Definition
Deliverability describes a sender's ability to get its messages into the inbox. Three outcomes are possible for any message: it is rejected by the receiving server, it is accepted and filed as spam, or it is accepted and placed in the inbox or a primary tab. Only the last is good deliverability.
The commonest confusion is with the delivery rate an email platform reports. That figure counts messages the receiving server accepted rather than bounced. A message filed straight into spam counts as delivered. A sender can see a delivery rate that looks perfect while a growing share of its mail never reaches an inbox. The email deliverability guide explains the mechanics and why placement figures are hard to observe directly.
In practice
Four foundations support good deliverability.
- Authentication. SPF, DKIM and DMARC records let receiving servers verify that mail claiming to come from a domain was authorised by it. Google's email sender guidelines set out authentication and other requirements for anyone sending to personal Gmail accounts, with stricter rules for bulk senders.
- Consent. People who asked for the mail open it, and people who did not mark it as spam. The list building and consent guide covers how opt-in practice protects reputation from the start.
- Complaints and unsubscribes. A spam complaint is the strongest negative signal a recipient can send. An easy unsubscribe is the safety valve that turns would-be complainers into unsubscribers, which is why the Unsubscribe RateShare of delivered emails that led the recipient to opt out. is read beside complaint data rather than minimised at any cost.
- List hygiene. Sending to addresses that bounce, or to people who have not engaged for a long time, signals a careless sender. The email list hygiene SOP sets out the routine for suppressing bounces and re-confirming or removing inactive contacts.
How mailbox providers judge senders
Providers do not publish their filtering rules, but they describe the signals they weigh: authentication, the sending domain's and IP address's history, complaint rates, bounce patterns, and engagement such as opening, replying, moving mail out of spam or deleting it unread. They judge senders as much as messages. A well-written email from a sender with a poor reputation still goes to spam, and a sender's reputation is shared across all its mail, so a careless promotional send can drag down the delivery of receipts and password resets from the same domain.
SegmentationDividing a market or customer list into groups that need different offers or messages. helps here: sending to engaged recipients more often and to unengaged ones less often concentrates positive signals. A pre-send routine such as the email campaign QA checklist catches the authentication, link and unsubscribe problems that would otherwise surface as complaints.
Warning signs and recovery
Deliverability usually fails gradually. Early signs include a falling Open RateShare of delivered emails recorded as opened. at one mailbox provider but not others, a rising complaint rate in the provider's postmaster tools, and replies saying a message was found in spam. Open rates are distorted by privacy features, so they are a weak signal on their own; a sudden drop at a single provider is more telling than the overall trend.
Recovery means removing the cause before rebuilding reputation: stop mailing the contacts that triggered complaints, fix authentication, reduce volume to the most engaged recipients, and increase it gradually as signals improve. The newsletter growth loop case study shows deliverability groundwork having to come before growth, and the email lead generation playbook treats sender reputation as a precondition of any list-building effort.
An audiobook shop pays for a bought list
Hushfeather Audiobooks buys 40,000 addresses to add to its 18,000 customers and sends a promotion to all 58,000. The platform reports 96% delivered. Within two sends, spam complaints climb from under 0.05% to 0.4%, and customer receipts start landing in spam at the largest webmail provider. Orders from email fall from about 300 a week to 110. The shop deletes the bought addresses, suppresses customers who have not opened or clicked in 12 months, and sends a re-permission email to 4,200 lapsed customers, keeping the 1,300 who confirm. Sending only to engaged customers for six weeks, it sees complaints fall back below 0.05% and email orders recover to 280 a week.
Related terms
Open RateShare of delivered emails recorded as opened. is the most visible early warning, despite its distortions. Unsubscribe RateShare of delivered emails that led the recipient to opt out. measures the safety valve that keeps unhappy recipients from complaining. SegmentationDividing a market or customer list into groups that need different offers or messages. lets senders concentrate mail on engaged recipients. Bounce RateShare of sessions that were not engaged (GA4); historically single-page sessions. in analytics is a different measure: it describes website sessions that were not engaged, not emails that were refused.
Common confusions
- Delivered is not inboxed. The delivery rate counts acceptance by the server, including mail filed as spam.
- Email bounces are not website bounce rate. A hard or soft bounce is a refused email; bounce rate in analytics describes sessions.
- Content is rarely the main cause. Reputation, consent and complaints usually matter more than particular words.
- A new sending tool does not reset reputation. Domain reputation follows the sender wherever it moves.
- Bigger lists are not better lists. Unengaged and bought contacts drag down delivery to everyone else.
