Skip to content

Authenticated email is not proof of delivery

A positive domain result does not explain where a message arrives. Separate technical checks, acceptance and visible receipt in your product.

Nora Henning2 min read
Envelope with a blue verification seal in front of a grey letter slot.

An application can complete a business task successfully and then fail when sending the associated email. If your interface combines both into one green tick, the team lacks the information needed to know where to intervene.

A message has several observable states

In the fictional product design, a support portal creates notification N-68. The sending service accepts the job. Later, feedback from the receiving mail system arrives. The interface should derive precisely the statement supported by that evidence.

Observation in the design

Suitable product message

Not yet established

Handover to the sending service confirmed

Handed over to sending service

Acceptance by the destination system

Acceptance by the destination system confirmed

Accepted by receiving mail system

Visible placement in the inbox

Test participant confirms arrival in the test mailbox

Visible in the checked test mailbox

Receipt by all other recipients or reading by customers

This table proposes a mapping for the design. It contains no sending tests performed here. The available feedback must be established for the service used and its events.

Domain checks answer a different question

A successful DMARC result confirms authorised use of the sender domain, not content quality or harmlessness. It is not a general acceptance guarantee either. The receiver may apply further checks and decide differently.

The product message should therefore not automatically turn a positive authentication result into “delivered”. Conversely, absence from the inbox alone does not explain which authentication mechanism might be faulty. Examine the sending route and the specific feedback together.

Treat feedback as part of the application

AWS describes separate events for bounces and complaints and notes that multiple notifications are possible. For this design, the application therefore needs to link events to the message and define how repeated feedback is processed. A second event should not trigger a second message without a reason.

If an event is missing, that part of the state remains unknown. Monitor such cases and assign handling to a responsible role. This lets the team manage the notification while the business task’s state remains traceable. The statement in the interface then follows the available observation.

Sources

Written by

  • Nora HenningEditor for IT SecurityNora Henning looks after the editorial content on IT security at ASES. She writes about application security and explains how access rights and the protection of company data are connected.

Follow ASES

Read more

All articles

Your project with ASES

Describe your project to us. We will get back to you with an initial assessment.