Abuse & compliance · Email
Email health classifies bounce, complaint, and delivery signals without automatically throttling or pausing sending. This page explains what each signal means, how automatic recipient suppressions work, and what you can do when deliverability is at risk. See the Abuse & compliance overview for the channel-agnostic abuse model.
The signals that drive reputation
We derive your email reputation from the same per-recipient events you already receive through webhooks and the events API, so you can watch exactly what we watch:
- Bounce rate: the share of your delivery attempts that bounce, across every bounce class. Hard bounces (the address does not exist) do the most damage. A high rate is the signature of a purchased, scraped, or badly stale list, and it is the signal mailbox providers punish hardest.
- Complaint rate: the share of delivered mail reported as spam, arriving as email.complained through mailbox-provider feedback loops. Complaints measure whether recipients wanted your mail at all.
- Unsubscribe behavior: email.unsubscribed for the in-body link and email.list_unsubscribed for the provider's one-click button. Opt-outs are a normal part of list life, but a spike on one campaign is an early warning that your targeting or your consent has drifted. Recipients who cannot find the unsubscribe link report spam instead.
Track bounce and complaint rates per campaign so you can respond before deliverability declines.
Health classifications
The Metrics dashboard classifies the current window as healthy, watching, or throttled. The overall verdict uses the worst classification among bounce rate, complaint rate, and delivery rate. Open rate has its own signal but does not affect the overall verdict.
These classifications describe deliverability risk. watching means at least one signal has entered its warning band. throttled means at least one signal has entered the highest-risk band. Neither classification changes your sending rate or pauses traffic.
Use the Metrics breakdowns to find the sending domain, recipient domain, or bounce reason driving the classification. Pause and clean an affected campaign or list before sending to it again. If the decline is broad rather than concentrated, contact your deliverability representative or Bird support for review.
Automatic recipient suppressions
The health classification never acts on your sending rate. We do act on individual recipients as their signals arrive, so a bounce or complaint needs no work from you. Hard bounces suppress the address for all mail. Complaints and unsubscribes suppress it for non-transactional mail, so password resets and receipts still arrive. Every automatic addition fires an email_suppression.created webhook event. The Suppressions guide covers the policy, record schema, and management API.
Suppression stops repeat sends to an address after a permanent failure or opt-out. A later send to that address is rejected before it reaches a mail server. Soft bounces and deferrals do not create suppressions. We retry deferrals because they report a temporary failure.
Your levers: list hygiene and consent
Health signals reflect how you build and maintain your list. Focus on two practices:
- List hygiene: never import purchased or scraped lists, drop addresses that have not engaged in months, and validate addresses at collection time, because a typo'd signup is a future hard bounce. The list hygiene guide has the workflow.
- Consent: send only to recipients who asked, provide a clear one-step unsubscribe path, and keep marketing in the marketing category so suppression policy and unsubscribe handling apply to it. Most complaint spikes trace back to mail the recipient never opted into.
If the overall classification is throttled, inspect the breakdowns and pause the affected campaign or list while you fix its source data. Resuming with the same list reproduces the same signals. The health classification itself does not pause sending.
Next steps
- Abuse & compliance overview: the channel-agnostic trust and safety model
- Suppressions: the four suppression reasons, the applies_to policy, and the management API
- Email events: full payloads for email.bounced, email.complained, the unsubscribe events, and bounce classification
- Email metrics: health classifications, per-sender breakdowns, and the signals to investigate
- List hygiene: keeping bounce and complaint rates low in practice