How many inboxes per domain for cold email, and how many domains you actually need

Short answer: two to three inboxes per domain, and no more than three. Domain reputation is shared across every mailbox on that domain, so a fourth or fifth inbox does not spread your risk, it concentrates it. Each of those inboxes should send 20 to 50 cold emails a day once fully warmed, which puts a three inbox domain at roughly 60 to 150 sends a day. To scale past that you add domains, not inboxes. Everything else on this page is the arithmetic for working out how many of each you actually need.

Last updated August 2026. Written for US teams budgeting an outbound motion. The threshold figures below come from Google's published bulk sender requirements, read on August 15, 2026. Mailbox providers change these, so confirm before you build to them.

How many inboxes per domain for cold email?

Two to three. Three is the working ceiling most experienced outbound teams settle on, and the reason is structural rather than a matter of taste. Gmail and Microsoft 365 evaluate reputation primarily at the domain level, not at the individual mailbox level. When one inbox on a domain gets marked as spam repeatedly, the penalty attaches to the domain, and every other mailbox sitting on it inherits the damage.

That inverts the intuition most people bring to this. Adding a fourth and fifth inbox to a domain feels like spreading load, the way adding servers spreads traffic. It is closer to the opposite: you are putting more of your sending capacity behind a single reputation, so when that reputation drops you lose all of it at once instead of a third of it. Three inboxes across three domains and nine inboxes on one domain can send similar daily volume, but the first survives a bad week and the second does not.

How many cold emails per inbox per day is safe?

20 to 50 a day from a fully warmed mailbox, and 30 is the number to plan with if you want a figure that will not get you into trouble. New mailboxes should start far lower, around 5 to 10 a day, and climb over several weeks. Sustained sending above 50 a day from one inbox is where bulk sender detection reliably starts paying attention.

Google publishes the thresholds that sit behind this and they are worth knowing exactly, because most advice paraphrases them badly. Google defines a bulk sender as one sending more than 5,000 messages a day to Gmail accounts. It requires spam rates reported in Postmaster Tools to stay below 0.30 percent, and recommends staying below 0.10 percent. It requires SPF and DKIM on your sending domain plus DMARC, and one click unsubscribe on marketing and subscribed mail. It also asks you to start at low volume with engaged recipients and increase slowly, and to avoid sudden volume spikes.

The 0.30 percent number is the one to internalize, because it is much less forgiving than it sounds. Three complaints in a thousand sends puts you at the limit. At 30 emails a day from one mailbox, a single complaint roughly every eleven days is enough to sit on the threshold. Cold outreach lives permanently close to that line, which is the real reason the daily caps are low: the caps exist to keep the complaint rate small in absolute terms, not because a number like 50 is magic.

How many domains do I need for cold email?

Divide your daily send volume by 30 to get inboxes, then divide inboxes by 3 to get domains. That is the whole formula. The part people get wrong is the input, because your daily send volume is not your lead count. A three touch sequence sends three emails per lead, so 2,500 leads a month is 7,500 sends a month, not 2,500.

Here is that worked through at volumes an outbound team would actually run, assuming a three step sequence, 22 working days a month, 30 emails per inbox per day and 3 inboxes per domain.

Two things in that table surprise people the first time they see it. The first is how fast the infrastructure grows: a motion contacting 6,000 leads a month needs somewhere near 28 mailboxes across 10 domains, which is a real operational footprint rather than a settings change. The second is that every one of those mailboxes is a paid seat. Google Workspace Business Starter and Microsoft 365 Business Basic both sit in the single digit dollars per user per month in the US, so check the current rate, but the shape is clear: 28 mailboxes is 28 seats, every month, before you have paid for any outbound software at all.

Why does domain reputation matter more than inbox reputation?

Because mailbox providers are trying to answer a question about the sender, not about the address. An address is trivially cheap to create and discard, so if reputation attached only to addresses, spammers would simply rotate them forever. Domains cost money, take time to age, and carry DNS records you had to configure, which makes them a far better proxy for whether a real organization stands behind the mail.

The practical consequence is that your unit of risk is the domain. Burn one and you burn everything on it, which is why cold outreach is run from secondary domains rather than the company's primary domain. Sending cold volume from the domain your invoices and customer support run on risks the mail your business actually depends on, and that trade is never worth it. Buy lookalike domains, authenticate them properly, and keep the primary domain out of the outbound motion entirely.

How long should I warm up a new inbox?

Two to four weeks before it carries meaningful cold volume, and the ramp matters more than the total. Start at 5 to 10 sends a day, increase gradually, and get real replies coming back in the other direction, because two way conversation is a strong positive signal and pure outbound with no responses is a weak one. Google's own guidance is explicit on the shape: start low with engaged users and increase volume slowly, avoiding sudden spikes.

Plan warm-up as lead time, not as a task. If you need 12 mailboxes live in March, they should be bought and warming in February. Teams that skip this almost always do so because the campaign was approved late, and the result is the same every time: a burst of volume from cold domains, a complaint rate that clears 0.30 percent in the first week, and a set of domains that never fully recovers. Domains are cheap and calendar time is not, so buy them early.

What does this add to an outbound budget?

More than most people model, and it is the line item that quietly breaks vendor comparisons. When you price an AI SDR or an outbound platform, the software quote is one number and the sending infrastructure underneath it is another, and only some vendors are upfront that the second exists. Artisan is unusually clear about this: its pricing FAQ states plans are sized to your lead volume, mailboxes, and dialer seats, so mailboxes are an explicit pricing variable rather than an afterthought. We worked through what that means for a quote on our Artisan AI pricing page.

Do the arithmetic before you compare vendors, not after. Take your monthly lead target, run it through the table above, and you have a mailbox count and a domain count. Price the seats, price the domains, add any warm-up or deliverability tooling, and add that total to every software quote you are weighing. Otherwise you are comparing one vendor who bundled the infrastructure against another who did not, and the cheaper looking quote is often the one that assumes you already own 28 mailboxes.

There is an operational cost as well as a financial one. Once you are running more than a handful of mailboxes, simply keeping track of what arrived where becomes its own job: replies land in whichever inbox sent the original, bounces scatter, and nobody wants to log into a dozen accounts every morning. Being able to read every mailbox from one place is the difference between a motion you can actually run and a pile of browser tabs. Budget for the monitoring, not just the sending.

Can I just send more from fewer inboxes instead?

No, and this is the most common shortcut and the most expensive one. Pushing a single mailbox to 100 or 200 cold sends a day does not get you the volume of three mailboxes at 30. It gets you a mailbox that trips bulk sender detection, a domain whose reputation falls, and a deliverability problem that is slow and awkward to reverse. Reputation recovers far more slowly than it degrades.

The honest reframing is that inbox count is a proxy for how much reputation you have built, and reputation is earned over time rather than bought. You cannot compress it by sending harder. What you can do is start earlier, keep per-inbox volume conservative, keep the messages different enough from one another that they do not share a fingerprint, and make sure enough of them are relevant that replies come back. Volume follows from that, not the other way around.

The practical setup

For most teams starting outbound, the first build looks like this. Buy two or three secondary domains that read plausibly as your company. Put three mailboxes on each. Configure SPF, DKIM and DMARC on every domain before a single send, since Google now requires all three of bulk senders and treats their absence as a strong negative signal regardless. Warm every mailbox for two to four weeks. Then run at 30 a day per inbox, watch Postmaster Tools, and hold your spam rate under 0.10 percent rather than merely under the 0.30 percent limit.

Scale by repeating that unit. Another domain, another three mailboxes, another two to four weeks of warm-up. It is deliberately unglamorous and it is the part of outbound that separates campaigns that keep working from campaigns that get one good month. If you would rather not operate any of it yourself, that is a reasonable conclusion to reach, and it is worth reading do AI SDRs actually work and hire an SDR or use an AI agent before you decide which way to go. Whichever you pick, the mailbox arithmetic does not change, because the constraint belongs to Gmail and Microsoft rather than to any vendor.