The pre-send cold email checklist: four gates, sixteen checks

Most campaign post-mortems find the same thing: the failure was visible before the first send — an unauthenticated domain, a list verified last year, a tracking link in email one, a mailbox doing 200 a day. None of it is subtle. All of it is checkable in an afternoon. So here’s the checklist we actually use, structured as four gates: if a gate fails, don’t send — fix it first. Copy earns replies; gates decide whether copy gets read at all.

Gate 1 — the domain

1. SPF, DKIM and DMARC all pass. Not “mostly” — since 2024 the majors increasingly reject unauthenticated mail outright. One pass through the domain health checker settles it; the explainers (SPF, DKIM, DMARC) fix what fails.
2. The sender has enough stable history for the target volume. Use two weeks as the minimum observation period for a low-volume mailbox and four to eight for a new or higher-volume domain. The decision framework is here and the warmup calculator turns a target into a planning schedule. On Norbelys, warmup is included.
3. Not on a blocklist. Thirteen DNSBLs, one free sweep — and if you’re listed, here’s the delisting playbook.
4. Cold sends on a secondary domain, never on the domain your invoices and password resets depend on. Setup: secondary domains for cold email.

Gate 2 — the list

5. Verified this month, not this year. Contact data changes continuously; over 2% bounces and Gmail starts rejecting. Free: clean the file. On Norbelys, imports verify automatically.
6. Every contact fits a written ICP. “Could plausibly buy” is not a segment — the list-building pipeline is the difference between outreach and littering.
7. Suppression applied. Unsubscribes, bounces and past “no”s are out — enforced at send time, not by memory.
8. Segmented by one shared pain. A Woodpecker platform benchmark associates smaller, tighter lists with higher reply rates. If one message fits the whole list, the list is too big.

Gate 3 — the copy

9. Under 100 words, one ask. The anatomy and templates if you’re starting from a blank page.
10. The first line is about them — it renders in the preview pane before any open. Rewrites here.
11. Plain text, no links, no images in email one. Links are the heaviest spam signal a no-history domain can carry. The tracking pixel counts too — and since most opens are robots anyway, you’re paying deliverability for a fictional metric.
12. Unsubscribe works, in one click. Required by Gmail/Yahoo above 5,000/day, wise at any volume, and the law is less demanding than you think — meet the higher bar anyway.

Gate 4 — the send settings

13. Under ~40 per mailbox per day. More volume = more mailboxes and domains, never more per mailbox — the per-mailbox math.
14. Human gaps between sends. Sixty emails at 9:00:00 sharp is a signature, not a schedule. Norbelys paces with per-sender budgets and jitter automatically.
15. Both thresholds on a dashboard you’ll see: bounce under 2%, complaints under 0.3% — the complaint-rate math explains why 1-in-300 annoyed readers is the whole ballgame.
16. Auto-pause armed. When a threshold trips mid-campaign, the campaign should stop itself — at 2am, without you. (Ours does, on both.)

Why order matters more than any single check

The four gates aren’t a flat list for a reason — later gates are worthless if earlier ones fail, but the reverse isn’t true. A domain with clean authentication and a real sending history will tolerate imperfect copy far better than perfect copy will tolerate a burned domain, because inbox placement is decided mostly upstream of anything the recipient reads. That’s also why gate 1 is the one worth doing weeks in advance rather than the afternoon before a send — authentication and blocklist status are fixable in minutes, but sending history and reputation are not something you can rush. Running the gates out of order, polishing copy on a domain nobody’s checked in months, is the single most common way a genuinely well-written campaign gets buried before a human ever sees it.

What skipping gate 1 actually costs

Picture a team that’s confident in their copy and their list, skips the domain gate because “it’s worked before,” and launches on a domain that’s been quiet for a couple of months. The first few hundred emails land fine — the domain still has some residual reputation — but as volume ramps toward the target, bounce and complaint rates creep past what the mailbox provider tolerates, and inbox placement starts silently dropping mid-campaign. By the time anyone notices reply rates cratered, the campaign has already sent to most of the list, the domain’s reputation has taken a real hit, and the fix isn’t “edit the copy” — it’s weeks of rebuilding sending history before that domain is trustworthy again. The checklist exists specifically to catch this before the first send, because there’s no version of this failure that’s cheap to fix once it’s already happened at volume.

The habit that makes it stick

Run the gates in order — domain, list, copy, send — because failures upstream make downstream checks meaningless: perfect copy from a blocklisted domain is a tree falling in an empty forest. First campaign on a new domain? Do gate 1 a full month before you need it; everything else fits in a day.

Pre-send checklist FAQ

How often should I re-run the full checklist, not just gate 4?+

Gate 1 (domain) needs a fresh check any time sending volume changes meaningfully or a domain has gone quiet for more than a couple weeks. Gate 2 (list) needs re-verification before every new campaign, not just the first one, since contact data decays continuously. Gates 3 and 4 are closer to per-campaign defaults you set once and revisit only when something changes.

What if a threshold trips mid-campaign after everything passed at send time?+

That's exactly what gate 4's auto-pause check is for. Passing the pre-send checklist doesn't guarantee a campaign stays healthy for its whole run, because bounce and complaint rates are measured against live recipient behavior you can't fully predict in advance. The checklist gets a campaign started safely; an armed auto-pause is what keeps a campaign that starts safely from degrading unnoticed.

Can I skip a gate for a small, low-volume test send?+

Skip gate 4's volume thresholds if you're genuinely testing at low volume — a handful of emails to a handful of contacts doesn't need per-mailbox pacing rules built for hundreds a day. Don't skip gates 1 through 3: a bad domain or a stale list produces misleading results even at low volume, and a test that passes on a broken foundation tells you nothing useful about how the real campaign will perform.

And after it sends: judge the campaign on replies, not opens, call A/B winners with real statistics, and triage every reply like the asset it is. The checklist gets you delivered; what happens next is the actual job.