How ColdMerge sends

Cold email has a bad reputation, most of it earned. Plenty of tools in this category sell speed, tricks to look human, and features whose only purpose is to help you avoid the consequences of a list you should not be mailing.

ColdMerge is built on the opposite bet: that the sender who lasts is the one whose sending is genuinely legitimate, not the one whose sending is disguised. This page states the rules the product holds itself to. They are not settings. You cannot turn them off, and neither can we.

We send from your real mailbox, and we say so

Every message ColdMerge sends leaves your own Gmail or Google Workspace account. Your address is the From:, it consumes your own Google quota, and replies land in your own inbox where you read them.

There is no shared sending infrastructure, no rented IP addresses, and no other customers whose behaviour can affect your reputation or yours theirs. What arrives in a recipient’s inbox is genuinely an email from you, because it is.

We also do not overclaim what happens next. When ColdMerge says a message was sent, that means Gmail accepted it and it is sitting in your Sent folder. That is all it means. It is never a claim that the message was delivered, or placed in an inbox rather than a spam folder, or read. Gmail gives no delivery receipt, so ColdMerge reports no delivery figure. A number nobody can measure is not a number worth printing.

Rotation spreads work across mailboxes, never mid-conversation

If you have connected several mailboxes to a campaign, ColdMerge spreads the sending across them. That is what keeps any one mailbox inside its own budget and its own pace.

The important limit is what rotation never does:

A contact is bound at their first send to one mailbox for the whole life of that campaign’s follow-ups. Every follow-up, and the reply thread it lives in, stays with that same mailbox. Rotation assigns senders across different people. It never rotates within a conversation with one person.

The reason is simple. Nothing looks more automated than a follow-up arriving from a different person than the one who wrote to you last week. A thread should read like one human wrote it, because one human’s mailbox did.

Two consequences worth knowing:

Which mailbox sends to a given person is decided at the moment of sending, not at launch. So there is no pre-launch forecast of the split between your mailboxes. The campaign shows the real split as it happens instead. This also means nobody is left stranded behind a mailbox that happened to be disconnected when you launched.
If the bound mailbox disconnects mid-sequence, that person’s follow-ups pause until you restore it. There is no silent fallback to a different mailbox, because that would break the thread. Reconnect and the follow-ups resume, oldest first, under the usual budget, window, and pacing.

One current limitation, stated plainly because it can surprise you: binding holds within a campaign, not across campaigns. The same contact reached by two different campaigns from your organization can hear from two different mailboxes. This is a known gap in the current version, not an oversight.

Suppression cannot be bypassed

Your suppression list is the organization-wide “never email this person again” record, and it is checked on every single send with no exceptions.

Not checked at launch and trusted thereafter. Checked at the moment each message is handed to Gmail, so somebody who unsubscribes while a campaign is running is protected on the very next send rather than the next campaign.

There is no override, no “send anyway”, no admin bypass, and no path through the import flow. A suppressed address is skipped at import time in every form: no contact created, no fields updated, no list membership added, checked before anything is written at all.

Even a preview send checks it. A preview is a real email, so it obeys real rules.

Suppression is matched by a one-way hash of the address rather than the address itself, which is what lets somebody be erased from your records entirely and stay suppressed forever anyway. The hash is the last thing standing between an erased person and being re-imported and re-emailed by accident.

Which entries can be removed follows from what they mean. An entry somebody typed by hand can be removed, because people make typos. A bounce can be removed, because that is ColdMerge’s own technical inference and inferences can be wrong. An unsubscribe can never be removed, because a person said stop. Full detail in Unsubscribes and suppression.

Unsubscribes are real, and they keep working

Two independent opt-out mechanisms, both writing to that same suppression list.

The List-Unsubscribe header is on by default. It is invisible in the body of the message and lets a mail client offer its own one-click unsubscribe control. It is a positive deliverability signal in its own right.

A visible footer link, with your postal address, is off by default and strongly advised on if you send to the US, the EU, or the UK. It cannot be turned on without a postal address on file, its wording is fixed rather than editable, and ColdMerge appends it at send time rather than asking you to paste it into a template. If either the link or the address is missing from any part of a message, the send is refused outright, naming what is missing, rather than going out incomplete.

Keep the visible footer in mind even with the header enabled. Mail clients decide for themselves whether to render their own unsubscribe control, and they do not always render one. The visible footer is the mechanism you control, and therefore the only one you can rely on.

The link itself is built to outlive everything around it. Every message carries an opaque, unguessable token that resolves to exactly one message and one person. It is deliberately not a signed value, because these links sit in strangers’ mailboxes for years and rotating a signing key would silently break every link ever sent. Someone can unsubscribe from an email you sent them three years ago, and it will work.

Verification informs, it never condemns

ColdMerge can check addresses before you send. The check returns one of four answers: alive, dead, risky, or no answer at all. You can also have no result, which simply means the address has not been checked.

Here is the rule that matters: a dead verdict never suppresses anybody. It does not add them to your suppression list. It does not mark them undeliverable. Only a real bounce or a real unsubscribe may do that, because those are facts that actually happened, and a third-party provider’s pre-send guess is not.

The reasoning is asymmetry of harm. Verification providers are sometimes wrong. If a false dead verdict fed the same list that unsubscribes feed, one bad guess would permanently condemn a real prospect, and unsubscribe entries can never be removed. So the guess-book and the compliance record stay separate on purpose.

In practice at launch: dead addresses are skipped by default with one click to include them anyway, and risky or no-answer addresses still go out with a caution on screen. It is a nudge, not a gate.

A few more things ColdMerge will not do with verification. It never runs a check automatically on import or at launch, only on a deliberate action where the cost is shown first. It charges only for an actual answer, so a provider timeout costs you nothing. And a result is stored against the address rather than the contact, so one address is checked and paid for at most once, and the answer is reused across every later import, list, and campaign. Results eventually go stale, currently after about 90 days, and become eligible for re-checking. Even then, nothing is re-checked or charged automatically; a stale result just counts as checkable again the next time you deliberately run a check.

No warmup pool, now or ever

Many tools in this category offer a warm-up network: your mailbox automatically exchanges chatty emails with thousands of other mailboxes in a pool, all of them opening and replying to each other, to manufacture the appearance of a healthy, engaged sender.

ColdMerge does not build this and will not. It is fake engagement. It is a network of machines lying to Google about how popular you are, and the entire value of it depends on Google not noticing. That is not a foundation to put your company’s mailbox on.

The legitimate substitute is ramp-up: starting low, growing on volume you actually sent, and staying under the ceiling. It is slower. It is also real, which means it does not stop working the day the provider decides to look closer.

What we do not collect

If you use open and click tracking, it runs through a tracking domain you own and records only the type of event, when it happened, and which link. No IP address, no user agent, no location. The tracking token names the message, not the person’s address. Your unsubscribe link is never rewritten through tracking, so opting out never routes through a redirect.

Reply detection only ever looks at threads ColdMerge started, and a notification about a reply carries a snippet and a link rather than the message body. Your mail is not ours to read.

And the analytics we use to improve the product are kept strictly separate from your customer data. Contact information never enters them.

Deliverability overview for the philosophy in short.
Bounces and the circuit breaker for the one place ColdMerge stops you automatically.
Unsubscribes and suppression for the compliance detail.