The four caps that keep a sending domain alive

Most cold email tools compete on how much you can send. It is an easy thing to advertise and a straightforward thing to build.

It is also the wrong axis. Sending more is trivial. Sending enough to matter without destroying a domain you spent three weeks warming is the actual problem, and the mechanism that solves it is a limit rather than a feature.

Raechly enforces four, and they stack.

Per mailbox

Every mailbox carries a daily limit, and it rises on a schedule as the mailbox warms. Early days are deliberately small; the ceiling lifts as history accumulates.

This is the cap that does the most work, because reputation is assessed per sending address as well as per domain. A single mailbox producing an obvious volume spike is the clearest possible signal that something automated just started, and providers respond to it accordingly.

The counter that matters here is sent_today. When campaign volume looks lower than expected, this is usually why: mailboxes hit their cap and stop. Nothing is broken. It is throttled, and the difference determines whether you should add mailboxes or wait out the ramp.

Per domain

Mailboxes on the same domain share its reputation, so limiting each one individually is not sufficient. Ten mailboxes at thirty a day is three hundred messages from one domain, which is a very different signal from thirty.

The domain cap bounds the total regardless of how the mailboxes underneath it are configured. It is the reason adding mailboxes to a single domain has diminishing returns, and the reason scaling volume properly means adding domains.

Per campaign

The first two protect infrastructure. This one protects you from a specific mistake.

A campaign-wide daily cap bounds how fast any single sequence can burn through its enrolled leads. Without it, one campaign with a large list can consume every mailbox’s daily allowance before anything else sends, which means an untested sequence hitting a fresh list gets maximum distribution before you have seen a single reply.

Setting it to something small on a new campaign is the cheapest insurance available. You see the first forty responses before the next four hundred go out.

The hard ceiling

Thirty per mailbox per day, applied in the pipeline, overriding whatever is configured on the mailbox record.

You can set a mailbox limit of two hundred. It will send thirty.

This is the one people push back on, and the reasoning is worth stating plainly. Every other cap is a number you chose, which means every other cap can be raised at 11pm on a Thursday when a client wants results by Monday. That decision always feels justified in the moment and is almost always wrong, because the cost does not arrive for two weeks and by then it is attributed to something else.

The ceiling exists because the failure mode is delayed and irreversible. Domain reputation is built over months and destroyed in days, and there is no button that undoes it. A limit you can raise under pressure is not really a limit.

How to actually scale

Given the ceiling, more volume means more mailboxes and more domains. That is not a workaround; it is the correct shape.

Rough arithmetic. Thirty per mailbox per day. Four mailboxes per domain, which keeps the domain total reasonable. So roughly 120 a day per domain, 2,400 a month on weekdays.

Need 10,000 a month? Four domains, sixteen mailboxes. That sounds like a lot of setup and it is, which is why create_domain and create_mailbox are exposed to your assistant rather than buried in a wizard.

The distribution is the point. Four domains at moderate volume is safer than one domain at four times the rate, and when a campaign goes badly you lose a quarter of your capacity instead of all of it.

The two that are not about volume

Two more constraints round this out, and neither is a rate limit.

Send windows. Evaluated per lead in the campaign timezone, using allowed weekdays and start and end hours. Sending at 3am local is a strong automation signal, and the reply you would have got at 9am does not arrive at 3am regardless.

Suppression that survives a re-import. Unsubscribed and bounced leads are never revived by importing the list again. This is the constraint most likely to be missing elsewhere, and the easiest to test: upload a list containing an address that previously opted out, and see whether it sends.

What this costs you

Real things, and it is worth being honest about them.

You cannot blast. If your model depends on 500 sends a day from one mailbox, this is the wrong tool and no configuration will change that.

Scaling takes setup. Sixteen mailboxes across four domains is four sets of DNS records and four verification waits.

Results arrive slower. Warmup ramps before volume, and volume is capped after it. You are trading week one for month six.

That trade only makes sense if you intend to be sending in month six. For a single burst against a list you bought, none of this is worth it and something else will serve you better. For anyone building outbound as a channel they keep, the limits are the entire value.

---

Related reading: sending reputation, email warmup, and never send cold email from your primary domain.