Warmup vs Sending Limits: What Actually Caps Your Volume

Three different ceilings sit above every mailbox: the provider’s hard limit, your warmup ramp, and your campaign’s own pacing. Only the lowest of the three is doing anything. Most senders tune the wrong one — raising a campaign setting that was never binding, while the real cap sits in a provider document they have not read.

Key takeaways

• Provider limits are hard and unmovable. Google Workspace allows 2,000 messages per user per day; Microsoft enforces 30 messages per minute per mailbox.
• The recipient count often binds before the message count — Google Workspace permits only 100 recipients per message over SMTP.
• Warmup is a ramp, not a licence. Finishing warmup does not raise a provider cap; it establishes sending history under one.
• Microsoft’s per-minute throttle produces no error at all — just delay. Absence of bounces is not evidence of health.
• The right lever is more mailboxes, never a higher per-mailbox number.

Which ceiling is actually binding?

Work down the list and find the smallest number.

1. The provider’s published limit. Fixed by the mailbox host, enforced at the server, unaffected by anything you do in a sending tool.

2. Your warmup ramp. A deliberately low, gradually increasing volume you run so the mailbox builds history. Always below the provider limit, by design.

3. Your campaign pacing. Daily send caps, sending windows, and per-mailbox rates you configure.

4. Reputation-driven throttling. Not a number at all — a provider slowing or deferring you based on how your mail is being received.

Raising #3 when #1 is binding changes nothing except how quickly you hit an error. Raising #3 when #2 is binding does nothing either, because the ramp is the constraint. Knowing which one you are actually against is the whole exercise.

What are the real provider numbers?

Google Workspace publishes exact per-user figures on a rolling 24 hours: 2,000 messages per day on paid accounts (500 on trial accounts), 10,000 total recipients per day, 3,000 unique recipients per day, and — the one that catches people — 100 recipients per message when sending over SMTP, POP or IMAP, against 500 via the API. Exceeding the caps suspends the account for up to 24 hours. Google’s sender guidelines carry the requirements that apply at any volume; the limits themselves sit in Workspace’s own documentation.

Microsoft 365 / Exchange Online enforces two independent limits per mailbox: 10,000 recipients per day on a 24-hour sliding window, and 30 messages per minute. Microsoft’s outbound sending guidance states plainly, “don’t exceed 30 messages per minute per mailbox,” and advises spreading bulk sends over time rather than sending all at once. It also states — worth reading twice if you have built on Exchange — that “Exchange Online isn’t designed for bulk mailing scenarios.”

There is also a tenant-level cap on external recipients per day, scaled by licence count, and reduced for new tenants during their first weeks. When a tenant exceeds it, every user in the tenant is blocked from sending externally, not just the offending mailbox.

Yahoo publishes no numeric per-account limit at all. Its best practices advise limiting messages per SMTP connection and opening concurrent connections respectfully, without specifying numbers — treat Yahoo’s limits as adaptive and reputation-driven.

Smaller hosts vary enormously, and several enforce hourly rather than daily caps, with much lower numbers on trial plans. Some general-purpose hosting control panels ship with the hourly limit set to unlimited at the software level, meaning the number you actually get is whatever your host configured — there is no safe default to assume. If you are on one, the practical approach is to discover the limit from rejections rather than guess it.

What does warmup actually change?

Not the ceiling. Warmup changes your position under the ceiling and the history you carry into a send.

Google’s guidance is explicit about the shape: “start with a low sending volume to engaged users, and slowly increase the volume over time”; “the more email that you send, the more slowly you should increase sending volume”; and, critically, “immediately doubling previously sent volumes suddenly could result in rate limiting or reputation drops.” Doubling is the named failure mode. Google also notes that daily sending permits a faster ramp than weekly sending, and that after a format or infrastructure change you should ramp the changed segment separately rather than swinging total volume.

What Google does not publish is a day-by-day schedule — and neither does Microsoft or Yahoo. Any calendar table promising “day 4: 12 emails” is tooling-industry invention.

So: a fully warmed mailbox on Google Workspace still has a 2,000/day ceiling and a 100-recipient-per-message SMTP limit. Warmup bought you history and a smoother ramp, not headroom.

Why do I get no error when I hit a limit?

Because not every limit produces one, and this is the most expensive misunderstanding in cold email.

Microsoft’s 30-messages-per-minute throttle produces no non-delivery report. Excess messages are queued and delivered over subsequent minutes. The only symptom is delay. Any process that judges health by “did we get a bounce?” is blind to it entirely.

When you do get an error, the code tells you which ceiling you hit — and they need different responses. Google’s 550-5.4.5 (“Daily user sending limit exceeded”) is a quota problem that resolves on the clock: stop that mailbox for the rest of the 24-hour window. Google’s 454-4.7.0 (“Too many login attempts”) is an authentication lockout, not a send block, and needs a long cooldown and a credentials check. Those two are routinely collapsed into a generic “mailbox error”, which produces both false alarms and missed real ones. Google publishes the full table in its SMTP error documentation.

A third category is neither quota nor auth: 450 4.2.1 means the recipient is receiving mail too quickly. Back off against that recipient, not your sender — misattributing it will wrongly quarantine a healthy mailbox.

What is the right lever when you need more volume?

More mailboxes. Never a higher per-mailbox number.

Every ceiling above is per-mailbox or per-tenant. Ten mailboxes each sending a comfortable daily volume beats one mailbox pushed toward its cap, on every axis that matters: you stay far below the hard limits, no single account carries the whole reputation, one bad day is contained, and you never trip a per-minute throttle you cannot see.

This is also why “unlimited sending” is a category error. The tool is not the constraint. The mailbox host is.

How WarmySender handles this

WarmySender enforces all four ceilings at once and lets the lowest win. Email warmup runs five adaptive ramp strategies with a deliberately sub-doubling, daily shape — the pattern Google’s own guidance describes — while campaign pacing spreads sends across your window rather than bursting. Per-mailbox and per-host caps sit underneath both, so a campaign configured above what a mailbox can safely take simply defers rather than pushing through.

Error handling distinguishes quota, authentication, throttle and block instead of collapsing them, which is what stops a temporary quota pause from being treated as a broken mailbox. And because verification keeps invalid addresses out of a send, your bounce and complaint rates stay inside what providers ask for — the reputation ceiling that has no published number.

An AI agent can drive all of it: read mailbox health and warmup stats, adjust warmup settings, build and launch campaigns, and get a plain-language answer to “why did this send nothing today?” What no agent can do is raise a cap. The scheduler paces every action inside safe limits regardless of who is driving. The deliverability FAQ and how cap enforcement works cover the details.

Frequently asked questions

Does finishing warmup raise my sending limit?

No. Provider limits are fixed by the mailbox host and are not affected by warmup. What warmup changes is your ramp position and your sending history — you arrive at real sending with an established pattern instead of a cold start. The ceiling was always there.

How many emails a day can one mailbox safely send?

Well below its provider limit. Google Workspace permits 2,000 a day, but that is a hard ceiling, not a target — and it counts every message including internal mail. A comfortable cold-outreach rhythm sits in the tens per mailbox per day. If you need more volume, add mailboxes.

Why did my campaign slow down without any errors?

The likeliest answer on Microsoft mailboxes is the 30-messages-per-minute throttle, which generates no bounce and delivers the excess over following minutes. On any provider it can also be reputation-driven deferral, which is temporary and reputation-linked rather than a fixed cap. Delay without errors is a signal in itself.

Does a bigger plan mean I can send more?

Not through the mailbox. A plan can allow more mailboxes, more campaigns and more verification, but the per-mailbox ceiling belongs to whoever hosts the mailbox. That is why capacity planning starts with counting mailboxes, not with picking a tier.

What is the single most common volume mistake?

Doubling. Google names it directly: immediately doubling previously sent volume can result in rate limiting or reputation drops. It usually happens the day a campaign goes from a test batch to the full list, which is exactly when the mailbox has the least history to absorb it.