How many cold emails per day per mailbox? Fewer than you want.

Every cold email forum has the same recurring question — how many can I send per day? — and the same recurring fantasy answer: as many as possible, as soon as possible.

Here’s the honest number for 2026: 20–50 cold emails per day per mailbox, after a proper warmup. Push past it and nothing announces the mistake; your inbox placement just quietly decays until replies stop and you blame your copy.

Where the number comes from

A mailbox provider watching a single mailbox send 200 unsolicited emails a day to strangers — with the reply rates cold email actually gets — has seen that movie before. The model that catches spammers is a volume-and- engagement model, and high-volume-low-engagement is its core signature.

A human salesperson writing real one-to-one email sends a few dozen messages a day. 20–50 stays inside the silhouette of human behavior. That’s the entire logic. It’s also why the ceiling can’t be negotiated with tips and tricks — the ceiling is the disguise.

Why this actually happens, not just what happens

Mailbox providers don’t have a rule that says “cold email is banned.” What they have is a model trained on the shape of abuse, and the shape of abuse looks like a mailbox that sends far more mail than it receives, to addresses it has no prior relationship with, that gets opened, replied to and forwarded at a much lower rate than normal mail. Every one of those signals is measurable at scale, and none of them requires reading a single word of your message.

A mailbox behaving like a human colleague — a few dozen sends a day, a healthy mix of sent and received mail, a normal reply rate, recipients who don’t immediately hit “report spam” — doesn’t trip that model, regardless of the fact that the content is still, technically, unsolicited outreach. A mailbox sending 300 a day with a 1% reply rate and a rising complaint count trips it fast, even if every message is thoughtfully written. This is why “better copy” alone never fixes a volume problem: the filter isn’t reading the copy, it’s reading the pattern the mailbox produces at scale. That’s also why the ceiling holds even as AI drafting makes each individual message look more human — a model can’t fake the aggregate behavior of a mailbox over weeks, only the surface of one message.

Three numbers above the ceiling worth knowing:

• 5,000/day to Gmail or Yahoo addresses makes you a “bulk sender” under the 2024 rules — stricter authentication, one-click unsubscribe, and the spam-complaint regime below. For cold outreach this line is a tripwire, not a target.
• 0.3% spam complaint rate is the hard limit where Gmail starts refusing mail. Keep under 0.1% to sleep at night. At 30 sends/day you effectively can’t afford complaints at all — which is the real argument for tight targeting.
• ~2% bounce rate is where reputation damage starts compounding. Verify before sending (free checker), and clean your lists before they age into liability.

Scaling without burning: horizontal, not vertical

The volume math still has to work — agencies and sales teams really do need hundreds of sends a day. The answer everyone lands on:

More mailboxes, not more per mailbox.

• 2–3 mailboxes per domain, ~30–40 sends each.
• Separate-but-related domains (never your primary), each warmed, authenticated, and health-checked.
• Need 300/day? That’s roughly 3 domains × 3 mailboxes × 35 sends — with no single point of reputational failure. One domain has a bad month; the campaign survives.

Yes, it costs more than one abused mailbox. It’s also the entire difference between an operation that runs for years and a domain graveyard.

The math at a few common targets

Working from ~30–35 sends per mailbox per day as a comfortable, sustained rate (not the absolute ceiling — a working margin under it):

Target volume/day	Mailboxes needed	Domains (at 2–3 mailboxes each)
100/day	3	1–2
300/day	9	3–4
600/day	18	6–8
1,000/day	30	10–13

The domain count matters more than the mailbox count. Each domain is a separate reputation asset with its own SPF/DKIM/DMARC history and its own warmup curve — spreading mailboxes across too few domains just recreates the single-point-of-failure problem one layer up. A domain that gets flagged doesn’t take down one mailbox, it takes down every mailbox living on it.

When more mailboxes stops being the answer

Horizontal scaling has a ceiling of its own, and it isn’t infrastructure — it’s targeting. Every mailbox added to a campaign needs its own share of genuinely qualified recipients; splitting a shallow list across 30 mailboxes doesn’t make the list deeper, it just spreads the same mismatched targeting across more reputation surface. Before adding the tenth mailbox, the honest question is whether there are enough real prospects in the segment to occupy it — if not, the fix is a bigger, better-built list, not another mailbox sending to the same names twice.

The volume nobody budgets: enough to learn

There’s a second reason per-mailbox volume disappoints people — at 30 sends a day, your A/B tests starve. A subject-line test needs hundreds of sends per variant before the difference means anything; eyeballing a winner at 50 sends is how bots pick your copy. Before splitting traffic, check what sample you actually need with the free A/B Test Calculator — the answer is usually “run it two more weeks.”

Run your own math

Whether the whole machine is worth building is a spreadsheet question: mailboxes × daily sends × reply rate × meeting rate, against what a closed deal pays you. We built the Cold Email ROI Calculator to make that an honest five-minute exercise — and the volume-math post walks through why 1,000 sends turns into fewer conversations than anyone advertises.

The summary fits in one line: send like a person, scale like an operation. Per-mailbox restraint is what keeps the per-domain asset alive — and warmed, trusted domains are the only inventory in this business that appreciates.

Two things worth settling before you scale

Does the 20–50 ceiling apply the same way to every provider? Roughly, yes — Gmail, Outlook and the smaller providers all watch the same volume-and-engagement shape, even though the exact thresholds and tolerances aren’t published and shift over time. Treat 20–50 as the sustained band to plan around, not a number any provider has printed verbatim.

Does warmup let you send more per mailbox, or just get to the ceiling safely? The latter. Warmup earns a mailbox the trust to send at 20–50 a day without tripping new-sender suspicion; it doesn’t raise the ceiling itself. A fully warmed mailbox and a freshly warmed one both live inside the same silhouette once they’re both established — warmup is about the runway, not the altitude.