The sending-infrastructure checklist before campaign one

Before the first cold email leaves, six things need to be true: you are sending from a secondary domain, its SPF, DKIM and DMARC records pass, its mailboxes have warmed for six to ten weeks, the list is verified, volume is capped at 25–40 emails per day per inbox, and every email identifies you and offers an opt-out. Miss any one and the campaign is compromised before the first subject line is written. This checklist is the order we build in.

Why does infrastructure come before copy?

Because copy only matters if it is read, and placement decides whether it is read. Mailbox providers score the sender before they score the message: domain reputation, authentication, volume pattern, list quality. The full model is in the practical guide to cold email deliverability — the short version is that a brilliant email from an unprepared domain goes to spam, and a plain one from a prepared domain gets read. Infrastructure is the multiplier on everything written afterwards.

What does the domain layer need?

Three items:

• A secondary sending domain. Cold outbound never runs on your main company domain. If a campaign damages a secondary domain, you retire it; your main domain — the one your invoices, proposals and client emails depend on — stays untouched.
• Authentication records that pass. SPF states which servers may send for the domain, DKIM cryptographically signs each message, and DMARC tells receivers what to do when the first two fail. All three configured, all three passing on a test send — not "probably fine".
• A custom tracking domain, if you track anything, so your links resolve through your own domain rather than a sending tool's shared one.

None of this is difficult. All of it is checkable in an afternoon, and commonly isn't checked at all.

What does the mailbox layer need?

Warmed mailboxes and hard volume caps. A new mailbox has no sending history, so it warms for six to ten weeks — gradually rising volume into a network of engaged mailboxes — until its placement score holds around 90 or above; 91/100 is the working example we use for "ready". Then it sends 25–40 cold emails per day, permanently. Not 41 on a good day. The cap is what keeps the sending pattern looking human, and mailboxes that respect it keep working for years.

Note the scheduling consequence: warm-up is the long pole. Everything else on this checklist fits inside those six to ten weeks, which is why the domain and mailboxes are bought in week one.

What do the list and the message need?

The list is verified — every address checked against a verification service before load, bounct-prone addresses removed — because bounces are the fastest way to undo six weeks of warm-up. And the message carries the two things UK law requires of B2B cold email: clear identification of the sender and a working opt-out. The mechanics of doing that without making a one-to-one email look like a broadcast are in opt-outs in cold email: compliant without killing replies.

What happens at launch?

The launch sequence is deliberately boring:

1. When every item above passes, the first campaign loads — small, one segment, one offer.
2. When sending starts, it starts at the bottom of the volume range, with warm-up traffic still running in the background.
3. When the first week's numbers come in — bounce rate, positive replies, complaints — they are checked against thresholds before volume rises. The numbers to watch and the thresholds that trigger action are in monitoring deliverability weekly.
4. When the numbers hold for a fortnight, volume steps up toward the cap. When they don't, sending pauses and the cause is fixed first.

The mechanism is: pass the checklist, launch small, read the gauges, then scale. Firms that invert this — launch big, then discover the gauges — burn the domain and start again minus two months.

Who answers the replies?

The checklist has one item that is not technical: someone owns the inbox. Replies from cold email are perishable — as with inbound leads, where contact rates drop roughly 8x after five minutes as an industry rule of thumb, a warm reply cools fast. If nobody is assigned to handle replies daily, the infrastructure delivers conversations to an empty room. Decide who answers, in what tone, within what window — before campaign one, not after reply one.

---

Next step: the Growth System Audit — £450, seven days, credited against any build — maps where your growth system leaks and what to build first.