Start with the message that failed
Collect the sending service, sender address, recipient domain, approximate time, and available error or message headers. Avoid changing several systems before you have a reproducible example.
Distinguish a failed application request, a provider accepting the message, a recipient server accepting it, and the recipient seeing it in the inbox. Each step answers a different question.
Inventory the approved senders
List employee mail, website forms, receipts, invoicing, newsletters, and customer-management tools. Record each service’s owner and sending domain.
Remove obsolete connections only after confirming they are no longer needed. Keep the inventory current when the business introduces another tool that sends email.
Verify authentication and alignment
Google requires SPF or DKIM for all senders to personal Gmail accounts and additional authentication for bulk senders. Its sender requirements explain the applicable conditions.
DMARC can pass when SPF or DKIM passes with an authenticated domain aligned to the visible From domain. Check the results on actual messages, following Google’s DMARC guidance. Authentication alone does not guarantee inbox placement.
Review recipients and unsubscribe handling
Send marketing mail to people who requested it and honor subscription choices. Google describes additional one-click unsubscribe requirements for certain bulk marketing and subscribed messages in its sender guidelines.
Check bounced, complained-about, and unsubscribed addresses before importing or reactivating contacts. Review both the visible unsubscribe route and the provider’s suppression handling.
Review content and sending behavior
Use a recognizable sender, accurate subject, working links, and a clear purpose. Test representative transactional and marketing messages separately.
Google advises against sudden sending-volume increases and recommends monitoring errors and reputation while increasing volume gradually in its sending guidance. There is no universal ramp that guarantees delivery for every sender.
Make a controlled correction
Document the current state, the proposed change, and the evidence you expect afterward. Test the affected service again and compare results for the same message type.
For a provider escalation, include example timestamps, recipient domains, headers, and error responses through an appropriate private support channel. Avoid posting customer information in public troubleshooting threads.
Establish ongoing ownership
Assign responsibility for authentication records, subscription handling, bounce monitoring, and new sending tools. Set a review process that fits the business’s sending activity.
Treat delivery improvement as an evidence-led operating task. Keep the change history and unresolved issues visible so the next person can continue the work without repeating guesses.