The newsletter went out on Tuesday with a 6% bounce rate. Nobody panicked. It was an old list, a few hundred dead addresses, a rounding error in the monthly report.
On Thursday the sales team noticed that nobody was replying to their follow-ups. On Friday a customer called to ask why the invoice had landed in spam.
Those look like three separate problems. They are one problem, and it started with the bounces.
A bounce is a message about you
When an email hard bounces, the receiving server is saying that the mailbox does not exist. Gmail, Outlook and Yahoo do not just deliver that message back to you. They remember who sent it.
A sender whose lists are full of dead addresses looks, from the outside, exactly like a sender whose lists were bought, scraped or never cleaned. That is the profile of a spammer, and filters are built to treat it that way. The judgment lands on your sending domain and the IP addresses that send for it, not on the campaign that caused it.
How the damage spreads
Bounces rarely hurt in one step. They start a chain:
- Bounces climb. Your domain's reputation with the big mailbox providers drops.
- Filtering tightens. Not for one campaign, for everything your domain sends: newsletters, sales outreach, invoices, password resets.
- Engagement falls. Mail that lands in spam is not opened, and low engagement lowers your reputation again.
- Your sending platform steps in. Amazon SES, for example, publishes its limits: keep bounces below 2%, the account is placed under review at 5%, and sending may be paused at 10%. Other platforms have similar rules, often unpublished.
By the time you see the last step, the first three have been happening for a while.
What providers actually measure
No mailbox provider publishes its formula, but the inputs are well known, and Gmail shows most of them back to you in Google Postmaster Tools.
| Signal | What a healthy sender looks like |
|---|---|
| Hard bounce rate | Under 2% on every send, ideally under 1% |
| Spam complaint rate | Under 0.1%, never close to 0.3% |
| Spam trap hits | None. One recycled trap is a warning; a pristine trap is worse |
| Authentication | SPF, DKIM and DMARC passing and aligned on every message |
| Engagement | Opens, replies and rescues from spam, measured against deletes and ignores |
| Volume pattern | Steady. A quiet domain that suddenly sends 50,000 messages looks bought |
Bounces sit at the top of that list because they are the cheapest signal a provider can measure and the hardest one for a sender to fake. Gmail condenses all of it into a domain reputation of High, Medium, Low or Bad. At Low, a noticeable share of your mail goes to the spam folder. At Bad, most of it is rejected outright or never shown, no matter how good the message is.
Reputation belongs to the domain, not to the campaign
There are two reputations in play, and they are not the same thing.
IP reputation follows the server that sends. On a shared pool, which is what most platforms give you by default, you inherit some of the behaviour of everyone else in the pool. On a dedicated IP, every bounce is yours alone, which is better when you send well and unforgiving when you do not.
Domain reputation follows the From domain, and it is the one that matters most now. It travels with you when you change platform, so switching sending tools after a bad month changes nothing.
Subdomains are the practical lever. Sending marketing from news.example.com and receipts from mail.example.com keeps one stream from dragging the other down, though a subdomain still inherits part of the parent domain's standing and needs its own warm-up. What does not work is starting again on a fresh domain every time reputation drops. Providers have seen that pattern for twenty years, it has a name in the filtering world, and a brand new domain begins with no history and gets treated cautiously anyway.
The mail you care about most goes down with it
Marketing email is the usual suspect, but it shares a domain with everything else. When reputation drops, the order confirmation, the password reset and the invoice reminder are filtered with the same suspicion as the promotion that caused the bounces.
That is where a deliverability problem becomes a support queue: customers who "never got the email", payments that arrive late, trials that never activate because the welcome message sat in spam.
The cost nobody budgets for
A bounce also costs money directly. Most sending platforms price by contacts or by messages, so every dead address on the list is paid for, every month, to deliver nothing.
The larger cost is time. You cannot undo a bounce. A damaged reputation rebuilds only as providers watch you send carefully to people who want your mail, which usually means weeks of smaller, more cautious sends while the pipeline waits.
The fix: stop mailing addresses you have not checked
Prevention is cheap compared with repair, and it happens at two moments.
- When an address is collected. A real-time check on signup, checkout and lead forms rejects typos, disposable inboxes and domains that cannot receive mail before they ever enter your lists.
- Before a list is mailed. Anything that has not been checked for a few months should be verified in bulk before the next send. People change jobs, companies close mailboxes, and a list decays while it sits.
Two rules keep the gains:
- Suppress hard bounces forever. An address that hard bounced once will bounce again.
- Treat unresolved addresses as risk, not as valid. A server that accepts every address proves nothing about a specific mailbox. Catch-all domains are the common case.
If you want to see what a check actually inspects, how email verification works follows one address through every stage.
Signs your reputation is already slipping
The support call is the last symptom, not the first. These come earlier:
- Open rates fall at one provider only. Gmail usually moves first. If Gmail opens drop while Outlook holds steady, that is filtering, not content.
- Postmaster Tools moves from High to Medium. The drop is gradual and the recovery is slower.
- More deferrals. Bounce logs fill with
4.7.xtemporary rejections, and messages take hours to arrive instead of seconds. - Your own test sends land in spam or in a tab you did not expect, on accounts that used to receive them fine.
- Unsubscribes and complaints rise on the same content that was fine last month, because the mail now looks unfamiliar to people who stopped seeing it.
Any one of these is worth checking the bounce report. Two at once means it has already started.
How long repair takes
Longer than the damage. Reputation is rebuilt by sending well, which means you need weeks of sends that providers can watch.
- Week one. Stop mailing anything unverified. Suppress every hard bounce you have ever had. Fix authentication first, because nothing else is judged fairly until SPF, DKIM and DMARC pass.
- Weeks two and three. Mail only people who opened or clicked in the last 30 to 90 days, at a steady daily volume. Small and boring is the point.
- Weeks four to six. Widen the audience a step at a time, watching the complaint rate rather than the open rate.
There is no way to buy that time back, which is the real argument for prevention. A list check costs a fraction of a cent per address. Six weeks of throttled sending costs a quarter of a pipeline.
What to do this week
- Open your sending platform and look at the bounce rate for the last 30 days, per campaign.
- Verify the next list you plan to send to, before it goes out.
- Add a verification call to your signup form, so new bad addresses stop arriving.
- Make sure hard bounces are suppressed automatically, not removed by hand when someone remembers.
Protecting a sender reputation is mostly about not giving it away. Bulk verification cleans the list you already have, and the verification API keeps new bad addresses out of it.
