When to pause a cold email mailbox: bounces and complaints

You have warmed up the mailboxes and set daily limits. Now campaigns are running and the numbers are coming in. The question is when a number means stop.

This post gives one published threshold, Google's spam rate line. It does not invent cutoffs for bounces or for pausing a mailbox. For those, it gives a method: pick the denominator, read each mailbox and each domain on its own, and decide in advance what you do at each level.

It follows the guides on daily limits per mailbox and on cold email deliverability. Read those first if you have not set up domains and warm-up.

What changed

This section lists the published sender rules the rest of the post relies on. Most are standing guidelines. Outlook's requirement is the one described as coming.

Spam complaints. Google says to "Keep spam rates reported in Postmaster Tools below 0.3%". [[1]](#sources) That is the one rate in this post that comes from a mailbox provider. It is a ceiling, not a target. If you wait until your reported spam rate reaches it before acting, you have waited too long. Set your own pause trigger below that line. The spam complaint rate glossary entry explains how the rate is calculated.

Authentication. Yahoo's sender best practices say to "Publish a valid DMARC policy with at least p=none". [[2]](#sources) Microsoft says "For domains sending over 5,000 emails per day, Outlook will soon require compliance with SPF, DKIM, DMARC." [[3]](#sources) A domain with weak authentication gives you a problem to chase that has nothing to do with your list or your copy.

SPF limits. RFC 7208 says "SPF implementations MUST limit the total number of those terms to 10 during SPF evaluation". [[4]](#sources) If you keep adding sending tools to one domain's SPF record, you can pass that limit without noticing. Check the record again after each change.

Sender identity. The FTC says "Your message must include your valid physical postal address." [[5]](#sources) See the CAN-SPAM entry for the rest of the rule.

Pick the denominator before you read any rate

A rate is only useful when you know what it is divided by.

Two published reply benchmarks show the problem. Belkins found that 0.45% of emails get a reply across industries, where reply rate is replies divided by emails sent. [[6]](#sources) Smartlead puts the median sender at 0.74% of contacts replying. [[7]](#sources) The figures look comparable, but one counts emails and the other counts contacts. They are not the same measure.

Bounces and complaints have the same trap. Decide once, and write down, whether your bounce rate is bounces divided by emails sent or by contacts in the campaign. Use the same definition for every mailbox and every domain. A sequence with several steps sends more emails than it has contacts, so the two definitions drift apart as a campaign ages.

How to read the numbers per mailbox and per domain

Read both levels. They answer different questions.

Per mailbox tells you which sender is in trouble. If one mailbox shows a much higher bounce rate than the others on the same domain, look at what is different about it. It may have been given a different list, a different campaign or a different start date.

Per domain tells you whether the problem is shared. If all mailboxes on a domain move together, the cause is probably the list, the copy or the domain's authentication. It is less likely to be one mailbox.

The domain counts are small. Builder includes 1 domain and 2 mailboxes. Growth includes 2 domains and 4 mailboxes. [[8]](#sources) With so few mailboxes per domain, one bad mailbox can drag the domain's picture. Always look at both numbers together.

What Outlevo shows you comes from the sending tool. Smartlead reports sent, reply, bounce, unsubscribe and category-change events back to Outlevo by webhook. [[9]](#sources) Outlevo also reads the mailboxes in your Smartlead account together with their warm-up state. [[9]](#sources) For managed mailboxes, Outlevo checks each mailbox's warm-up state and reputation daily. [[8]](#sources)

If you want to script your own checks, the public API has resources for mailboxes, sender-domains, suppressions and sending. [[10]](#sources)

Small samples mislead. A mailbox that has sent a handful of emails can show a dramatic rate from one bounce. Set a minimum number of sends before a rate counts, and write that minimum down with your triggers.

What to do

Set four levels, in this order. For each, write down your own number, using your baseline from the first weeks of healthy sending, and the action. Do this before the campaign starts, not while a rate is climbing.

Level 1: slow down

When a mailbox's bounce rate moves above its usual range, cut its daily volume before changing anything else. This costs little and gives you time to look. The daily limit guide covers how to set and ramp volume. Do not raise volume again until the rate has settled.

Level 2: clean the list

Start with the list. Verify addresses before they go into a campaign, and re-verify any list that has been sitting for a while.

In Outlevo, email verification costs 1 credit per address. [[11]](#sources) See the email verification page for what it covers. If you prefer a separate tool, ZeroBounce positions itself as email verification and deliverability: "verify emails, clean your list, build your sender reputation, and monitor inbox health". [[12]](#sources) Its Pay as You Go option starts at $39 for 2,000 email validations. [[13]](#sources) Its Freemium plan is $0 a month with 100 validation credits. [[14]](#sources) Prices were read on October 26, 2026. See the ZeroBounce pricing page for the current quotes.

After cleaning, add the bounced addresses to your suppression list so they are never emailed again.

Level 3: pause

Pause when the rate stays above your trigger after you have slowed down and cleaned the list, or when your complaint rate is approaching Google's 0.3% line. [[1]](#sources)

Outlevo's kill switches pause a campaign, a channel or the whole workspace at once. [[15]](#sources) Use the smallest one that covers the problem. If one campaign is the cause, pause that campaign. If the whole email channel is affected, pause the channel. Reserve the workspace switch for the case where you do not yet know the cause.

Policies also matter here. They set who can be contacted, on which channels, how often and with what budget, and every external action is checked against the approved policy before it runs. [[15]](#sources) If you change a limit while troubleshooting, change the approved policy, not just the campaign.

Telegram alerts cover hot leads, approvals, problems that need you and credit alerts. [[16]](#sources)

Level 4: investigate content and setup

If the list is clean and the rate is still high, look at what you are sending and how the domain is set up.

Authentication. Before a campaign sends through Smartlead, every mailbox must be connected and warmed up, and its domain must have SPF, DKIM and DMARC checked in the last 7 days. [[9]](#sources) If you are past that window or have changed DNS since, check again. Count the lookups in the SPF record against the limit of 10. [[4]](#sources)
Warm-up. A new managed mailbox sends nothing until it is warm: at least 14 days, usually about 3 weeks. [[8]](#sources)
Message. Check that the email includes your valid physical postal address. [[5]](#sources) Read the first lines as the recipient would. Ask whether they would know why they got it.
Targeting. A high complaint rate can mean the list does not fit the offer. The free ICP assessment is one way to check who you are writing to.

Change one thing at a time and restart at a lower volume. If you change the list, the copy and the DNS together, you will not know which one fixed it.

A one-page checklist

Before the next campaign, write down:

Your bounce rate definition and its denominator.
The minimum number of sends before a rate counts.
Your own trigger for each level, per mailbox and per domain.
A complaint trigger set below Google's 0.3%. [[1]](#sources)
Who pauses, and with which switch.
What has to be true before volume goes back up.

The numbers matter less than having them agreed before you need them.