Playbook · 8 min read

The deliverability checklist your outbound workflow needs

Sending an email is easy. Knowing whether it arrived where it should, reached the right person, and preserved the sending domain is an operating discipline.

Key idea

Treat deliverability like site reliability: define the service, monitor its health, set stop conditions, and investigate failures before adding volume.

Why “sent” is not a success metric

A historical HN launch about inbox-placement testing points to a useful distinction: delivery, inbox placement, and response are different layers. An email can be accepted by a receiving server and still be filtered away from the person you hoped to reach.

Do not treat a commercial tool’s claims as a benchmark. Use the underlying lesson — test the channel before trusting the campaign report.

Define the health model

LayerWhat to monitorAction when it degrades
IdentityDomain alignment and sender authentication.Pause new sending and fix configuration.
List qualityVerification status, bounce patterns, suppression coverage.Quarantine the source or cohort.
PlacementSeed or panel tests across relevant mailbox providers.Reduce volume and investigate content or infrastructure.
Recipient responsePositive, negative, opt-out, and complaint signals.Review targeting, relevance, and offer.

Run the pre-send checklist

  • Audience: Is every record eligible, current, verified, and unsuppressed?
  • Evidence: Can the message’s claims be traced to a current source?
  • Infrastructure: Are domains, mailboxes, authentication, and send limits documented?
  • Content: Is the message short, specific, and honest about why it was sent?
  • Fallback: Does the team know who can pause a sequence and what triggers that decision?

Use stop conditions

A healthy outbound programme does not only define targets. It defines the signals that make the programme unsafe to continue: unexpected bounce changes, rising complaints, a broken suppression sync, poor placement, or a surge of “why are you emailing me?” replies.

The operator should be able to pause a workflow without waiting for a quarterly review. That is how you protect a channel that takes months to rebuild.