Accepted vs delivered vs inboxed
Accepted — your MTA handed the message to the recipient MTA, which returned a 2xx SMTP code. Says nothing about whether the message reached the inbox.
Delivered — the recipient mail server actually placed the message somewhere visible to the user (inbox, spam, promotions). For Gmail / Outlook receivers we infer this from a combination of acceptance plus engagement signals.
Inboxed — the message went to the recipient's primary inbox, not spam or a folder they rarely check. This is the only number that affects business outcomes. SNDR's dashboard reports per-recipient-domain inbox rate.
The five levers
Authentication — SPF, DKIM, DMARC aligned and passing. The baseline; everything else compounds on top.
Reputation — your sending domain and IPs have to be trusted by the receiver. Slow to build, fast to lose.
Content — clean HTML, plaintext alternative, no spammy phrases, link domains that match the From: domain.
List quality — addresses are real, recently active, and want your mail. Old / purchased lists degrade reputation fast.
Engagement — recipients open, click, and reply. Mail that gets ignored degrades; mail that gets "Not spam" clicks recovers.
Per-ISP reality
Receivers do not coordinate. Gmail, Outlook, Yahoo, Apple, and Microsoft 365 each compute reputation independently and apply different policies. A sender can be loved by Gmail and dropped by Outlook at the same time.
This is why aggregate "inbox rate" is misleading. SNDR ships per-recipient-domain inbox rate so you see a Yahoo slide before it shows up as a global average decline.
Common deliverability problems and what fixes them
Mail landing in spam at Gmail only — usually a Gmail-specific reputation issue. Check Postmaster Tools; tighten DMARC; ensure List-Unsubscribe-Post header is present.
Bounce-rate spike — usually a list-quality problem. Check for recent CSV imports or test-environment leakage. Add to suppression and stop re-sending.
Complaint-rate spike — usually a content or relevance problem. Check what changed in the most recent campaign.
Sudden delivery latency — usually a single-MTA reputation issue. The SNDR per-MTA dashboard shows which pool member is the culprit.
How SNDR optimises for deliverability
Multi-MTA pool with weighted routing — sticky-by-recipient-ISP. Gmail consistently sees you from the same IP, Outlook from another, Yahoo from a third. Reputations build up in parallel.
Circuit-breaker drop-out for misbehaving pool members. A flapping box stops accepting traffic automatically until it recovers.
Automatic warmup for new sending domains — fresh IPs are weighted lower until they accumulate reputation.
Per-recipient-domain inbox rate in the dashboard so you see ISP-specific issues before they affect aggregate metrics.
Suppression handled automatically; hard bounces and complaints stop being retried in real time.