Concepts

Metrics & optimization

What's measured, the status ladder, and the ranked action feed that tells you what to fix.

Every send is logged to the message spine, and provider webhooks advance each message up a status ladder as events arrive. Metrics are aggregations over that spine.

The status ladder

queued → sent → delivered → opened → clicked, with bounced, complained, failed, and suppressed as terminal branches. A hard bounce or a complaint auto-suppresses the address (complaints are never removable).

Rates that matter

  • Delivered / sent — reaching the inbox at all.
  • Opened / delivered, clicked / delivered — engagement.
  • Bounced / sent — list hygiene. Above 2% is a problem.
  • Complained / sent — the reputation killer. Above 0.1%, stop and investigate before sending more.

Where to read them

  • The dashboard Metrics page: headline tiles, an engagement trend, and a per-template breakdown.
  • Per automation: a step-keyed funnel and, for A/B steps, per-variant results.
  • Over MCP: email_get_email_metrics (rates by template/day), email_get_automation_metrics (per-step + per-variant funnel with an A/B winner readout).

The optimization plan

email_get_optimization_plan turns those numbers into a ranked action feed — deliverability risks, setup gaps, A/B winners ready to promote, zero-open templates, stale drafts — each with evidence. It's the "what should I fix next" answer, and the one machine-applicable action (promoting a called A/B winner) carries the exact call to apply it. See A/B testing.

On this page