ever asks, "we clean quarterly with a verification service, here's the report" ends the conversation.
bounces are the only deliverability number you can fix before it happens ✦
The free evaluation checks a sample of your list end to end and tells you straight whether it needs cleaning. Verdicts in plain English, about ten minutes, no card.
Evaluate my list free →