Most email marketing checklists list thirty things to do before you press send. Five of them decide whether the message arrives at all, and the other twenty-five only matter once it does. Working through them in the published order is how founders end up with a beautifully formatted email sitting in a spam folder.
This is the shorter version, ordered by what actually breaks.
Deliverability is not a step in the checklist. It is the checklist.
An email that does not reach the inbox cannot be improved by a better subject line. Yet almost every published checklist puts authentication somewhere near the bottom, under formatting and send times, if it appears at all.
The rules also stopped being advisory. Google's email sender guidelines, in force since February 2024, require anyone sending more than 5,000 messages a day to Gmail to authenticate with SPF and DKIM, publish a DMARC record, support one-click unsubscribe, and keep spam complaints below 0.30 percent. They advise aiming for 0.10 percent, because a single bad week at 0.30 leaves no margin.
Below that volume you fall outside the strict threshold and should meet it anyway. The work is one afternoon and it does not need repeating.
The five that decide whether it arrives
01 Authenticate the sending domain.
SPF, DKIM and DMARC, configured and verified rather than assumed. Send a test to an address you control and read the raw headers. If any of the three fails, nothing further on this list matters.
02 Send from a domain you can afford to lose.
A separate sending domain, never the one your invoices come from. If a campaign draws complaints the harm lands somewhere you can retire.
03 Warm it before you use it.
Volume ramped over weeks, not switched on. A domain with no history that suddenly sends two thousand messages is behaving exactly like a spammer, and receivers score it that way.
04 Make unsubscribing trivial.
One click, visible, honoured immediately. A hard-to-find unsubscribe does not retain anyone. It converts people who would have left quietly into people who report you as spam, which is the single most damaging thing a recipient can do.
05 Remove people who never open.
A list full of dead addresses drags your sender reputation down and flatters your list size. Removing them improves both delivery and honesty.
The ones that matter once it arrives
One idea per issue. A digest of five links teaches people to skim, then to stop opening. One argument, properly made, with the link for those who want more.
A subject line that describes rather than teases. Curiosity gaps work once. The second time, the reader has learned the email will not be what the subject implied.
One call to action. Two competing asks reliably produce fewer clicks than one clear one.
Plain text that survives images being blocked. Many clients block images by default, so an email whose message lives inside a graphic arrives blank.
A real reply address. Replies are the most valuable thing a newsletter produces, and no-reply addresses discard them by design.
What is not worth the time
Optimising send times before you have enough sends to know anything. The published best hours are averages across millions of unrelated campaigns and they will not describe your two hundred readers.
A B testing subject lines on a small list. At a few hundred recipients the difference between two subject lines is indistinguishable from chance, and acting on it teaches you something untrue.
Chasing the open rate. Since Apple began pre-loading images, a share of recorded opens are machines. Clicks and replies are the honest signals, which is also why journey stages should be defined by clicks rather than by opens.
Before the first send
Send the issue to yourself and to one colleague on a different email provider. Read it on a phone. Click every link. Check the unsubscribe works and that your physical address is present, because omitting it is a legal problem in several jurisdictions as well as a trust one.
Then check the thing nobody checks: does the email say something specific? The newsletter we write for founders lives or dies on that question rather than on any of the mechanics above. The mechanics only determine whether it arrives. Whether it gets read again next week is decided by whether this one was worth opening.