Follow the sending evidence
Silent video.

[0–6]
Define delivered • Server acceptance does not establish inbox placement. • Attempt • Accepted • Inbox • Junk • M17 · receiving server accepted • Placement still needs its own evidence • One test message ≠ universal placement rate
The reviewed diagram shows an attempted message accepted by the receiver before a branch to inbox or junk. Illustrative M17 highlights server acceptance while placement remains a separate evidence question. SendGrid delivery means receiving-server acceptance and does not guarantee inbox placement. One test inbox describes one recipient and message, not a universal placement rate; engagement tracking is a different observation. Distinguish unattempted, deferred, rejected, accepted and broken-message cases before diagnosing.

[6–12]
Inventory every sender • Staff, campaigns, stores and apps may use different paths. • One visible From domain · inspect every actual path • Staff mail • Mailbox provider + owner • Campaigns • Campaign account + owner • Store notices • Store / app + owner • App messages • Relay / API + owner
A shared visible From domain remains above four separately traced origins: staff correspondence, campaigns, store notices and application messages. Each links to its actual provider, account or relay and its change owner. This is an inventory of existing paths, not a product recommendation. A provider migration or new store app warrants reviewing the affected row. Identify provider-owned infrastructure versus DNS records the team can change, and use current recipient-provider requirements for the actual sender category.

[12–18]
Inspect actual messages • Compare authentication with the configured sender. • Campaign sender • Application sender • Received message • Compare results • M17 · campaign evidence • A17 · application evidence • Check auth, domains and setup
The full reviewed hypothetical-store figure retains separate campaign and application paths and received-message results. Native M17 campaign evidence and A17 application evidence highlight independently before comparing actual authentication, domains and intended settings. A campaign setup screen or successful newsletter test does not verify an older application relay. Use representative received messages and provider-supplied values with a reviewed change plan; inventory legitimate senders before a stricter domain policy.

[18–24]
Test the opt-out • A later sync must not silently restore eligibility. • O17 · test marketing opt-out • C17 · initially eligible fixture • C17 · marketing withheld • Later customer sync • Preference persists; customer record may still exist • Controlled fixture · test supported header and body routes where required
In a controlled C17 fixture, marketing eligibility is replaced by withholding after test opt-out O17. A later customer-sync token moves toward the record but the withheld state remains visible and never reopens. This is expected behaviour to test in the actual audience systems, not a claim that a production test passed. Customer-record existence is distinct from marketing permission. Check the visible opt-out and the applicable supported header mechanism: Google bulk-sender rules for marketing and subscribed messages require one-click unsubscribe plus a visible body link. Verify the resulting suppression.

[24–30]
Record and retest • Keep the response, owner and remaining uncertainty. • A17 · path + time + recipient provider • Exact response + auth + audience state • One diagnostic record · named owner · reviewed change • Retest the affected path; keep remaining uncertainty
Illustrative A17 path, time and recipient-provider evidence and the exact response, authentication and audience state feed one diagnostic record. Assign a named owner and reviewed targeted change, then retest the affected sending path and audience conditions. Keep actual results, observation period and uncertainty; do not change unrelated settings blindly. Share minimal private references and redacted examples through the agreed channel. Compare metrics only with their provider definitions and denominators. Published hosting email-DNS checks remain included; broader campaign diagnostics need an agreed scope.