How Many Cold Emails Can You Send Per Inbox Per Day?

On a mature domain with clean sending history, 20 to 30 cold emails per inbox per day is the working range. During the ramp it is closer to 5 to 10. On a domain you registered last week it is zero. Anyone who gives you one universal number is guessing, because the ceiling is set by your complaint rate and your reply rate, not by a send count. That is not a dodge. There is a real formula underneath it, and once you can see it you can calculate your own ceiling instead of copying someone else’s. This post walks through the complaint math, a warmup ramp you can actually run, how domain history moves the number up or down, and how to size an inbox pool for a target volume. It also makes an argument you will not like if you came here looking for permission to send more: volume is almost always the wrong lever. If you want the underlying spam-filter mechanics first, read our breakdown of cold email deliverability in 2026. This post picks up where that one stops and deals purely with volume.

The Real Ceiling Is a Complaint Budget, Not a Send Limit

Start by ruling out the limits people assume are binding. Google Workspace publishes a cap of 2,000 messages and 2,000 unique external recipients per user per day. Microsoft’s Exchange Online allows 10,000 recipients per rolling 24 hours with a throttle of 30 messages per minute. Nobody running cold outbound gets within an order of magnitude of either. The provider hard limit is not what stops you.

What stops you is the spam rate. Per Google’s sender guidelines, you should keep the spam rate reported in Postmaster Tools below 0.10 percent and avoid ever reaching 0.30 percent. Run that as arithmetic on a single inbox. At 30 sends a day across 22 working days you send 660 messages a month. Two spam complaints on 660 messages is 0.30 percent. One complaint is 0.15 percent, already over the target Google actually recommends.

Two details make the budget tighter than that calculation suggests. First, the denominator is Gmail traffic only. If 45 percent of your list is on Gmail, roughly 300 of those 660 messages carry the entire budget, and a single complaint puts you at 0.33 percent. Second, the rate is aggregated at the domain level, not per mailbox. Every inbox on a sending domain draws from one shared budget.

That second point kills the most common workaround. Spreading the same volume across four inboxes on the same domain does not lower your spam rate, because the rate is a ratio and the domain is the unit of measurement. You have moved the sends, not the math.

• Provider hard caps: 2,000 messages per user per day on Google Workspace, 10,000 recipients per 24 hours on Microsoft 365
• Reputational cap: below 0.10 percent spam rate target, 0.30 percent hard line, per Google’s sender guidelines
• Spam rate denominator counts Gmail-delivered mail only, so a Gmail-heavy list burns budget faster
• Spam rate aggregates per domain, so more inboxes on one domain dilutes nothing

A Warmup Ramp You Can Actually Run

Google’s guidance is explicit about pacing: start with a low sending volume, increase it slowly over time, and avoid sudden volume spikes if you have no history of sending large volumes. Cold outbound has no pre-engaged audience to start with, which is exactly why warmup tools exist. They manufacture the positive engagement history that a normal sender would accumulate naturally.

The ramp below is what we run on new infrastructure. It is deliberately slower than most tool defaults, and the extra two weeks cost far less than a burned domain pool costs.

Two mistakes undo a good ramp. The first is switching warmup off the day real sending starts. Keep it running permanently at reduced volume, because it holds up the positive-signal ratio that mailbox providers score you on. The second is ramping every inbox in the pool on the same day. Stagger new inboxes by a week or so, otherwise a single bad list or a bad copy test takes the whole pool down together and you have no clean control group to compare against.

Spread the daily volume across a six to eight hour window with randomized gaps between sends, which both Instantly and Smartlead let you configure directly. Firing 25 messages in four minutes is a pattern no human account produces, and Microsoft throttles at 30 messages per minute regardless.

• Days 1 to 14: automated warmup only, zero real sends, with SPF, DKIM and DMARC live before day one
• Week 3: 5 real sends per inbox per day
• Week 4: 10 per day
• Week 5: 15 per day
• Week 6: 20 per day
• Week 7 onward: 25 to 30 per day, only if bounce rate is under 2 percent and reply rate is holding steady

Domain Age Moves the Ceiling, But Not the Way People Think

Reputation attaches to sending history, not to registration date. An aged domain with no mail history is not warm, it is just old. This is why buying aged domains as a shortcut buys you nothing useful, and can actively buy you someone else’s problem: a domain that previously ran affiliate offers or was hijacked for spam starts from a negative position, and you inherit it. Before you buy any used domain, run blacklist lookups and check the Internet Archive for what it used to host.

What genuinely raises the ceiling is accumulated clean sending. A domain that has been sending 20 a day at a 6 percent reply rate for four months has earned headroom. A domain three months old that has never been pushed above 10 has not earned it yet, even though the calendar says it should be fine.

One practical wrinkle: Postmaster Tools needs meaningful volume before it will display a domain reputation grade. Most cold email domains sit below that threshold and show nothing at all, which means the diagnostic everyone tells you to check is often blank. That is not a reason to skip it. It is a reason to treat per-inbox reply rate as your primary instrument instead, which is covered further down.

• Under 30 days since registration: 0 real sends, warmup only
• 30 to 90 days, clean history, warmup complete: 10 to 20 per day
• 90 days or more with a medium or high reputation grade in Postmaster Tools: 25 to 40 per day
• Any domain that has taken a complaint spike or a blacklist hit: drop to 5 per day and rebuild over three to four weeks
• Purchased aged domain with unknown history: treat it as new until it has 60 days of your own clean sending

The Inbox Math for a Target Volume

Say you want 10,000 sends a month. At 25 per inbox per day across 22 working days, one inbox carries 550 messages a month. That gives you 19 inboxes, and you should round up to about 21 because at any given moment one or two will be paused, bouncing, or re-warming. At three inboxes per domain, that is seven sending domains.

Now apply the correction almost everyone skips. Sends are not leads. If your sequence runs four steps and repliers exit early, each lead consumes roughly three and a half sends. Your 10,000 monthly sends represent something closer to 2,800 new leads a month, not 10,000. People build infrastructure for the wrong number constantly because they size against sends and forecast against leads.

Then apply the calendar correction. Twenty-one inboxes each need six to eight weeks from purchase to full volume. If you want that number producing in September, you buy domains in July. Infrastructure lead time is the most commonly missed line item in an outbound plan, and it is the one you cannot compress with money.

• Inboxes = monthly send target ÷ (daily per-inbox volume × working days in the month)
• Add 10 to 15 percent for inboxes paused, bouncing, or re-warming at any given time
• Domains = inboxes ÷ 3
• New leads per month = monthly sends ÷ average steps consumed per lead
• Calendar: 6 to 8 weeks from domain purchase to full sending volume

More Inboxes Is Not a Strategy

Here is the position we take with clients, including the ones who do not want to hear it. Adding inboxes does not fix anything. It multiplies whatever your current message already does. If 100 sends produce one complaint and zero replies, 1,000 sends produce ten complaints and zero replies. You have scaled the problem and paid for the privilege.

The complaint rate is a rate. It does not improve by being divided across more mailboxes, and as covered above it is aggregated at the domain level anyway. What does change is your operational surface area. Every inbox is a seat cost, a set of DNS records, a warmup cycle, a monitoring line, and one more place a misconfigured DKIM key can quietly poison a domain. Twenty-one inboxes means seven DNS zones you now maintain.

The pattern we see most often when we open an outbound audit looks like this: someone scaled to 40 inboxes to hit a volume target, reply rate stayed flat at around half a percent, and then the entire pool degraded within the same fortnight because it was all bought from the same registrar in the same week, warmed on the same schedule, and running the same copy. Nothing was isolated, so nothing could be diagnosed. Recovery took longer than the campaign had been running.

Infrastructure should scale to match a message that already works. Scaling infrastructure to compensate for a message that does not work is how domain pools die.

Reply Rate Is the Cheapest Deliverability Asset You Own

Mailbox providers score engagement, and a reply is the strongest positive signal available to you. It establishes two-way conversation history between two specific addresses, and mail between addresses with conversation history lands with far less scrutiny. This is the mechanism most volume advice ignores: reply rate is not only a conversion metric, it is an input that raises your sending ceiling.

Compare two inboxes. One sends 15 a day and gets an 8 percent reply rate, producing roughly 26 replies a month. The other sends 50 a day at 0.5 percent, producing about 5 replies from more than three times the volume. The first inbox is building reputation with every campaign. The second is spending it. Six months later, the first inbox can safely send 40 a day and the second cannot safely send anything.

Use reply rate as a per-inbox instrument, not a per-campaign one. Track it at the mailbox level. When the same copy going to the same list segment produces 5 percent replies on four inboxes and 0 percent on the fifth, you have a placement problem on one inbox, not a copy problem. Pull that inbox and re-warm it. Do not rewrite the email.

It is also your earliest warning. A reply rate that falls with no change to copy, list, or targeting usually means placement has slipped, and it shows up before bounce rates move. By the time bounces climb, you have already been in spam for a week or two.

What We Do When a Client Asks for More Volume

Usually we say no to the volume and yes to the pipeline number sitting behind the request. Nobody actually wants 10,000 sends a month. They want the meetings they believe 10,000 sends will produce, and there is almost always a cheaper path to the same meetings.

The order of operations matters. Tighten targeting first, using Clay for signal-based enrichment and Prospeo for contact data, so you are writing to people with an actual reason to reply this month. Verify hard with Reoon and drop catch-alls and risky addresses rather than gambling on them, because bounces cost reputation at the same domain level complaints do. Split the load across channels so email is not carrying the whole number alone, which is what HeyReach handles on the LinkedIn side. Only then, if the arithmetic still demands it, add inboxes.

The reason this is worth the effort is straightforward. Three thousand sends to a signal-qualified list at a 6 percent reply rate produces roughly 180 conversations. Twelve thousand sprayed sends at 1 percent produces 120. The first version fits in six inboxes and two domains. The second needs 22 inboxes and eight domains, four times the warmup time, and four times the complaint surface, to deliver fewer conversations.

Across 118+ GTM systems built and $500M+ in new pipeline generated, the accounts that hold inbox placement longest are almost never the highest-volume ones. They are the ones where reply rate was fixed before the infrastructure was scaled. If you want a second opinion on your own numbers, our free 30-minute outbound audit produces a written fix list covering your current per-inbox volume, ramp schedule, and domain pool sizing.

Related reading

• Cold email spam test: how to check inbox placement before you send
• SPF, DKIM and DMARC for cold email, explained properly
• Cold email deliverability in 2026: why your emails land in spam

Frequently asked questions

How many cold emails can I send per inbox per day?

On a mature domain with established clean sending history, 20 to 30 cold emails per inbox per day is the safe working range. Pushing to 30 to 50 is only defensible when Postmaster Tools shows a medium or high domain reputation and your bounce rate sits under 2 percent. During the first two weeks of real sending, stay under 10. On a domain registered in the last 30 days, the safe number is zero.

How long should I warm up a new cold email inbox before sending?

Plan on two weeks of automated warmup with zero real sends, followed by a four to six week ramp that climbs from 5 sends a day to full volume. Total time from domain purchase to full sending is six to eight weeks. Keep warmup running at reduced volume permanently after you go live, because it maintains the positive engagement ratio mailbox providers score you on.

How many inboxes do I need to send 10,000 cold emails a month?

About 21. At 25 sends per inbox per day across 22 working days, one inbox carries 550 messages a month, so 10,000 sends requires 19 inboxes plus roughly 10 percent headroom for mailboxes that are paused or re-warming. At three inboxes per domain that is seven sending domains. Budget six to eight weeks before they all reach full volume.

What spam complaint rate is safe for cold email?

Google’s sender guidelines say to keep the spam rate reported in Postmaster Tools below 0.10 percent and to avoid ever reaching 0.30 percent. At 660 sends a month from one inbox, two spam complaints puts you at 0.30 percent. The rate is calculated on Gmail-delivered mail only, and it aggregates at the domain level, so every inbox on a domain shares one budget.

Does sending more emails per inbox directly hurt deliverability?

Volume by itself does not hurt deliverability. Volume that outruns your engagement does. Mailbox providers score the ratio of positive signals such as replies and moves out of spam against negative ones such as spam reports, bounces, and deletes without reading. Doubling your sends without doubling your replies halves that ratio, and the ratio is what determines placement.

Should I add more inboxes or send more per existing inbox?

Neither, until reply rate is fixed. Adding inboxes multiplies whatever your current message already produces, including complaints, and each one costs a seat, a DNS setup, and six weeks of warmup. Raising per-inbox volume on a domain with a weak engagement ratio degrades reputation faster. Improve targeting and reply rate first, then size infrastructure to match the message that works.