A practical checklist before your first production send
Check recipients, content, failure handling, and support visibility before your app starts sending real customer email.

On this page
Make test delivery deliberate
Separate development, staging, and production credentials and recipient policies. In non-production environments, consider routing mail to a test inbox or an explicit allowlist. A staging database copied from production should not become an accidental mailing list.
Use recognizable environment labels in test messages and make the production send path an intentional configuration choice. Keep secrets on the server, outside browser bundles and committed files.
- Verify the intended From and Reply-To addresses.
- Confirm the sending domain’s configuration.
- Check that staging cannot send to arbitrary real recipients.
Exercise the uncomfortable inputs
Build fixtures for empty optional fields, very long names, non-ASCII text, and invalid or expired action links. Preview both HTML and plain text. Then send through the real integration to mailboxes you control and inspect the result.
Test a replayed application event, a provider timeout, and a delayed webhook. Decide which failures should retry and which should be surfaced for intervention. Record enough information to investigate without logging tokens or complete message bodies unnecessarily.
Give the team a way to see and stop problems
Before launch, agree on who responds to a sending incident and where they can inspect message history. Have a way to pause a faulty notification job without disabling the entire product. Document how to resume it safely.
Start with a controlled set of real workflows and review the results. Did the intended person receive the right message? Did the action link finish the task? Could support explain what happened from the stored events? Those questions catch more useful problems than checking only whether an API call returned a success code.
