Email deliverability: getting into the inbox, and knowing that you did

Email deliverability is whether your message reaches the inbox rather than the spam folder, a promotions tab, or nowhere at all. It is distinct from delivery rate, which only tells you the receiving server accepted the message — a mail accepted and then filed into spam counts as delivered, and that is the gap most reporting hides.

The distinction matters because the two numbers can move in opposite directions. A list with many long-dormant addresses can show a delivery rate above 99% while placement quietly collapses, because mailbox providers accept the mail and then route it away from the inbox based on how recipients behave. By the time an open-rate decline is visible, the reputation damage is typically weeks old.

Deliverability is also not a single setting. It is the aggregate of authentication, sending-domain reputation, IP reputation, list hygiene, content signals, and recipient engagement — evaluated independently by each mailbox provider, using rules none of them publish in full.

Placement is decided by a filter weighing several signals at once, and reputation carries more weight than content. A perfectly written email from a domain with a poor sending history will be filtered; a plain one from a trusted domain generally will not.

The signals below are the ones you can influence. None is sufficient alone, and the first three are effectively prerequisites — without correct authentication, the rest barely matters, because major providers now reject or quarantine unauthenticated bulk mail outright.

The four standards answer different questions, and they are cumulative rather than alternative. SPF asks whether this server is permitted to send for this domain; DKIM asks whether the message was altered in transit; DMARC ties the two to the visible From address and tells receivers what to do on failure; BIMI displays your logo once DMARC is enforcing.

The subtlety that catches most senders is alignment. A message can pass SPF and DKIM and still fail DMARC, because DMARC requires the authenticated domain to match the domain in the From header the recipient actually sees. This is precisely what breaks when mail is sent through a third-party platform on its own domain rather than one you have delegated.

A seedlist is a set of real mailboxes you control across the major providers. You include them in a send, then check where each copy landed — inbox, promotions, spam, or missing entirely. It is the only method that observes placement directly rather than inferring it.

Its value is that it is empirical. Engagement metrics tell you about the people who saw the mail; a seedlist tells you about the mail nobody saw. That is the population you most need information about, and the one that is invisible by definition in ordinary reporting.

Its limits are equally real and worth stating plainly. Seed mailboxes have no genuine engagement history, so a provider treats them differently from an engaged subscriber. Results indicate how a message is filtered on its content, authentication, and reputation — not a guaranteed outcome for every recipient. Read it as a smoke alarm, not a census.

Apple's Mail Privacy Protection, introduced in 2021, pre-fetches tracking pixels for Apple Mail users regardless of whether the message was read. Because Apple Mail is one of the most common email clients, a large share of recorded opens are now machine-generated.

The practical effect is that open rate is no longer usable as an absolute measure, and is only weakly usable as a trend. Engagement-based segmentation built on opens quietly degrades: subscribers who never read anything still register as opening, so they survive re-engagement rules designed to remove them — which then damages the sending reputation those rules existed to protect.

Clicks, replies, conversions, and unsubscribes remain trustworthy because they require deliberate action. The correct response is not to discard opens entirely but to stop treating them as ground truth, and to rebuild any automation that silently depends on them.

Reputation recovers by demonstrating sustained good behaviour to each provider, and there is no way to accelerate it. Recovery is measured in weeks, and the single most common mistake is resuming full volume as soon as placement improves, which resets the clock.

The sequence that works is unglamorous: stop sending to unengaged segments entirely, fix authentication so every message aligns under DMARC, remove hard bounces and role addresses, then rebuild volume gradually starting with your most engaged recipients. Providers are watching the ratio of wanted to unwanted mail — shrinking the denominator works as well as growing the numerator, and works faster.

Changing sending domain to escape a bad reputation is a trap worth naming. A brand-new domain has no reputation at all, which providers treat with suspicion rather than neutrality, and you also forfeit whatever positive history remained.