The Cold Email Work That Happens Before You Write a Single Email

Nikhai Jaysen · September 2, 2026

You can write the best cold email of your career and have it land in spam. Deliverability is infrastructure, and it is built weeks before the first send.

Most advice about cold email is about the email. Subject lines, opening lines, how many sentences, whether to include a link. All of it is downstream of a question nobody finds interesting: will the message arrive at all?

Deliverability is not a writing skill. It is infrastructure, it takes weeks to build, and it is the single most common reason a campaign produces nothing. A team can spend a month on messaging and lose to a DNS record they never configured.

How the filter actually decides

It helps to know roughly what is being evaluated. When mail arrives, the receiving provider weighs a set of signals, and content is well down the list. Ahead of it sit authentication, the sending domain's history, the sending IP's history, and how recipients have engaged with mail from you before.

This ordering has a blunt implication. A perfect message from an unauthenticated, brand-new domain sending two thousand emails on its first day is filtered on the signals that were read first. The content never gets a fair hearing.

The setup, in the order it has to happen

Separate sending domains

Never run cold outreach from your primary domain. Register variations — adding a word, or using a different top-level domain — and redirect them to your main site so a curious prospect lands somewhere real.

The reason is containment. Cold outreach is inherently reputation-risky: some recipients will mark it as spam no matter how relevant it is. If that damage accrues to the domain your contracts, invoices, and support conversations run on, you have traded a marketing channel for your ability to do business over email.

Authentication: SPF, DKIM, DMARC

These three DNS records are how a receiving server verifies you are who you claim to be.

SPF publishes the list of servers permitted to send mail for your domain. DKIM attaches a cryptographic signature to each message so the recipient can verify it was not altered and did come from you. DMARC ties the two together and tells receiving servers what to do when a check fails, while sending you reports on what is being attempted in your name.

None of them is optional any more. Major providers have progressively tightened requirements for bulk senders, and an unauthenticated domain is now treated as suspicious by default rather than merely unverified.

Mailbox provisioning

Each sending domain carries a small number of mailboxes — a handful, not dozens. Volume is spread across them rather than concentrated, because per-mailbox sending limits are one of the signals providers watch. Spreading also means one mailbox developing a problem does not take the campaign down.

Every mailbox needs a real name, a real signature, and ideally a profile photo. These are trust signals for the human reading, and their absence is a pattern that filters have learned.

Warming

A new domain has no history, and no history is treated closer to bad history than to good. Warming solves this by sending small volumes of mail that gets opened and replied to, ramping up gradually over three to four weeks until the domain has a track record.

This is the step teams skip, because it is three weeks of doing nothing visible. It is also the step whose absence is unrecoverable in the short term: a domain that gets flagged in its first week does not recover by lowering volume afterwards. You start again with a new domain and another month.

The volume ramp

Even after warming, volume goes up in steps rather than in one jump. Sudden changes in sending pattern look like a compromised account, which is exactly the behaviour filters are tuned to catch.

Keep per-mailbox daily volume modest and add mailboxes to scale rather than pushing more through each one. This is slower than founders want, and it is the difference between a channel that compounds and one that burns out in a quarter.

List hygiene, which is deliverability too

Sending to invalid addresses produces bounces, and a high bounce rate is read as a sign that you acquired your list rather than built it. Verify every address before it enters a sequence, and remove anything that hard bounces immediately.

Relevance belongs in this section as well, which is not obvious. Mail that gets deleted unread and reported as spam teaches the provider that your domain sends unwanted mail. Tight targeting is therefore not only a conversion tactic — it is how you protect the infrastructure. We go into that side of it in the piece on why a broad ICP breaks outbound.

What to monitor once you are live

Three numbers tell you whether the infrastructure is holding.

Bounce rate should be low and stable. A rise usually means list quality has slipped rather than anything technical.

Spam complaint rate is the one to watch most closely, because the tolerated threshold is very small and crossing it has consequences that outlast the campaign.

Reply rate by sending domain will reveal a domain quietly falling out of the inbox before any other metric does. If one domain's replies drop while others hold steady on the same copy and the same segment, that domain is being filtered.

The honest cost

Done properly, this is three to four weeks of setup before the first real campaign, and an ongoing monitoring job after that. There is no version of cold email where you buy a tool on Monday and book meetings on Wednesday, and any vendor implying otherwise is selling you a domain that will be unusable by month three.

The upside of treating it as infrastructure is that it compounds. Warmed domains with clean reputations keep working, and the second campaign starts from a much stronger position than the first.

We build and maintain this layer as part of AI email outreach — domains, authentication, warming, rotation, and monitoring — so the messaging work is not wasted on mail nobody sees. Get in touch if you want a look at your current setup before you scale sending.

Frequently Asked Questions

Why do cold emails go to spam even when they are well written?

Because filtering decisions are made largely on sender reputation and authentication before content is weighed at all. An unauthenticated domain with no sending history and a sudden volume spike gets filtered regardless of how good the message is.

Should you send cold email from your main company domain?

No. Use separate sending domains that redirect to your main site. If a cold campaign damages a domain's reputation, you want that damage contained away from the domain your invoices, contracts, and support email depend on.

How long does domain warming take?

Plan for three to four weeks before meaningful volume. Warming ramps sending gradually so mailbox providers build a positive history for the domain. Rushing it is the most common cause of campaigns that fail from day one and never recover.

What are SPF, DKIM, and DMARC?

Three DNS records that prove your mail is legitimate. SPF lists which servers may send for your domain, DKIM cryptographically signs each message, and DMARC tells receiving servers what to do when the first two fail. Missing any of them materially increases the chance of being filtered.